You might want to wait for Sandy Bridge in Q2+ 2011 instead of buying Westmere

[c.hilding]

Before you start reading; this thread has nothing to do with performance of Sandy Bridge vs Nehalem/Westmere. Just wanted to get that out of the way since some people didn't read the post and just assumed things and jumped immediately at the performance argument. That's not what this is about at all. Moreover, all relevant information that surfaces in the thread is moved to this first post so that it's always up-to-date.

It should be clear to everyone by now that we have entered the era of the iDevices (iPhones and iPads). Steve Jobs' zeal on stage while talking about the iPad and now the iPhone 4 shows that he's on a tunnel-vision crusade. Apple is concentrating all their efforts on becoming the market leader in this new field. Heck, they didn't even give ONE Apple Design Award to a Mac application this year, despite this being a traditional award given out every year to the best MAC applications. Suddenly, it's all about the iDevice toys... Portable consumer devices for browsing the internet and checking email are expected to overtake computers completely by 2015, to a point where content creators and gamers will be the only people still using desktop machines. They also predict that most applications and services will be in that disgusting "cloud" thing which requires Internet connectivity just to access apps. I don't really believe either of those predictions, but that doesn't matter, Apple (and when I say Apple I mean Steve Jobs) is dead-set on shaping, grooming and dominating this new market.

So... it's been over 460 days since the last update to the Mac Pro, which is highly unusual, and of course this massive delay (along with Intel starting mass production of Westmere Xeons now, June) lead everybody to the highly logical conclusion that we should expect new Mac Pros at WWDC '10, which as we know didn't happen... Did you hear the developer-audience's cheers at "one more thing", only to be let down by a damn video chat application? The audience consists mainly of professional developers who depend on Mac Pros in their offices, and this iToy business was a huge slap in their faces by Apple. It's clear that WWDC is no longer a developer convention, as far as the keynotes go. It's now a marketing arm of Apple, designed to create cute Steve Jobs-quotes and video clips to feed to the consumer public via the massive amounts of press that hang out at the event. Heck, if anyone doubts Steve's tunnel vision; Steve Jobs can't even promise that we'll get Apple Design Awards for Mac DESKTOP APPLICATIONS next year either (http://forums.macrumors.com/showpost.php?p=10088596&postcount=4) (note the emphasis on "Maybe"). There should be NO question about what Steve Jobs' latest baby is, and it's NOT the Mac desktops.

Waiting for new Pro machines is of course made even more painful by Mac Pro 2009's insane prices. Apple could at least drop the current price! You can build your own Hackintosh with the exact same components as the current Mac Pro 2009s for less than half the price using easily accessible off-the-shelf parts. To top it off, the current machines were released over 460 days ago, and have kept the same price for all this time despite being overpriced even back then (the 2009 Mac Pros actually cost Apple a lot less to manufacture than the 2008 models due to much lower CPU prices, yet Apple significantly raised the price of the 2009 models), it's insanity... It has gone far enough that lots of people that used to despise Hackintoshes have actually considered building one of their own, since Apple has completely dropped the ball. Heck, they've thrown the ball away! Over 460 days without updates so far, and no price reductions... jeeze.

This new iToys-horny Apple is giving me a headache.

So WWDC came and went without any news of new Mac Pros. However, today, June 10th, Steve Jobs confirmed that Apple has NOT forgotten about the Mac Pros (http://www.macnn.com/articles/10/06/10/promises.better.future.ahead/) and that we'll have to "wait and see". Updates are coming. What you'll have to ask yourself is if you should buy now or wait for Sandy Bridge.

See, on April 13th, 2010, Intel's CEO announced that they will begin mass-producing their latest architecture, Sandy Bridge, in Q4 2010:
http://www.pcmag.com/article2/0,2817,2362601,00.asp

Server-versions (Xeons) of Sandy Bridge will follow a few months later, around April+ (Q2) 2011.

Basically, to get the most out of your buck, you have these choices:

If you have a perfectly good workstation now and can use it until April+ 2011, then wait. You'll be getting Sandy Bridge's USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores per processor, and higher performance, and it'll be a much better buy than ANY Westmere of today (which is merely a die-shrink of November 2008's Nehalem).

If you don't have a workstation now but need one, then either get a Westmere Mac Pro workstation when they come out if you don't mind the entire architecture becoming obsolete (http://forums.macrumors.com/showpost.php?p=10100265&postcount=30) in 9-10 months (as opposed to just a die-shrink (http://en.wikipedia.org/wiki/Intel_Tick-Tock#Roadmap) such as Nehalem -> Westmere), or get something cheaper for now such as last year's technology; either via a refurb, second-hand or a cheap Hackintosh. I'll most likely hit this camp. They are still powerful machines and will be excellent to tide you over into 2011, where you'll be able to get the vastly superior Sandy Bridge (SB) architecture (http://www.homestarrunner.com/sbemail.html) <-- (click for unrelated relaxation with "SB" :p).

If you don't have a need for a workstation now, but want one later, then wait for Sandy Bridge and use something like an iMac, MacBook or Mac Mini for now.

As some apologists will point out, of course Nehalem/Westmere will still be incredibly powerful as far as CPU performance goes after Sandy Bridge is released, but it will no longer have the latest technology, and we're talking about groundbreaking things like USB3, SATAIII, LightPeak and PCI Express 3.0 (which both doubles the bandwidth over 2.0 and has many more lanes per CPU, giving you more PCI Express ports in the computer), where we'll eventually have devices and cards that require these ports. Wouldn't you rather have a computer that contains the ports that are about to become the new standards? Anyone who doesn't at least look over the implications of buying now versus waiting is foolish. This is a completely new architecture right around the corner, with multiple new standards technologies, things that will just be in more and more demand as modern devices are released that will require USB3, SATAIII, LightPeak or PCI Express 3.0. It's your money and I'm not saying you shouldn't buy a Westmere Mac now. As I said, it's perfectly fine to buy a Westmere now as long as you know what you're getting into.


Responses to a couple of possible objections:

Q: Who cares?
A: I'm glad you asked that. Depending on who you are, this might not be interesting to you at all, but for the rest of us, this factor of a whole heap of emerging technology should at least be taken into account before spending 3 years of saved up cash on a fully loaded Westmere Mac, when you could buy a second-hand/refurb/make a Hackintosh to tide you over until the real quantum leap with Sandy Bridge. Unless you are made of money, that will be a much wiser choice. I happen to be made of money and I still won't be buying any fully kitted out Westmere, it's just not a wise choice with such a quantum leap coming up quite soon. The best thing to do is just to get something to get by with while waiting; and the current machines are very powerful, so a refurb/second-hand/Hackintosh will be the best choice for most people.

Q: But... but... Westmere has 6 cores instead of 4, isn't that better?
A: Those that need the 4 extra cores (from 2x4 (8) to 2x6 (12)) are free to buy a Westmere now, if they really need it, but even most editors would get by excellently with the 2009 Mac Pro. As for Westmere (6 cores) vs Nehalem (4 cores), the performance is negligible since most applications cannot even use more than 1 or 2 cores simultaneously. Come back when programs are written to use "n" number of cores without bottlenecks and I'll say your NEHALEM will come to use (let alone the WESTMERE). Heck, stuff like 3D Studio Max crashes if you have over 8 cores (or something like that, I forgot the exact number), and most applications still only use one or two cores. There are programs that are exceptions to this, along with people who use said programs to make money, and for them I say go for it (Westmere) if you need the extra cores.

Q: Aren't there current-generation PC motherboards with USB3 and SATA III that could be used in a HackPro?
A: That's a two-part question; "is it available?" and "is it supported under Mac OS X?". Current implementations exist, yes; however, they are unofficial non-Intel controller chipsets. The Intel solution is the one most likely to be in the Mac Pro, and Intel is releasing an all-in-one USB 3 / SATA III controller chip in conjunction with Sandy Bridge's release in 2011, for OEMs such as Apple to use on their motherboards. Therefore, there will NOT be Mac OS X drivers for ANY of those unofficial chips that are out today; just in case you were planning to build a Hackintosh with a PC motherboard. One such board is the EVGA SR-2, and if you were to use it, your only chance at getting support for your precious USB3 and SATA3 ports is if you try to port open-source Linux or UNIX drivers to Mac OS X, which is something that the Hackintosh community does at times, but don't bet on being able to boot from a SATA3-connected disk. Basically, NO these things don't exist under Mac OS X until Intel officially brings them out, which they'll be doing with Sandy Bridge's release in 2011. Until then we don't know which of the USB 3 controller chips Apple will be using (most likely the Intel one), and therefore you shouldn't build any Hackintosh today with 3rd party chips hoping for drivers, because they'll most likely never come. So chill out and wait for Apple's Sandy Bridge board, it's too early to know which controller they'll be using (in other words; which chipset they will be writing Mac OS X drivers for).

Q: Okay so USB3 and SATA III won't work under Mac OS X with current non-Intel implementations. What about Light Peak and PCI Express 3.0, can those be added to current computers in any way?
A: Light Peak - YES, at lower performance; Intel has designed a PCI Express 2.0 card that will supply Light Peak to existing motherboards (not as fast as native Light Peak though). PCI Express 3.0 - NO, that is dependent on the CPU. For instance, each Nehalem/Westmere Xeon provides thirtytwo (32) "1x" lanes of PCI Express 2.0; with two CPUs that's sixtyfour (64) "1x" lanes, or a total of "64x" of bandwidth. This is then split across the various connectors on the board in any way the manufacturer prefers. Apple, for instance, chose to go for a 4x16 (=64) setup, but could just as well have gone for 2x16 (=32) + 4x8 (=32) = 64 (which would have given six ports, and still only used 64 lanes). Other manufacturers, like EVGA, use Nvidia NF200 line doublers, which is basically an intelligent queuing/messaging passthrough device that allows you to connect multiple fast devices to a single lane, and therefore artificially increases the number of lanes and lets you create configurations such as EVGA SR-2's seven (7) 16x lanes. For PCI Express 3.0, you will need new Sandy Bridge CPUs to take care of the increasing demands such as doubled polling rate and more lanes from the CPU.

Q: What is all this good for? USB 3? SATA III? Light Peak? PCI Express 3.0? How do I know if I will need it for what I'm doing?
A: USB 3: Offers a maximum transfer rate of 5 Gbps, which is 10 times faster than USB 2. USB 3 is also full duplex (bi-directional; allowing simultaneous upload and download) whereas USB 2 is only half-duplex (one direction at a time). USB 3 is also superior to SATA 1.0/eSATA, which has a maximum rate of 3 Gbps, and to FireWire 800 (which only offers full duplex at 800 Mbps). It remains to be seen how much CPU it will use (USB has historically been a CPU hog compared to FireWire).
SATA III: This will offer 6 Gbps transfer rates, which is mainly going to benefit SSD users, since fast SSD RAID arrays will be bottlenecked by the current SATA bus.
Light Peak: LightPeak is a high-speed optical transfer technology developed by Intel, which will initially offer 10 Gbps over a single wire (and eventually scale up to 100 Gbps), and is something you would have in addition to things like USB3 and SATA III, and only for special purposes; it is meant for connecting super high resolution screens, networking, storage, etc, basically anything that benefits from its huge bandwidth. It offers a massive amount of bandwidth and will be excellent for a lot of purposes such as network-accessed RAID arrays. It all travels over one, tiny cable as well. The long-term goal by Intel is to eventually connect most devices using LightPeak and get rid of dedicated cables such as DVI, HDMI, DisplayPort, Audio cables, Mouse cables, Keyboard cables, Network cables, etc, but for the foreseeable future the most likely use will be in high-speed transfers such as networking; with everyone still connecting their regular devices through their dedicated ports (USB, FireWire, DVI, etc...), the way it's done today.
PCI Express 3.0: As mentioned above, it will have double the current bandwidth of PCI Express 2.0, as well as more lanes. This leads to computers with many more PCI Express slots than today, allowing you to plug in lots of cards. For instance, if you are a music producer using a current Mac Pro and the UAD2 DSP card platform, you would be limited since the current Mac Pro only has four PCI Express slots, and you would need to use at least one for a graphics card, leaving 3 slots free for DSP cards. PCI Express 3.0 will solve all of that, offering many more and faster ports on a single motherboard.

Q: Well, even though Sandy Bridge is coming out soon, Apple might not use it right away?
A: My take is: Apple has received early exclusive access from Intel in the past (http://news.cnet.com/apple-beats-intel-to-nehalem-ep-chip-launch/), so it's not impossible that they'll be first out. Even if they do decide to screw us over and delay a Sandy Bridge update next year, it's always easy to build one yourself with off-the-shelf parts (Hackety-hack (http://forums.macrumors.com/showpost.php?p=10101281&postcount=44)), which in my opinion is perfectly fine to do when we're dealing with a company that doesn't treat their Pro customers' needs right.

Yes, it's often the case that the most powerful Macs are Hackintoshes, it's freaking horrible. :( You don't have that problem on the PC side, where you can just get the most powerful components and build anything cheaply. That's why most of the 3D rendering industry is back on Windows or Linux workstations now. I've seen one Windows machine with 288 GB Ram, dual X5680s, 24 SSDs in a RAID configuration. To say that it edited HD movies quickly is an understatement. You can't get anything even close to that in a Mac.

Let's hope that Westmere upgrade comes soon so that 2009's second-hand value drops further. ;) The 2009's will be nice and cheap and powerful enough while waiting for Sandy Bridge.

[ronweathers]

Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 3_1_3 like Mac OS X; en-us) AppleWebKit/528.18 (KHTML, like Gecko) Mobile/7E18)

Agree

[ronweathers]

Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 3_1_3 like Mac OS X; en-us) AppleWebKit/528.18 (KHTML, like Gecko) Mobile/7E18)

Agree

[Peace]

WWDC did not "came and went". It is still going on through tomorrow.

And did it ever occur to you that perhaps there might be some stuff about a new Mac Pro that only developers need to know about right now ?

For example XCode 4.0 has been given to developers attending WWDC. I'm guessing that version has to do with Macs since the iPhone 4 GM has been seeded meaning the "Toys" you refer to can be used using XCode 3.2.

[c.hilding]

Thanks Ron. Heck, at least we're voting with our feet. Not many people are willing to buy the outdated and overpriced Mac Pros right now, and loads of people are being pushed towards Hackintoshes. It'll be very interesting to see what Apple's next move is. They recently (mid-May) decided to hire 300 new employees at their facilities in Ireland during next year, and the article says that these people are "expected to work on the production of the firm's high-end desktop computer, the Mac Pro, which is targeted at professional users":
http://www.careersportal.ie/news/news.php?Heading=Apple+plans+to+recruit+300+new+employees&ID=19783158

I don't know where the article got that idea from (that they'll be working on the Mac Pros), but if that's true, the "hiring 300 people next year" statement doesn't look good.

[c.hilding]

WWDC did not "came and went". It is still going on through tomorrow.

If they would have released Mac Pros, it would have been during the keynote or on the Tuesday the day after, that's how they usually do it.

PS: The new release of Xcode is simply that, a new version of Xcode, with the Interface Builder integrated into the workflow, that's it, there's no hidden meaning. ;)

[Hellhammer]

I don't want to read that but just by the way, Mac Pro suitable Sandy Bridge CPUs won't be available before late 2011

[Umbongo]

Sandy Bridge Xeons are not coming in a few months. Q3 2011 is a rumoured release date. The 32nm socket 1366 line isn't even all out yet, there is more to come at lower price points.

[PeterQVenkman]

why would they refresh the Macs with Westmere Xeons now, only to replace them with the totally new Sandy Bridge architecture a few months later?

Because Apple loves charging tomorrow's prices for yesterday's workstations. It would make sense, in the screwed up Apple way, for the Pros to be updated right before a new architecture becomes available.

[c.hilding]

I don't want to read that but just by the way, Mac Pro suitable Sandy Bridge CPUs won't be available before late 2011

Sandy Bridge Xeons are not coming in a few months. Q3 2011 is a rumoured release date. The 32nm socket 1366 line isn't even all out yet, there is more to come at lower price points.

Did you at least read the linked article? It's from April 13, 2010. Intel's CEO announced that production of Sandy Bridge will begin in Q4 2010. This is the first time that Intel has confirmed a production date; all previous dates are old speculation and not based on facts in any way. Also, the announcement doesn't in any way separate the Xeon server processors from that date. Besides, both consumer (Core XXXX) and server (Xeon XXXX) processors are made from the exact same design these days, the only difference being that the Xeons have DUAL QPI links to allow them to communicate with a second processor in a dual CPU setup. Apart from QPI link count; both the consumer and server chips are identical nowadays. Therefore it's trivial for them to manufacture and release both at the same time, and the only constraint would be supply where they might have to choose whether to supply Server or Consumer users first. Either way, Q4 2010 is the first and only official information we have to go by right now.
http://www.pcmag.com/article2/0,2817,2362601,00.asp

Update: Confirmed that server versions will come a few months after the consumer versions. Yes, this has historically been what Intel has always done; but I thought that the recent unification of chipset designs, where nowadays both Server and Consumer versions use the same design, could have lead to a simultaneous release for both pro and consumer users. That turned out not to be the case, sigh... Still, that's Sandy Bridge Xeons in April+ 2011 so if you can wait it's very much worth it. Updated first post.

Because Apple loves charging tomorrow's prices for yesterday's workstations. It would make sense, in the screwed up Apple way, for the Pros to be updated right before a new architecture becomes available.

Haha, what a funny and true observation. Apple loves charging tomorrow's prices for yesterday's processors. Well put.

[nanofrog]

Sandy Bridge Xeons are not coming in a few months. Q3 2011 is a rumoured release date. The 32nm socket 1366 line isn't even all out yet, there is more to come at lower price points.
Exactly.

If Apple waited until Sandy Bridge, it'll be at least 2.5 years between systems, and that's basing it on shipped parts in the beginning of Q3 2011. :eek:

[VirtualRain]

This rhetoric is getting rediculous. I wish someone would sticky the Mac Pro historical release cycle...

Thanks to Topper...

G5 to 2006 Mac Pro - 304 days.
2006 to 2008 Mac Pro - 518 days
2008 to 2009 Mac Pro - 425 days
2009 to 2010 Mac Pro - 463 days and counting


Also, if anyone has tunnel vision, it's a large chunk of the users in this particular forum. Apple is not focusing ALL their efforts on iOS. While it is the focus of this particular WWDC (and for good reasons from a business perspective) Apple just refreshed the MacBook lines, there are rumors of a Mac Mini update next week, and that could very well be accompanied by a Mac Pro update. The last rumors I heard were that an update to the Mac Pro would occur around June, and the month is just underway.

Also, anyone that thinks the value is going to significantly improve, or pricing will significantly drop with a refresh is delusional. The best value in the current architecture today is a refurb 2009 and after the refresh, that won't change.

The only thing that will change after the refresh is all this whining about the absence of an update will be replaced with more whining from the same people about how the prices are ridiculous and the value isn't there.

[Hellhammer]

Did you at least read the linked article? It's from April 13, 2010. Intel's CEO announced that production of Sandy Bridge will begin in Q4 2010. This is the first time that Intel has confirmed a production date; all previous dates are old speculation and not based on facts in any way. It doesn't in any way separate the Xeon server processors from that date. Besides, both consumer (Core XXXX) and server (Xeon XXXX) processors are made from the exact same design these days, the only difference being that the Xeons have DUAL QPI links to allow them to communicate with a second processor in a dual CPU setup. Apart from QPI link count; both the consumer and server chips are identical nowadays. Therefore it's trivial for them to manufacture both at the same time, and the only constraint would be supply where they might have to choose whether to supply Server or Consumer users first. Either way, Q4 2010 is the first and only official information we have to go by right now.
http://www.pcmag.com/article2/0,2817,2362601,00.asp


Low-end chip production will indeed start Q4 this year but high-end with LGA 2011 socket won't be out before H2 2011 and Mac Pro uses high-end chips

This doesn't seem to be something you are into so please, don't argue about this

[Peace]

If you want Steve's view on the Mac read this (http://www.macnn.com/articles/10/06/10/promises.better.future.ahead/).

[cs4160]

Because Apple loves charging tomorrow's prices for yesterday's workstations. It would make sense, in the screwed up Apple way, for the Pros to be updated right before a new architecture becomes available.

What a choice...not sure what i prefer. Outdated HW out the gate, but i get my MP in a month, or wait until Xmas...With my luck, the MP comes out next month, i buy it, and then they break all the rules and do what we have been asking, tech turn quicky: MP 2011 in January....i cant even imagine the posts...

:)

[c.hilding]

Low-end chip production will indeed start Q4 this year but high-end with LGA 2011 socket won't be out before H2 2011 and Mac Pro uses high-end chips

This doesn't seem to be something you are into so please, don't argue about this

Found some articles with further information:
http://www.brightsideofnews.com/news/2010/4/16/intel-shows-off-sandy-bridge-at-idf-2010.aspx
http://www.xbitlabs.com/news/cpu/display/20100413235813_Intel_Begins_to_Ship_Samples_of_Sandy_Bridge_amid_Superior_32nm_Production_Ramp.html

Dual and Quad-Core low-end models come first, before 6-core and 8-core high-end models in Q2 2011. It still begs the question though: Why get Westmere now? Remember that Westmere is just a "tick" 32nm die-shrink of Nehalem with two more cores in (for a total of 6) in the highest-end CPUs. Why not wait for the "tock" (new architecture) in Q2 2011 and get USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores, and the higher performance of the new architecture?

If you want Steve's view on the Mac read this (http://www.macnn.com/articles/10/06/10/promises.better.future.ahead/).

Thank you very much for that article, it's obviously damage control due to the huge disappointment at WWDC '11 but at least it shows that Steve is aware.

[Umbongo]

You're right:
http://www.brightsideofnews.com/news/2010/4/16/intel-shows-off-sandy-bridge-at-idf-2010.aspx
http://www.xbitlabs.com/news/cpu/display/20100413235813_Intel_Begins_to_Ship_Samples_of_Sandy_Bridge_amid_Superior_32nm_Production_Ramp.html

Dual and Quad-Core low-end models come first, before 6-core and 8-core high-end models in Q2 2011. It still begs the question though: Why get Westmere now? Remember that Westmere is just a "tick" 32nm die-shrink of Nehalem with two more cores in (for a total of 6) in the highest-end CPUs. Why not wait for the "tock" (new architecture) in mid-2011 and get USB3, LightPeak, SATA3, native support for 1600 MHz RAM, 8 cores, and the higher power of the new architecture?

Because the processors and graphics cards currently used have all been replaced and customers expect new products, half of the ones Apple use have received 50% more cores. While we don't know why they aren't out yet for sure, it does seem processor supply may be a key issue rather than Apple holding off. If they don't update to Westmere then they will be selling 2 year old systems at some point.

[Digital Skunk]

Dual and Quad-Core low-end models come first, before 6-core and 8-core high-end models in Q2 2011. It still begs the question though: Why get Westmere now? Remember that Westmere is just a "tick" 32nm die-shrink of Nehalem with two more cores in (for a total of 6) in the highest-end CPUs. Why not wait for the "tock" (new architecture) in mid-2011 and get USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores, and the higher power of the new architecture?.

I think you have a point, but Apple just doesn't work that way. Not to be negative, but Apple will update the Mac Pros when they feel it's time to release them.

Honestly, we may be waiting this long because of a . . . I hate to say this but . . . a new case design or something else insane that Apple doesn't need to do with the Mac Pro.

Very rarely does Apple hold off on a product release because of new useful, up to date, technology. The last time this happened was for the switch from the G4 chip to the G5 chip and a new case design.

I don't see Apple putting in state of the art tech in a Mac Pro in Q4 and keeping the price at the crazy level it is now, or lowering it to where it should be.

[longball11]

If you want Steve's view on the Mac read this (http://www.macnn.com/articles/10/06/10/promises.better.future.ahead/).

Even though he said that. Sound bull ****. MAY emphasize next year's WWDC for mac? hahahaha. It's been iphone that and this the last two or three years in WWDC. Even with the iPad now...

[c.hilding]

Because the processors and graphics cards currently used have all been replaced and customers expect new products, half of the ones Apple use have received 50% more cores. While we don't know why they aren't out yet for sure, it does seem processor supply may be a key issue rather than Apple holding off. If they don't update to Westmere then they will be selling 2 year old systems at some point.

Yeah I thought about that after posting the reply; it'd obviously still be nice if Apple would get some sort of Westmere 6/12-core Mac Pros out, even though they'll be replaced with 8-core Sandy Bridge in April-June (Q2) 2011! At least current models wouldn't be so hopelessly old. The fact that Sandy Bridge isn't that far off is making me rethink everything though. Might just build a cheaper Hackintosh for now using last year's components and use that while waiting for the proper Sandy Bridge models in 2011, which is a sad state of affairs.

Oh well this thread isn't about me. I certainly do hope Apple releases Westmere models soon!

I think you have a point, but Apple just doesn't work that way. Not to be negative, but Apple will update the Mac Pros when they feel it's time to release them.

Honestly, we may be waiting this long because of a . . . I hate to say this but . . . a new case design or something else insane that Apple doesn't need to do with the Mac Pro.

Very rarely does Apple hold off on a product release because of new useful, up to date, technology. The last time this happened was for the switch from the G4 chip to the G5 chip and a new case design.

I don't see Apple putting in state of the art tech in a Mac Pro in Q4 and keeping the price at the crazy level it is now, or lowering it to where it should be.

Yes, it's often the case that the most powerful Macs are Hackintoshes, it's freaking horrible. :( You don't have that problem on the PC side, where you can just get the most powerful components and build anything cheaply. That's why most of the 3D rendering industry is back on Windows or Linux workstations now. I've seen one Windows machine with 288 GB Ram, dual X5680s, 24 SSDs in a RAID configuration. To say that it edited HD movies quickly is an understatement. You can't get anything even close to that in a Mac.

Even though he said that. Sound bull ****. MAY emphasize next year's WWDC for mac? hahahaha. It's been iphone that and this the last two or three years in WWDC. Even with the iPad now...

I know, depressing...

[Umbongo]

Yeah I thought about that after posting the reply; it'd obviously be nice if Apple would get some Westmere 6/12-core Mac Pros out, even though they'll be replaced with 8-core Sandy Bridge in April-June (Q2) 2011! At least current models wouldn't be so hopelessly old. The fact that Sandy Bridge isn't that far off is making me rethink everything though. Might just build a cheaper Hackintosh for now and wait for the proper Sandy Bridge models in 2011, eww.

Oh well this thread isn't about me. I certainly do hope Apple releases Westmere models soon!

Don't forget we are only talking about all this because Apple haven't released new systems almost 3 months after the processors were launched. Sandy Bridge's release date doesn't much matter.

[VirtualRain]

The reason for this years focus on iOS should be obvious.

Apple has three lines of business that are all approx. $10B business units:
- Mac
- iPod
- iPhone

Their Mac market is growing at a tremendous rate, almost all by itself (see my post on why iDevices are good for the Mac).

The iPod market is somewhat stagnant, and to some extent being cannibalized by Apple's own more sophisticated iOS devices. It's a market in decline.

The iPhone market is under assault by Google and it's critical that Apple focus on competing strongly here.

Finally, Apple is trying to develop a new line of business around tablets that they hope will be a fourth $10B business, and it too is already in the sights of many competitors. If they want to achieve $10B like revenues in this area, it's going to require a massive effort.

To have done anything other than focus this years WWDC on catering to iOS developers would have been a bad decision for their two business units which are most vulnerable.

You guys need to think with your Apple business hat on. Presumably, all of the new customers Apple is bringing on board, is great news for the pros who frequent this forum... it means more customers to consume whatever you are developing.

[c.hilding]

Don't forget we are only talking about all this because Apple haven't released new systems almost 3 months after the processors were launched. Sandy Bridge's release date doesn't much matter.

The reason Sandy Brudge was brought up was that Intel's CEO has finally given a release date for Sandy Bridge (before this it was all speculation); and that the date given was not far off (Q4 2010). However, as was pointed out in the thread, it then turned out that the high-end 6- and 8-core models will come 3 months after the Sandy Bridge release, which puts it a bit further into the future (April+ 2011), basically removing it from the radar as far as what will be in the next Mac Pro goes, but the date is close enough that it's still worth carefully thinking about your next purchase.

Basically, to get the most out of your buck, you have these choices:

If you have a perfectly good workstation now and can use it until April+ 2011, then wait. You'll be getting Sandy Bridge's USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores per processor, and higher performance, and it'll be a much better buy than ANY Westmere of today (which is merely a die-shrink of November 2008's Nehalem).

If you don't have a workstation now but need one, then either get a Westmere Mac Pro workstation when they come out if you don't mind the entire architecture becoming obsolete in 9-10 months (as opposed to just a die-shrink (http://en.wikipedia.org/wiki/Intel_Tick-Tock#Roadmap) such as Nehalem -> Westmere), or get something cheaper for now such as last year's technology; either via a refurb, second-hand or a cheap Hackintosh. I'll most likely hit this camp. They are still powerful machines and will be excellent to tide you over into 2011, where you'll be able to get the vastly superior Sandy Bridge (SB) (http://www.homestarrunner.com/sbemail.html) <-- (click) architecture.

If you don't have a need for a workstation now, but want one later, then wait for Sandy Bridge and use something like an iMac, MacBook or Mac Mini for now.

[Hellhammer]

You're right:
http://www.brightsideofnews.com/news/2010/4/16/intel-shows-off-sandy-bridge-at-idf-2010.aspx
http://www.xbitlabs.com/news/cpu/display/20100413235813_Intel_Begins_to_Ship_Samples_of_Sandy_Bridge_amid_Superior_32nm_Production_Ramp.html

Dual and Quad-Core low-end models come first, before 6-core and 8-core high-end models in Q2 2011. It still begs the question though: Why get Westmere now? Remember that Westmere is just a "tick" 32nm die-shrink of Nehalem with two more cores in (for a total of 6) in the highest-end CPUs. Why not wait for the "tock" (new architecture) in Q2 2011 and get USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores, and the higher power of the new architecture?



Thank you very much for that article, it's obviously damage control but at least it shows that Steve is aware.

Q2 is just a rumor, many other articles have stated H2 2011, e.g. this (http://translate.google.com/translate?js=y&prev=_t&hl=en&ie=UTF-8&layout=1&eotf=1&u=http://pc.watch.impress.co.jp/docs/column/kaigai/20100409_360112.html&sl=ja&tl=en). They also stated in the article you linked that Q2 2011 or later so it's just a pure guess and I wouldn't bet my money on Q2. Production may start then thus if Apple gets them earlier, we may see SB Mac Pro in summer 2011, but that's just IF.

There has also been some rumors about Becton going 32nm in early 2011 so that would fulfill some Xeon market before SB hits the high-end market.

Too early to give other than guesses. There is no official info about Sandy Bridge Xeons

[ValSalva]

If you want Steve's view on the Mac read this (http://www.macnn.com/articles/10/06/10/promises.better.future.ahead/).

Is SJ responding to the Newsweek column or the sender of the email?

The sender said: "I believe and hope that the Mac will remain a vibrant, vital part of Apple's future and one of its (admittedly many) product lines" He then asks SJ: "So, as you view it, does the Mac have a long and important history ahead of it?" And SJ responds: "Completely wrong. Just wait."

Who's wrong is SJ's estimation, the emailer or those that speculate about the Macopalypse? Maybe SJ is too cryptic :D

[ValSalva]

Basically, to get the most out of your buck, you have these choices:

If you have a perfectly good workstation now and can use it until April+ 2011, then wait. You'll be getting Sandy Bridge's "USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores, and higher performance", and it'll be a much better buy than ANY Westmere of today (which is merely a die-shrink of November 2008's Nehalem).

If you don't have a workstation now but need one, then either get a Westmere workstation when they come out if you don't mind the entire architecture becoming obsolete in 9-10 months, or get something cheaper for now such as last year's technology (I'll most likely hit this camp).

If you don't have a need for a workstation now, but want one later, then wait for Sandy Bridge and use something like an iMac, MacBook or Mac Mini for now.

What cheaper option are you referring to in your second scenario? A refurb? And are you really disciplined enough to pass on the new Mac Pro that hypothetically comes out at the end of this month :D

I agree with the tone of your posts and share in your frustration.

[xgman]

It's Apple. Add a year on to the Intel release before Apple allows us to buy it. :rolleyes:

[c.hilding]

Is SJ responding to the Newsweek column or the sender of the email?

The sender said: "I believe and hope that the Mac will remain a vibrant, vital part of Apple's future and one of its (admittedly many) product lines" He then asks SJ: "So, as you view it, does the Mac have a long and important history ahead of it?" And SJ responds: "Completely wrong. Just wait."

Who's wrong is SJ's estimation, the emailer or those that speculate about the Macopalypse? Maybe SJ is too cryptic :D

Very nice catch, I wondered the same when reading the article, but assumed they were just quoting some reply that some regular person had received and tried to pass it off as Steve Jobs responding to their article. :p Either way I'm glad Steve Jobs says it's "Completely wrong" to think that they've given up on the Pro market.

What cheaper option are you referring to in your second scenario? A refurb? And are you really disciplined enough to pass on the new Mac Pro that hypothetically comes out at the end of this month :D

I agree with the tone of your posts and share in your frustration.

Thanks, there are quite a lot of us now, Pro users that are fed up with being treated like dirt just because it's all about the "iToys" now.

What I meant about the second scenario was getting last year's technology either via a refurb, second-hand or a cheap Hackintosh. They are still powerful machines and will be excellent to tide you over into 2011, where you'll be able to get the vastly superior Sandy Bridge (SB) (http://www.homestarrunner.com/sbemail.html) <-- (click) architecture. I'll update the post to clarify this.

It's Apple. Add a year on to the Intel release before Apple allows us to buy it. :rolleyes:

Hopefully not. Apple has received early exclusive access from Intel in the past (http://news.cnet.com/apple-beats-intel-to-nehalem-ep-chip-launch/), so it's not impossible that they'll be first out. Even if they do decide to screw us over and delay a Sandy Bridge update next year, it's always easy to build one yourself with off-the-shelf parts (Hackety-hack), which in my opinion is perfectly fine to do when we're dealing with a company that doesn't treat their Pro customers' needs right. Let's hope that Westmere upgrade comes soon so that 2009's second-hand value drops further. ;-) The 2009's will be nice and cheap and powerful enough while waiting for Sandy Bridge.

[strausd]

By the time these are put in Mac Pros, Intel will announce new CPUs. Then when those are put in, they will announce even newer CPUs. Technology is always growing and always will.

[c.hilding]

By the time these are put in Mac Pros, Intel will announce new CPUs. Then when those are put in, they will announce even newer CPUs. Technology is always growing and always will.

Westmeres are not "new CPUs", they are die-shrunk (size-reduced) Nehalems, giving them slightly better performance and allowing them to fit more cores inside each processor (they went from 4-core to 6-core) but doesn't offer any new features.

Intel alternates between releasing new architectures and then shrinking those architectures, in a tick-tock fashion. You can read about it here:
http://en.wikipedia.org/wiki/Intel_Tick-Tock#Roadmap

Sandy Bridge is a big deal and will bring loads of the latest technology (USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores per processor, and higher performance); whereas the Westmere that came out 3 months ago is just a Nehalem that has been reduced in size.

The reason it's being discussed is that this completely new architecture is not that far off, and will replace Nehalem/Westmere with a bunch of new technology and ports that will soon be required by more and more modern devices. Consumer models of Sandy Bridge will be mass-produced in Q4 2010, and server-models in April+ (Q2) 2011. Read the first post for a summary and possibly some help on whether you should wait or not.

[chrmjenkins]

Two points. Westmere will not become "obsolete" when Sandy Bridge launches. It will still be an incredibly capable architecture. Second, correct me if I'm wrong, but didn't the 08 Pro upgrade include CPUs that no one else had access to yet?

[Hellhammer]

Sandy Bridge is a big deal and will bring loads of the latest technology; whereas Westmere is just a Nehalem that has been reduced in size (giving them slightly better performance and allowing them to fit more cores inside each processor).

You call ~50% increase just a slight one? I think two extra cores are pretty significant in terms of performance, especially when they run at some clocks as quad cores. Sandy Bridge may not bring eight core for Mac Pro yet because of price, who knows? Of course SB will bring other improvements e.g. AVX.

Sandy Bridge is still at least a year away and we never know when Apple will use them, just like they have done with Westmere. See my post from previous post for some other details about the release date

[Digital Skunk]

Yes, it's often the case that the most powerful Macs are Hackintoshes, it's freaking horrible. :( You don't have that problem on the PC side, where you can just get the most powerful components and build anything cheaply. That's why most of the 3D rendering industry is back on Windows or Linux workstations now. I've seen one Windows machine with 288 GB Ram, dual X5680s, 24 SSDs in a RAID configuration. To say that it edited HD movies quickly is an understatement. You can't get anything even close to that in a Mac.

Very true. The other side of the spectrum has the best hardware in the industry and more options to choose from compared to the Mac side. The OS we have is a dream, and our hardware design is top notch, but we just can't get bleeding edge hardware for the life of us.

Sounds like that cutter is using a server as his workstation, but I can see him using it and upgrading it for years to come. Oh, and it must have paid close to $100,000 for it too.

Just once I'd like to see Apple give us bleeding edge in workstation or 17" laptop. I think they did back when they offered SAS drives, but I am sure they didn't sell much of them, since anyone with half an IT mind would just buy them from somewhere else.

I'd also love to see Apple offer a barebones Mac Pro kit to bring the price down. Just give me the mobo, system discs, processors, and daughter board w/risers. Let me grab everything else.

[c.hilding]

Two points. Westmere will not become "obsolete" when Sandy Bridge launches. It will still be an incredibly capable architecture. Second, correct me if I'm wrong, but didn't the 08 Pro upgrade include CPUs that no one else had access to yet?

Of course Nehalem/Westmere will still be incredibly powerful as far as CPU performance goes after Sandy Bridge is released, but it will no longer have the latest technology, and we're talking about groundbreaking things like USB3, SATAIII, LightPeak and PCI Express 3.0 (which both doubles the bandwidth over 2.0 and has many more lanes per CPU, giving you more PCI Express ports in the computer), where we'll eventually have devices and cards that require these ports. Wouldn't you rather have a computer that contains these ports? Anyone who doesn't at least look over the implications of buying now versus waiting is foolish. This is a completely new architecture right around the corner, with multiple new standards technologies, things that will just be in more and more demand as modern devices are released that will require USB3, SATAIII, LightPeak or PCI Express 3.0. It's your money and I'm not saying you shouldn't buy a Westmere Mac now. Re-read the last portion of the 1st post, where you see the different scenarios, and you'll see that I said it's perfectly fine to buy a Westmere now as long as you know what you're getting into. And yes, Apple had early access in 2009; check post #28 here, the last paragraph, for more on that.

You call ~50% increase just a slight one? I think two extra cores are pretty significant in terms of performance, especially when they run at some clocks as quad cores. Sandy Bridge may not bring eight core for Mac Pro yet because of price, who knows? Of course SB will bring other improvements e.g. AVX.

Sandy Bridge is still at least a year away and we never know when Apple will use them, just like they have done with Westmere. See my post from previous post for some other details about the release date

You misread the portion you quoted; I said "Sandy Bridge is a big deal and will bring loads of the latest technology; whereas Westmere is just a Nehalem that has been reduced in size (giving them slightly better performance and allowing them to fit more cores inside each processor).". The important word is in the portion you yourself quoted: "and". ;) I was talking about performance per-core only being "slightly better"; in other words, the die-shrink leads to slightly more effective electrical paths giving slightly better performance per-core AND the reduced size ALSO allows more cores per processor.

Now, when it comes to technology; as mentioned in my reply to chrmjenkins in this post (above the reply to you), Sandy Bridge brings very important new technology that will eventually be required by devices, and then your Nehalem/Westmere will officially be obsolete. That is why post #1 of this thread talks about the different scenarios we're all going to have to face, and just wants to inform people so that everyone can make the call on what they want to do. For a more elaborate reply on this technology-point of what Sandy Bridge brings, see the reply to chrmjenkins in this very post.

Lastly, regarding "we never know when Apple will use them", that is a non-issue, see the last paragraph of post #28 here.

[nanofrog]

You call ~50% increase just a slight one? I think two extra cores are pretty significant in terms of performance, especially when they run at some clocks as quad cores.
This is only correct under specific conditions though, and won't be the case all the time (depends on the software run).

There's a few multi-threaded applications that seem to be fairly common, that can't actually utilize all the cores. Photoshop certainly comes to mind. Granted, it's possible to attempt running 3x instances simultaneously, but from what I understand, it's not all that viable due to the input requirement from the user.

I'd also love to see Apple offer a barebones Mac Pro kit to bring the price down. Just give me the mobo, system discs, processors, and daughter board w/risers. Let me grab everything else.
I don't see this ever happening. The inexpensive parts they do put in, allows for high margins that ultimately benefit their bottom line.

Then there's the technical side, primarily around the problem of requiring them to have parts for validation testing (and there are other issues as well, such as attracting 3rd party suppliers for graphics cards, Intellectual Property,...).

[Hellhammer]

Lastly, regarding "we never know when Apple will use them", that is a non-issue, see the last paragraph of post #28 here.

That has nothing to do with it :confused: You can't use Sandy Bridge in old Mac Pro when they come out and Hackintosh isn't same as real Mac Pro. If you are fine with Hackintosh then you shouldn't even complain about Mac Pro as it will never be as expandable as Hack is. Apple may screw us up like they are doing now.

Of course we all, including me even though you might think I'm the "party ruiner" now, hope that Apple still care about Mac Pro and gets early access for SB. Maybe Apple indeed think that Westmere is too small upgrade so they are waiting for SB to get early access and surprise us all (by that time Southern Islands from ATI should be out as well) with enormous update ;)

This is only correct under specific conditions though, and won't be the case all the time (depends on the software run).

There's a few multi-threaded applications that seem to be fairly common, that can't actually utilize all the cores. Photoshop certainly comes to mind. Granted, it's possible to attempt running 3x instances simultaneously, but from what I understand, it's not all that viable due to the input requirement from the user.

In terms of raw processing power, it's about 50% faster :p I know software is unfortunately lagging behind thus Sandy Bridge will be so significant (better clock fro clock, core for core performance).

Hopefully we'll see some actual benchmarks soon so we can start the real drooling :D

[iMacmatician]

Server Sandy Bridge is Q3 2011 for -EP (and high-end desktop) and Q4 2011 for -EN. Earlier reports said Q1 or Q2.

There has also been some rumors about Becton going 32nm in early 2011 so that would fulfill some Xeon market before SB hits the high-end market. Most recent reports say mid 2011.

[Digital Skunk]

Then there's the technical side, primarily around the problem of requiring them to have parts for validation testing (and there are other issues as well, such as attracting 3rd party suppliers for graphics cards, Intellectual Property,...).

True, and i know it's a pipe dream.

I don't know if HP and Dell still sell barebones kits, but there was a time that one could go to their sites, and get just the case, risers, boards; and pick up RAM, HDD, optical drives, GFX cards, etc. 3rd party.

[Hellhammer]

Server Sandy Bridge is Q3 2011 for -EP (and high-end desktop) and Q4 2011 for -EN. Earlier reports said Q1 or Q2.

Has that been confirmed or is it just an educated guess made by PC Watch? :p

Most recent reports say mid 2011.

Couple months this or that way, doesn't matter, we won't see them in Mac Pros anyway (Becton is 3500$ chip) :cool:

[Umbongo]

Two points. Westmere will not become "obsolete" when Sandy Bridge launches. It will still be an incredibly capable architecture. Second, correct me if I'm wrong, but didn't the 08 Pro upgrade include CPUs that no one else had access to yet?

There were rumours Apple purchased all the 1600MHz parts, and they were the only one to use them for a while. There were 1333MHz parts available at the same time though. In 09 they launched a month before other's had Nehalem Xeons.

[nanofrog]

Maybe Apple indeed think that Westmere is too small upgrade so they are waiting for SB to get early access and surprise us all (by that time Southern Islands from ATI should be out as well) with enormous update ;)
This time around is like 2007, when they had an option for Quad core parts, creating the first run of Octad systems (all '06 DP systems were only Quads).

They only had to drop-in a new chip (didn't even need any changes in the firmware). They will have to update the firmware this time, but it's not that big a deal in terms of time or investments (unlike the Tick portion of the cycle, where the entire board changes due to the new architecture).

Due to that cost, Intel has created the Tick Tock cycle, as that huge chunk is amortized over 2 years (same boards = same chipset, socket,....).

By skipping the Gulftown parts, they're doing their MP segment a disservice, as some users (those that can in terms of finances, and may have been thinking about it for a little while on the whole, which more importantly IMO, includes software), switch over to a different OS and system vendor (i.e. Dell, HP,...).

In terms of raw processing power, it's about 50% faster :p I know software is unfortunately lagging behind thus Sandy Bridge will be so significant (better clock fro clock, core for core performance).
Only under ideal conditions for Symetric Multiprocessing capable applications (can actually use n cores, not limited to a fixed number, and not hampered by bottlenecks that may exist).

Real world performance won't be that much. More like what we saw between Harpertown and Nehalem's real world data (so 10% - 20% is more along what we'll see, and this is for SMP capable apps).

Single threaded applications would benefit, assuming the processors used are in fact what's expected/hoped for, as they've higher clocks for the same cost segments (P/N breakdown in terms of cost, and matching them with the initial cost for Nehalem parts). Same architecture @ a faster clock = better performance, single or multi-threaded applications. But single threaded could also gain a bit more due to Turbo Mode, assuming the other cores aren't loaded enough to prevent it from kicking in.

I don't know if HP and Dell still sell barebones kits, but there was a time that one could go to their sites, and get just the case, risers, boards; and pick up RAM, HDD, optical drives, GFX cards, etc. 3rd party.
I don't recall seeing that ability any longer, though it's worth a look. ;)

I'm more accustomed to "bare bones" requiring a graphics card (assuming it's separate as it would be in a workstation) and HDD, though they're low cost enough most will upgrade them to what's needed, and keep them as spares (without any real complaint, though this may be to do with the fact it's usually funded by a corporate budget, not out of their own pocket :p).

Has that been confirmed or is it just an educated guess made by PC Watch? :p
It's information that's out there, though the dates will change over time (usually getting later as time goes on). This type of information filtration is rather common, as Intel does release information to industry concentric sites.

I know there can be issues with Wiki, but generally speaking, it's gotten a lot better, and is a good source to find a good bit of information once it's available (articles get scrounged for info, and it's compiled/cobbled together at a single location).

[iMacmatician]

Has that been confirmed or is it just an educated guess made by PC Watch? :pNothing's really confirmed at this point in time.

An old Intel roadmap from late last year (.pdf was pulled but I got it in time) put -EP at Q2 2011 and -EN at mid 2011. A number of CPUs were probably pushed back since then (Nehalem-EX was slated for turn of 2010 on that roadmap, it ended up launching late Q1 2010).

As for PC Watch, they seem fairly reliable, except that any stuff in question marks is most likely speculation.

[Hellhammer]

It's information that's out there, though the dates will change over time (usually getting later as time goes on). This type of information filtration is rather common, as Intel does release information to industry concentric sites.

I know there can be issues with Wiki, but generally speaking, it's gotten a lot better, and is a good source to find a good bit of information once it's available (articles get scrounged for info, and it's compiled/cobbled together at a single location).

Nothing's really confirmed at this point in time.

An old Intel roadmap from late last year (.pdf was pulled but I got it in time) put -EP at Q2 2011 and -EN at mid 2011. A number of CPUs were probably pushed back since then (Nehalem-EX was slated for turn of 2010 on that roadmap, it ended up launching late Q1 2010).

As for PC Watch, they seem fairly reliable, except that any stuff in question marks is most likely speculation.

I only asked because iMacmatician usually has some very cool articles and roadmaps to show, so by asking his source, I get a hold of them :D

I know it's what we currently know and I also use Wikipedia as my main source (look at their sources for extra info) because it's so simple and usually has pretty decent stuff.

We'll know more when the time passes, there will still likely be a Westmere Mac Pro before Sandy Bridge so... How about speculating Ivy Bridge and Haswell, maybe even Rockwell Mac Pro now :p

[c.hilding]

If you are fine with Hackintosh then you shouldn't even complain about Mac Pro as it will never be as expandable as Hack is. Apple may screw us up on sandy bridge like they are doing now with westmere.

Of course we all, including me, even though you might think I'm the "party ruiner" now, hope that Apple still cares about Mac Pro and gets early access for SB. Maybe Apple indeed thinks that Westmere is too small of an upgrade so they are waiting for SB to get early access and surprise us all.

Naw. You're definitely not a party pooper by saying that you hope Apple gets Sandy Bridge before everyone else. ;) I am an Apple fan and would love for them to get early access, I'll explain my stance on all of this:

Official hardware is always more fun than Hackintoshes, since it all "just works" without having to fiddle. That beings said; today, Hackintoshes have come very far and are very easy to install and manage. I've been in the Hackintosh scene since 2004 with PearPC (yes, the extremely slow PowerPC emulator for PCs that could just about run Mac OS X at about 1 FPS). I've built numerous hacks throughout these six years, and it's so easy nowadays that it's ridiculous, and they run as stably as real Mac Pros now (just be sure to choose Intel hardware that is close to what's in real Macs and all the builtin Mac OS drivers will "just work"). However, despite the ease of current Hackintosh installations, I'll always prefer Apple hardware even with their price premiums because of the fact that you can focus on working instead of having to set anything up; BUT if Apple doesn't put out Sandy Bridge in a timely fashion, then nothing will stop me and numerous other people from buying the components and building Sandy Bridge machines ourselves. :D :apple:

That should hopefully explain my stance on what will happen as Sandy Bridge nears release. The action to be taken all depends on how much Apple likes their Pro customers next year, I'd prefer an official SB machine but there's nothing wrong with building one yourself if they slip again like they did this year! ;)

PS: Hackintosh people have already built dual-processor Westmere Xeon machines, thanks to EVGA's SR-2 motherboard released now (early June). Pretty nice eh! Those machines can have 2x X5680's and overclock to speeds around 4.5 GHz per core on air-cooling. That's another benefit Hackety-hacks bring; overclocking. Even people on a budget could get a cheap X5650 and overclock it to reach the same speeds as the twice as pricy X5680. ;)

[iMacmatician]

According to this (http://forum.coolaler.com/showthread.php?t=240578), there are some Sandy Bridge (4-core) benchmarks for 2.5 GHz model.

[Eidorian]

Sandy Bridge Xeon parts are an eternity from this point. Mobile and desktop mainstream are coming first. If they're going to do anything this year it feels like it's going to be a silent introduction of Gulftown and a new high end GPU option.

[c.hilding]

We'll know more when the time passes, there will still likely be a Westmere Mac Pro before Sandy Bridge so...

Yes, looking at all the known facts, a Westmere Mac Pro is very likely considering the dates for SB's release and the statement issued by Steve Jobs today, saying that we shouldn't worry that they're ignoring the Mac Pros and to "just wait" for the coming update. His comment even makes it sound like it's not far off. Strange that he didn't mention a word about it at WWDC, as that would have spared a lot of frustration for their Pro customers. Apple really seems to underestimate the need for a new Mac Pro, and Steve Jobs had to issue that statement due to the overwhelming negativity directed at Apple right now, but at least his statement implies that new Mac Pros ARE on the way, and the brief and smug "Just wait" implies it could be something massive (click for full story behind his comment (http://www.macstories.net/news/steve-jobs-on-the-death-of-the-mac-completely-wrong/)). Of course we could speculate all we want about Apple getting early exclusivity on Sandy Bridge and preparing a huge announcement, skipping Westmere, keeping quiet due to Sandy Bridge NDA, etc, but that would just be setting ourselves up for massive disappointment since that's not a very realistic scenario (it's Q3 2010, and Intel just now confirmed that they'll start shipping Consumer SB's in Q4 2010). The best we can hope for now is Westmere and I have a feeling we should "just wait", at least we've had a life-sign from the man himself and that's good enough for me. It was the silence combined with the waiting that was killing me and most other people. ;)

How about speculating Ivy Bridge and Haswell, maybe even Rockwell Mac Pro now :p

Why would we do that? Look, there's nothing to ridicule about this thread: Sandy Bridge is imminent, in the near future, and will bring no less than FOUR entirely new standards and evolutions (USB3, SATAIII, PCI Express 3.0 and LightPipe), which will make all of today's computers obsolete, as these new technologies and ports will be required by more and more modern devices as time goes on. That is why it's something that should be taken into account before spending 3 years of saved up cash on a fully loaded Westmere Mac, when you could buy a second-hand/refurb/make a hackintosh to tide you over until the real quantum leap with Sandy Bridge. Unless you are made of money, that will be a much wiser choice. I happen to be made of money and I still won't be buying any fully kitted out Westmere, it's just not a wise choice with such a quantum leap coming up quite soon. The best thing to do is just to get something to get by with while waiting; and the current machines are very powerful, so a refurb/second-hand/hackintosh will be the best choice for most people. Those that need the 4 extra cores (from 2x4 (8) to 2x6 (12)) are free to buy a Westmere now, if they really need it, but even most editors would get by excellently with the 2009 Mac Pro.

[Hellhammer]

According to this (http://forum.coolaler.com/showthread.php?t=240578), there are some Sandy Bridge (4-core) benchmarks.

Wow! Seems to deliver decent bump in performance. Over 40% faster than 3GHz Phenom x4, looks very nice. This is the reason I asked some confirmation, you always seem to find these great sites ;)

[Eidorian]

Wow! Seems to deliver decent bump in performance. Over 40% faster than 3GHz Phenom x4, looks very nice. This is the reason I asked some confirmation, you always seem to find these great sites ;)TechPowerUp! (http://www.techpowerup.com/124273/Intel_Sandy_Bridge_Quad-Core_Processor_Tested.html) has a much cleaner compilation of benches from that same thread.

[Ravich]

Yes, looking at all the known facts, a Westmere Mac Pro is very likely considering the dates for SB's release and the statement issued by Steve Jobs today, saying that we shouldn't worry that they're ignoring the Mac Pros and to "just wait" for the coming update. His comment even makes it sound like it's not far off.
What else was he going to say?


Question about lightpeak: will there be a way to get lightpeak to older mac pros when it becomes available?

[Eidorian]

Question about lightpeak: will there be a way to get lightpeak to older mac pros when it becomes available?Get a PCIe expansion card.

[Hellhammer]

Question about lightpeak: will there be a way to get lightpeak to older mac pros when it becomes available?

Likely yes with a PCIe card

[Salavat23]

IMO, for professionals looking for more performance, Sandy Bridge is irrelevant. Look at this sneak peak of SB performance and you will see why:

http://forum.coolaler.com/showthread.php?t=240578

Performance is inline with what is on the market right now. If you extrapolate the data, SB may be 1-2% faster at most at CPU heavy tasks.

SB is not worth getting if all you want is performance. However, if you're looking for better efficiency and more features, only then is it worth taking a look IMO. Intel's Nehalems will be neck and neck with the SB cpus two years from now in terms of performance.

[Hellhammer]

IMO, for professionals looking for more performance, Sandy Bridge is irrelevant. Look at this sneak peak of SB performance and you will see why:

http://forum.coolaler.com/showthread.php?t=240578

Performance is inline with what is on the market right now. If you extrapolate the data, SB may be 1-2% faster at most at CPU heavy tasks.

SB is not worth getting if all you want is performance. However, if you're looking for better efficiency and more features, only then is it worth taking a look IMO. Intel's Nehalems will be neck and neck with the SB cpus two years from now in terms of performance.

How is that? Sandy Bridge doesn't that much if any software support like adding cores do. Nehalem provided up to 30% better performance than Core architecture did.

For example in this (http://www.techpowerup.com/img/10-06-10/42g.png) benchmark, the 2.5GHz SB beats 3.33GHz i7. Power efficiency means better performance because more cores and faster clocks can be used

[Just Switched]

Of course Nehalem/Westmere will still be incredibly powerful as far as CPU performance goes after Sandy Bridge is released, but it will no longer have the latest technology, and we're talking about groundbreaking things like USB3, SATAIII, LightPeak and PCI Express 3.0 (which both doubles the bandwidth over 2.0 and has many more lanes per CPU, giving you more PCI Express ports in the computer), where we'll eventually have devices and cards that require these ports. Wouldn't you rather have a computer that contains these ports? Anyone who doesn't at least look over the implications of buying now versus waiting is foolish. This is a completely new architecture right around the corner, with multiple new standards technologies, things that will just be in more and more demand as modern devices are released that will require USB3, SATAIII, LightPeak or PCI Express 3.0. It's your money and I'm not saying you shouldn't buy a Westmere Mac now. Re-read the last portion of the 1st post, where you see the different scenarios, and you'll see that I said it's perfectly fine to buy a Westmere now as long as you know what you're getting into. And yes, Apple had early access in 2009; check post #28 here, the last paragraph, for more on that.



You misread the portion you quoted; I said "Sandy Bridge is a big deal and will bring loads of the latest technology; whereas Westmere is just a Nehalem that has been reduced in size (giving them slightly better performance and allowing them to fit more cores inside each processor).". The important word is in the portion you yourself quoted: "and". ;) I was talking about performance per-core only being "slightly better"; in other words, the die-shrink leads to slightly more effective electrical paths giving slightly better performance per-core AND the reduced size ALSO allows more cores per processor.

Now, when it comes to technology; as mentioned in my reply to chrmjenkins in this post (above the reply to you), Sandy Bridge brings very important new technology that will eventually be required by devices, and then your Nehalem/Westmere will officially be obsolete. That is why post #1 of this thread talks about the different scenarios we're all going to have to face, and just wants to inform people so that everyone can make the call on what they want to do. For a more elaborate reply on this technology-point of what Sandy Bridge brings, see the reply to chrmjenkins in this very post.

Lastly, regarding "we never know when Apple will use them", that is a non-issue, see the last paragraph of post #28 here.

There are alreay westmere boards that have sata 3, usb3, 7 pci-e 16 lane slots etc already. Pci e 3 boards are right around the corner These things will by no means be new when sandy bridge comes out. When sandy bridge comes out alot of the features you speak of as being new wil have already been out for year integrated into motherboards or available as add ons with pci e cards. Sandy Bridge is going to be great and there will be bus bandwith improvements for sure as well as genuinely new processor but as far as features there's not a whole lot new.

[c.hilding]

IMO, for professionals looking for more performance, Sandy Bridge is irrelevant. Look at this sneak peak of SB performance and you will see why:

http://forum.coolaler.com/showthread.php?t=240578

Performance is inline with what is on the market right now. If you extrapolate the data, SB may be 1-2% faster at most at CPU heavy tasks.

SB is not worth getting if all you want is performance. However, if you're looking for better efficiency and more features, only then is it worth taking a look IMO. Intel's Nehalems will be neck and neck with the SB cpus two years from now in terms of performance.

Not once has this thread touted PERFORMANCE as the main reason to wait for Sandy Bridge (it will be faster, though). From post one, and consistently throughout the thread, the issue has been the same. I'll quote a summary from a previous reply to give you an overview of the issue:

Sandy Bridge is imminent, in the near future, and will bring no less than FOUR entirely new standards and evolutions (USB3, SATAIII, PCI Express 3.0 and LightPipe), which will make all of today's computers obsolete, as these new technologies and ports will be required by more and more modern devices as time goes on. That is why it's something that should be taken into account before spending 3 years of saved up cash on a fully loaded Westmere Mac, when you could buy a second-hand/refurb/make a hackintosh to tide you over until the real quantum leap with Sandy Bridge. Unless you are made of money, that will be a much wiser choice. I happen to be made of money and I still won't be buying any fully kitted out Westmere, it's just not a wise choice with such a quantum leap coming up quite soon. The best thing to do is just to get something to get by with while waiting; and the current machines are very powerful, so a refurb/second-hand/hackintosh will be the best choice for most people. Those that need the 4 extra cores (from 2x4 (8) to 2x6 (12)) are free to buy a Westmere now, if they really need it, but even most editors would get by excellently with the 2009 Mac Pro.

There are alreay westmere boards that have sata 3, usb3, 7 pci-e 16 lane slots etc already. Pci e 3 boards are right around the corner.

I think I've already mentioned what you're saying in this thread. Current implementations exist, yes; however, they are unofficial non-Intel controller chipsets. The Intel solution has not been released yet, and therefore they will not be in the Mac Pro. Nor will there be Mac OS X drivers for those unofficial chips if you build a Hackintosh with current motherboards. One such board is the EVGA SR-2, and your only chance at getting support for your precious USB3 and SATA3 ports is if you try to port open-source Linux or UNIX drivers to Mac OS X, which is something that the Hackintosh community does at times, but don't bet on being able to boot from a SATA3-connected disk. Basically, NO these things don't exist until Intel officially brings them out, which they'll be doing with Sandy Bridge and the corresponding new reference motherboards. As for your last comment, "PCI Express 3.0 boards are just around the corner"; no, they are not. PCI Express is dependent on the CPU. For instance, each Nehalem/Westmere Xeon provides thirtytwo (32) "1x" lanes of PCI Express 2.0; with two CPUs that's sixtyfour (64) "1x" lanes, or a total of "64x" of bandwidth. This is then split across the various connectors on the board in any way the manufacturer prefers. Apple, for instance, chose to go for a 4x16 (=64) setup, but could just as well have gone for 2x16 (=32) + 4x8 (=32) = 64 (which would have given six ports, and still only used 64 lanes). Other manufacturers, like EVGA, use Nvidia NF200 line doublers, which is basically an intelligent queuing/messaging passthrough device that allows you to connect multiple fast devices to a single lane, and therefore artificially increases the number of lanes and lets you create configurations such as EVGA SR-2's seven (7) 16x lanes. For PCI Express 3.0, you will need new Sandy Bridge CPUs to take care of the increasing demands such as doubled polling rate and more lanes from the CPU. Please think and be informed before you speak.

Question about lightpeak: will there be a way to get lightpeak to older mac pros when it becomes available?

Yes, Intel has designed some Light Peak PCI Express cards that you will be able to use in existing computers, but it probably won't be as fast as a builtin solution.

[Salavat23]

How is that? Sandy Bridge doesn't that much if any software support like adding cores do. Nehalem provided up to 30% better performance than Core architecture did.

For example in this (http://www.techpowerup.com/img/10-06-10/42g.png) benchmark, the 2.5GHz SB beats 3.33GHz i7. Power efficiency means better performance because more cores and faster clocks can be used

In another Everest benchmark, the SB CPU scores much lower. We don't know what part of the CPU they are benchmarking (cache latency, etc).

[Eidorian]

Intel's USB 3.0 implementation is still scheduled for 2012 on their controllers. Several other USB 3.0 controller vendors stepped up during Computex as well in addition to NEC, VIA, and Texas Instruments.

(LGA 2011) processors are going to bring PCIe 3.0 but SATA 6 Gbps is going to be available across the board on all sockets.

[brentsg]

Dual and Quad-Core low-end models come first, before 6-core and 8-core high-end models in Q2 2011. It still begs the question though: Why get Westmere now? Remember that Westmere is just a "tick" 32nm die-shrink of Nehalem with two more cores in (for a total of 6) in the highest-end CPUs. Why not wait for the "tock" (new architecture) in Q2 2011 and get USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth), native support for 1600 MHz RAM, 8 cores, and the higher performance of the new architecture?


Why? Because once it's available, the hexa 3.33GHz part is an absolutely huge performance bump beyond what is currently available in the MP lineup. If someone truly needs to be using a MP, then there's a pretty good chance that the upgrade from whatever box they have now, to a 6 or 12 core will pay for itself before the new architecture becomes available.

I upgraded from a quad 2.66 i7 to a hexa 3.33 i7 in my Linux/Win7 video encoding box and the performance increase is huge.

[Salavat23]

Sandy Bridge is imminent, in the near future, and will bring no less than FOUR entirely new standards and evolutions (USB3, SATAIII, PCI Express 3.0 and LightPipe), which will make all of today's computers obsolete, as these new technologies and ports will be required by more and more modern devices as time goes on.

Ok, that is just plain silly.

USB 3 and SATA 6 are already flooding the PC market. As for PCI-E 3.0, we aren't even close to saturating PCI-E 1.1 in many cases.

As for Lightpeak, I believe it is useless especially among the growing USB 3 and eSata markets. If you don't believe that, then you're still in luck. Read this:


"Intel has designed a prototype PCI Express card for desktop PCs as an add-on. This would mean many people wouldn't need to buy a new motherboard for the new cable type. The card has two optical buses powering 4 ports. Note that such a card might not be able to keep up with the 40Gbit/s bandwidth of four Light Peak ports. Most desktop motherboards in 2010 have one or more PCIe 16x slot and a few PCIe 1x slots and a few Standard PCI slots. A PCIe 1x slot is limited to 4Gbit/s; a PCIe 16x slot would be enough"

And there you have it. It would need but merely a PCI-E 1.1 16x slot.

What does Lightpeak even have to do with SB? You are linking technologies that are not interrelated at all, and then using them as a basis by which to convince people to to buy any computers in the near future.

[Asylum Design]

So, I've bought and evangelized Macs for years and years and years...

I was waiting for the new Mac Pros because the current hardware pricing for the performance wasn't worth it...

I was sooo wrong. I just didn't realize in which direction and just HOW much I was wrong.

It's sooooo much worse than I ever thought... wow... I must have been absolutely RETARDED to EVER buy ANY hardware from APPLE.

:confused::eek::eek::confused::eek::confused::rolleyes:

Good God Man, What Was I Thinking???????

I almost bought a Mac Pro Dual 2Ghz 8 Core / 16 Thread box... for $3500 roughly after shipping, tax, etc.

Instead?

I bought a Core i5 750 Motherboard / Combo, 4 Gig Ram, GTS 250 1GB from Fry's. Spent ~$400...

And I'm running OS X 10.6.3 on it... and I'm simply wiping the floor with that $3500 machine.

It's rock - hard stable, faster than a bat out of hell, overclocked to 4.2Ghz, and literally I can't get over how badass OS X runs with a Vanilla Kernel on this thing.

I'm getting WELL over 500+ in XBENCH, and with only 4 Cores, (only 2 are being identified by Cinebench) my Cinebench Scores are almost 6 for the CPU and near 40 for the GPU... with Geekbench, I'm at near 12000, not to mention I'm able to encode a 1 1/2 hour movie with Handbrake in High Definition (720p) in about 15 minutes flat. Ripping the movie from the disk takes roughly 3-4 minutes.

FOR $400!!!!!!!!!!!!!!!!

I can ONLY imagine how fast it would be if I bought something other than a POS Mid-Level CPU and GPU!!!!

So thank you Apple for making a great OS, but your hardware has GOT TO GO BABY! :-)

It will be a loooooooong time in Hades before I spend money on Apple hardware again.

Thoughts? Am I crazy?

http://www.pcpathology.com/images/xbench.jpg

[Icaras]

As for Lightpeak, I believe it is useless especially among the growing USB 3 and eSata markets. If you don't believe that, then you're still in luck. Read this:


Will the technology behind Lightpeak allow for compatibility between other digital interconnected interfaces? If so, then I still think Apple will take this route, just as it did with the mini Displayport, amidst an industry dominated by HDMI.

[c.hilding]

--> Post #56 (http://forums.macrumors.com/showpost.php?p=10101817&postcount=56) <--

Eidorian: See reply #56. USB3+SATAIII will be provided by Intel for OEMs as an all-in-one chip (http://forums.macrumors.com/showpost.php?p=10102199&postcount=67) that they can use on their Sandy Bridge motherboards. The thing you're talking about (2012) is the first REFERENCE BOARDS slated to have that AS A STANDARD INSTALLATION, but nothing prevents OEMs like Apple from putting Intel's all-in-one chip on their boards in the meantime.

brentsg: See reply #56, this is not about performance, it's about new standards. As for Westmere (6 cores) vs Nehalem (4 cores), the performance is negligible since most applications cannot even use more than 1 or 2 cores simultaneously. Come back when programs are written to use "n" number of cores without bottlenecks and I'll say your NEHALEM will come to use (let alone the WESTMERE). Heck, stuff like 3D Studio Max crashes if you have over 8 cores (or something like that, I forgot the exact number), and most applications still only use one or two cores.

Salavat23: See reply #56 regarding the USB3/SATAIII motherboards that are already "flooding the market" and why that does not matter to Macintosh users, not even to Hackintosh users. Hint: Intel's official chipsets are not out; those motherboards all use 3rd party non-Intel chipsets that completely lack drivers for Mac OS X. Read post #56 for further information.

Asylum Design: Good job mate.

Icaras: LightPeak is something you would have in addition to USB3 and SATA III; and is meant for connecting super high resolution screens, networking, storage, etc, basically anything that benefits from huge bandwidth. It offers a massive amount of bandwidth and will be excellent for a lot of purposes. It all travels over one, tiny cable as well.


I'm going to bed soon. I hope people will keep behaving. We've been having a good discussion so far!

[apolloa]

Sigh, I said EXACTLY this: http://www.macnn.com/articles/10/06/10/promises.better.future.ahead/

In the other thread, but some of you lot seem dead set on being as negative as you can. Yes I understand your frustration but at the end of the day Apple is making s*** loads of money of 'iToys' so can you blame them? I mean, they are a business after all.
But fret if you want, I would expect a very nice update soon however, and I think it will offer some unique features, I mean NO ONE predicted Apple would make there own version of Nvidia's Optimus for the MB Pro's.

And I would also like to add I for one am VERY great full for this latest iToy and totally love it, I cannot wait to get one. So my sympathy for you is a lot less then it was before Tuesday!! Sorry....


Thoughts? Am I crazy?

I love the name of your start up disk :D:D

[VirtualRain]

Intel's Tick-Tock cycle... Both parts of the cycle have merit.

You will always see some significant architectural improvements on the Tock swing of the cycle. However, those CPU's will be at a premium, compared to the following die-shrink Tick.

Thus buyers looking for the most value should buy now (Gulftown/Westmere) or the next Tick (22nm Ivy Bridge). The buyers who want to be on the bleeding edge (for a premium) should buy on the Tock.

As far as what I/O technologies will have the most impact...

I believe SATAIII is the most anticipated since SSD's are already pushing SATAII to the limit.

USB3 is nice to have, but the only peripherals which can exploit that kind of speed are fast drives, and there's eSATA for that job already. Never-the-less, it's a welcome improvement. LightPeak is a cable simplification play, not a performance driver, since everything it's touted to replace is not a bottleneck now anyway. PCIe 3.0 is also a nice evolution of the bus, but I don't anticipate any performance gains coming out of it for a few generations of graphics cards which are increasingly thermally limited, not bus constrained.

I think it's very possible that most people will be on a second computer after Sandy Bridge before any of these technologies (except SATAIII) are adding value.

[Eidorian]

Eidorian: See reply #56. USB3+SATAIII will be provided by Intel for OEM's as an all-in-one chip...Link please.

[c.hilding]

Link please.

http://www.guru3d.com/news/intel-not-planning-usb-30-until-2012/

Our sources here in Taipei tell us that Intel has no plans to integrate USB 3 into its chipsets until 2012 at the earliest - there are no new platforms due this year, and next year's roadmaps currently show none featuring USB 3. Apparently Intel plans to make a USB 3/SATA 6Gbps all-in-one chip for optional use by motherboard manufacturers on its products, however it’s currently finding it difficult to get the pin-count down to an appropriate size.

It appears that the quote comes from the April conference where Intel simultaneously announced Sandy Bridge's production date (Q4 2010). Note that "integrate into its chipsets" means building it right into their reference motherboards; they won't do THAT until 2012, but in the meantime they'll provide it to OEM motherboard manufacturers as an all-in-one USB3/SATAIII addon chipset.

I really need to go to bed now, hope there aren't more questions. ;)

[Icaras]

Icaras: LightPeak is something you would have in addition to USB3 and SATA III; and is meant for connecting super high resolution screens, networking, storage, etc, basically anything that benefits from huge bandwidth. It offers a massive amount of bandwidth and will be excellent for a lot of purposes. It all travels over one, tiny cable as well.


I'm well aware of Lightpeak, but my question was whether it will mimic how HDMI, DVI, DP, and mDP behave with one another in that you can easily adapt between different types of cables.

[c.hilding]

I'm well aware of Lightpeak, but my question was whether it will mimic how HDMI, DVI, DP, and mDP behave with one another in that you can easily adapt between different types of cables.

Ahh I suspected that you meant that from your wording. Nope, it's an entirely new connector that uses an optical signal rather than electricity.

It's possible that some manufacturer will release some box that connects to LightPeak and sends out a DVI/HDMI/DP/mDP signal, although that would be in the far-far-far future if the day comes when LightPeak is all there is on the computer and you need to connect some old monitor. ;)

The current display connectors are around to stay for a long time, and I can see Mini DisplayPort gaining momentum thanks to stuff like ATI EyeFinity's 6-output cards (made possible thanks to the small size of mDP). I expect we'll see more and more monitors that feature both Mini DisplayPort and DVI connectors and work with either type of graphics card, until the transition is complete. If I'm not mistaken, DisplayPort also costs a lot less for manufacturers to license (compared to DVI and HDMI), so there is incentive to adapt to it.

[c.hilding]

The initial post has once again been updated and includes everything that has been discussed thus far. I'm heading to bed for the day.

[Icaras]

Ahh I suspected that you meant that from your wording. Nope, it's an entirely new connector that uses an optical signal rather than electricity.

It's possible that some manufacturer will release some box that connects to LightPeak and sends out a DVI/HDMI/DP/mDP signal, although that would be in the far-far-far future if the day comes when LightPeak is all there is on the computer and you need to connect some old monitor. ;)

Hmm, but you can currently change an optical cable into a mini 1/8" digital cable just via a very simple small adapter on the connector itself. Why couldn't we expect that with Lightpeak? Going from lightpeak to USB, Firewire, DVI, HDMI, DP, mDP, etc. via a simple adapter?

[c.hilding]

Hmm, but you can currently change an optical cable into a mini 1/8" digital cable just via a very simple small adapter on the connector itself. Why couldn't we expect that with Lightpeak? Going from lightpeak to USB, Firewire, DVI, HDMI, DP, mDP, etc. via a simple adapter?

I'll answer this, but it's the last question for the day. ;) It seems as though you are thinking of typical digital S/PDIF audio connections in your example? Yes, most computers and audio interfaces offer both an electrical RCA plug AND a Toslink (http://en.wikipedia.org/wiki/TOSLINK) optical plug, allowing you to hook up digitally no matter which type of connector your audio receiver uses.

Well, that's all fine and dandy for something low-bandwidth like S/PDIF.

With LightPeak we are talking 10 Gbps for starters, and they are planning to scale it up to 100 Gbps. It's also mainly intended for high-bandwidth transfers like networking and super high resolution displays (with more pixels than the bandwidth of current display cables allows for). Current connectors for USB/FireWire/eSATA devices, audio or display connectors like DisplayPort/miniDisplayPort and DVI/HDMI are not going away, so this conversion you are talking about is a complete non-issue. Yes, eventually they plan to COMPLETELY replace ALL of our device connections with this single Optical wire, but that's in the far future. Initially its biggest use is likely to be networking.

So, why Optical?

Well, Intel chose an Optical transfer method due to the inherent problems with Electrical transfers; those being lower speeds, higher rates of errors and corruption, lower possible cable lengths, crosstalk between wires, and the fact that electrical transfers require DIFFERENT WIRE CONFIGURATIONS depending on what the protocol is. The last reason is why we have different cables for HDMI, DVI, DP, Audio, USB, FireWire, etc instead of a single cable for everything. That's because each electrical protocol needs its own collection of individual wires to transfer the data it needs.

With Optical, ALL of these problems go away and you can transfer as many protocols as you want at once over a single wire, since it's all sent digitally with an Optical connection. The fact that it is Optical is the reason that it is able to transfer so much data over a single tube. Electrical would never be able to do that and would require multiple wires to reach the same bandwidth. So if you wanted to make something called "ElectricalPeak" and transfer at 10 Gbps, you would have to design a cable with loads of wires in it, all designed for transferring data simultaneously (in parallel) over every wire in order to achieve the same 10 Gbps bandwidth that LightPeak gets over a single wire. And if you're going for LightPeak's upper limit of 100 Gbps with an electrical solution? Forget about it. ;) LightPeak's Optical solution does all of that over a single wire, it makes no sense to stick to electrical cables anymore.

So, for all the reasons above, particularly the fact that we will still have all the old connectors on our computers for the foreseeable future, there will not be converters from LightPeak into existing connectors like USB and FireWire. You would have to design a converter box that accepts the LightPeak signal, decodes it, grabs the packets relating to its transfer, sends it to some converter chip that outputs the relevant electrical signal (such as USB or FireWire), and then go from there. It would make very little financial sense to create such a box since our computers will STILL have all the old connectors for a very long time. LightPeak devices require advanced circuitry to decode and pass through the light signals and therefore will initially be reserved for niche uses such as Networking and ultra resolution displays.

[Eidorian]

http://www.guru3d.com/news/intel-not-planning-usb-30-until-2012/



It appears that the quote comes from the April conference where Intel simultaneously announced Sandy Bridge's production date (Q4 2010). Note that "integrate into its chipsets" means building it right into their reference motherboards; they won't do THAT until 2012, but in the meantime they'll provide it to OEM motherboard manufacturers as an all-in-one USB3/SATAIII addon chipset.

I really need to go to bed now, hope there aren't more questions. ;)You've only reinforced my point.

Intel has no USB 3.0 offerings until 2012 on the hardware it provides. This leaves many other manufacturers open to provide solutions.

[Icaras]

I'll answer this, but it's the last question for the day. ;) It seems as though you are thinking of typical digital S/PDIF audio connections in your example? Yes, most computers and audio interfaces offer both an electrical RCA plug AND a Toslink (http://en.wikipedia.org/wiki/TOSLINK) optical plug, allowing you to hook up digitally no matter which type of connector your audio receiver uses.

Well, that's all fine and dandy for something low-bandwidth like S/PDIF.

With LightPeak we are talking 10 Gbps for starters, and they are planning to scale it up to 100 Gbps. It's also mainly intended for high-bandwidth transfers like networking and super high resolution displays (with more pixels than the bandwidth of current display cables allows for). Current connectors for USB/FireWire/eSATA devices, audio or display connectors like DisplayPort/miniDisplayPort and DVI/HDMI are not going away, so this is a complete non-issue. Yes, eventually they plan to COMPLETELY replace ALL of our device connections with this single Optical wire, but that's in the far future. Initially its biggest use is likely to be networking.

So, why Optical?

Well, Intel chose an Optical transfer method due to the inherent problems with Electrical transfers; those being lower speeds, higher rates of errors and corruption, lower possible cable lengths, crosstalk between wires, and the fact that electrical transfers require DIFFERENT WIRE CONFIGURATIONS depending on what the protocol is. The last reason is why we have different cables for HDMI, DVI, DP, Audio, USB, FireWire, etc instead of a single cable for everything. That's because each electrical protocol needs its own collection of individual wires to transfer the data it needs.

With Optical, ALL of these problems go away and you can transfer as many protocols as you want at once over a single wire, since it's all sent digitally with an Optical connection. The fact that it is Optical is the reason that it is able to transfer so much data over a single tube. Electrical would never be able to do that and would require multiple wires to reach the same bandwidth. So if you wanted to make something called "ElectricalPeak" and transfer at 10 Gbps, you would have to design a cable with loads of wires in it, all designed for transferring data simultaneously (in parallel) over every wire in order to achieve the same 10 Gbps bandwidth that LightPeak gets over a single wire. And if you're going for LightPeak's upper limit of 100 Gbps with an electrical solution? Forget about it. ;) LightPeak's Optical solution does all of that over a single wire, it makes no sense to stick to electrical cables anymore.

So, for all the reasons above, particularly the fact that we will still have all the old connectors on our computers for the foreseeable future, there will not be converters for existing types of connections like USB and FireWire. They are not competing, yet. Mainly because LightPeak devices require advanced circuitry to decode and pass through the light signals and therefore will initially be reserved for niche uses such as Networking and ultra resolution displays

Thanks for the thorough explanation. I wasn't well versed into the fine details of Lightpeak so I figured the same practices today with adapting different cable protocols would just easily as work in the future with lightpeak. But I guess that isn't the case.

[Apollo21]

Yes, eventually they plan to COMPLETELY replace ALL of our device connections with this single Optical wire, but that's in the far future. Initially its biggest use is likely to be networking.

I thought a big point of LightPeak was that it would be used on laptops and replace all the connections you currently need to hook up your display/audio/power/USB peripherals, presumably by keeping a little converter box at your desk that everything would connect to with its usual connector. Is this not expected in the near future at all? (I was really looking forward to that the next time I buy a laptop.)

[c.hilding]

Icaras: No problem. :)

You've only reinforced my point.

Intel has no USB 3.0 offerings until 2012 on the hardware it provides. This leaves many other manufacturers open to provide their own USB 3.0 solutions.

Hey, there's no need for smugness like "you've only reinforced my point," especially not when you have completely forgotten why you were arguing with me in the first place. ;) Let's take a stroll through memory lane, through the far-far back time of "about an hour ago":

Your initial argument and claim was that "Intel isn't making *any* USB 3 chips (controllers) until 2012", for example this post:

Intel's USB 3.0 implementation is still scheduled for 2012 on their controller chips.

I replied to that post of yours by mentioning that Intel indeed plans to make an official Intel all-in-one chip in 2011 in conjunction with the Sandy Bridge launch, that's free for OEMs to use for both USB 3 and SATA III, a fact which you immediately questioned with a quick reply of "Link please":

Eidorian: See reply #56. USB3+SATAIII will be provided by Intel for OEMs as an all-in-one chip that they can use on their Sandy Bridge motherboards.

Link please.

So I dutifully replied with a link:

"Intel plans to make a USB 3/SATA 6Gbps all-in-one controller for optional use by motherboard manufacturers on its products"
From: http://www.guru3d.com/news/intel-not-planning-usb-30-until-2012/

Upon which you replied smugly in a "told you so!" manner, completely forgetting what your original argument had been!

So, let's recap: Your argument was "Intel won't have any USB 3 chips before 2012", and when I showed you that Intel will indeed create a USB 3 / SATA III chip, your argument is now suddenly "Yes, Intel will have USB 3 but so will many other manufacturers, you are only reinforcing my point". I guess you're a bit tired since you completely forgot what your initial argument was and changed it to something that has nothing to do with the original one, and then gloats about how "right" you are. Interesting argument technique. :rolleyes:

I fully expect some sort of silly retaliation from you to save face, but I'm going to bed, who cares about internet arguments?

[Eidorian]

Where does it say USB 3.0 from Intel in 2011 for Sandy Bridge?

[c.hilding]

I thought a big point of LightPeak was that it would be used on laptops and replace all the connections you currently need to hook up your display/audio/power/USB peripherals, presumably by keeping a little converter box at your desk that everything would connect to with its usual connector. Is this not expected in the near future at all? (I was really looking forward to that the next time I buy a laptop.)

The long-term goal is to have most devices running off LightPeak so that you can indeed connect everything with a single wire. I don't think we'll see too many devices in the short term that support light peak, simply because it costs money to implement LightPeak in a device, and computers already have all the connectors you need. Over time we'll see more and more LightPeak devices, but initially it'll primary be useful for things like networking between two computers via LightPeak, allowing 10 Gbps transfers, which is actually something most people won't even need, but would be useful for example if you were to connect two computers together and write some software so that they help each other compute large data sets.

Seriously, this thread is keeping me up with all the questions. Bed. Now. :D

Where does it say USB 3.0 from Intel in 2011 for Sandy Bridge?

See, I knew you'd change the argument a third time, you did not fail to disappoint. :D So it has morphed from "Intel won't make any USB 3 chips until 2012" to "haha, I told you so, Intel is going to be making USB 3 chips in 2011", to "show me where that chip (that I, just a moment ago, dismissed as not even existing) will sit on the reference boards for Sandy Bridge". This is pointless. But since you asked I'll give you one last dog bone: Read the article, or at least the quote, it mentions that Intel will provide the all-in-one USB 3 / SATA III chip in conjunction with the Sandy Bridge launch, so that OEMs will be able to easily implement USB 3 and SATA III on the new Sandy Bridge motherboards.

Good night all!

[Eidorian]

See, I knew you'd change the argument a third time, you did not fail to disappoint. :D So it has morphed from "Intel won't make any USB 3 chips until 2012" to "haha, I told you so, Intel is going to be making USB 3 chips in 2011", to "show me where that chip (that I, just a moment ago, dismissed as not even existing) will sit on the reference boards for Sandy Bridge". This is pointless. But since you asked I'll give you one last dog bone: Read the article, or at least the quote, it mentions that Intel will provide the all-in-one USB 3 / SATA III chip in conjunction with the Sandy Bridge launch, so that OEMs will be able to easily implement USB 3 and SATA III on the new Sandy Bridge motherboards.Do you have another article?

Computex already showed Intel onboard SATA 6 Gbps on H67 and P67. There's still no final word on X68 (I guess.) but I expect it to include SATA 6 Gbps as well.

From your April Guru 3D article, Intel won't be adding USB 3.0 to its offerings. When it does so in 2012, it'll still be optional.

[deconstruct60]

F
Dual and Quad-Core low-end models come first, before 6-core and 8-core high-end models in Q2 2011. It still begs the question though: Why get Westmere now? Remember that Westmere is just a "tick" 32nm die-shrink of Nehalem with two more cores in (for a total of 6) in the highest-end CPUs. Why not wait for the "tock" (new architecture) in Q2 2011


1. Sandy Bridge is also 32nm, the exact same manufacturing process as Westmere. The "tock" (in terms of process) is when do not improve process and do optimization tweaks along with microarchitecture improvements. Sandy Bridge will be slightly better, but there is no general huge jump coming on Xeon.

So 8-cores will only mean the die gets bigger and the costs go up. Notice that the new Westmere versions all dropped at higher costs this spring? The "old" tech didn't go down in price. That is all the more likely to happen again next Summer/Fall2011.

What going to pick up is another memory controller. So if fill four, single DIMM slots all four DIMMs accessible in parallel. ( currently on the 7500/6500 series ). Also get a bump in L3 caches. Also get a bump in intercore communication. On the low end the core count will still be 4!!! With 4 , 6 , and 8 Apple is likely to even release the same 4 core , 6 core , 6 core line up as they likely will with Westmere. That would keep the 3 subversions of the SP and DP models to still have 6 total offerings. Don't really see them going to more than 6 ( maybe a Config-to-order with 8 core option, but it won't be a focus nor particularly affordable for the vast majority. )

The majority of folks here can only afford the low and middle versions so won't be that huge of a change. It is extremely misguided to set any sort of expectation that 8 core package set-ups will be common place in the Sandy Bridge era.


This whole argument of "old, ancient, decrepit" tech at 12 months is ridiculous; especially in the high reliability space. Intel's pricing says that isn't true (or else prices would have collapsed on the decrepit Xeons from last year. ). Intel knows nothing about the market for their chips... yeah right. Intel disagrees with that. So do lots of other customers who are seeking regularity as oppose as chasing specs.

Also since Intel had trouble ramping up on Core-i then Xeon they are more likely to space out the Xeon update. They also will be rolling out the Nehalem-EX later this year... so there is another Xeon roll out to go this year. That makes even more stuff in the queue before a Xeon 5000/3000 series refresh. Won't be surprising if it rolls out later next year so that release time more closely matches when they can ship it in volume.


Sandy Bridge is far more important for Intel's mobile line up because the current mobile Core-i packages have both a 32nm and 45nm die in them. If can collapse that to one 32nm die or at worse two 32nm dies that would be a huge bump (to IGP. also to memory access if can get to one die) . The desktop versions mostly get a L3 cache bump. Don't think that is going to be revolutionary changes in performance (since for the mainstream ones still have the memory controllers throttled at 2. ).




and get USB3, LightPeak, SATA3, PCI Express 3.0 (twice the bandwidth),


Mabye ( US3 PCI 3.0)although Apple is likely to do a substitution if something new appears ( remove 'old' or the 'new' so that board space is approximately the same.). Lightpeak is a separate break out chip and hasn't even come close to getting standardization consensus. You are not likely giong to get numerous new break out chips on the board. . Like the CPUs, the server support chispsets tend to lag behind the consumer ones.




native support for 1600 MHz RAM, 8 cores, and the higher performance of the new architecture?

What is bleeding edge 1600 ECC RAM going to do with pricing ? Better to wait and see what Apple does with Westmere offering before setting expectations that they are going to switch modes and go with the highest possible RAM option.


This is the classic if you wait 6 months there will always be something better on the horizon trap. Ivy Bridge is going to be faster/better than Sandy Bridge... so maybe should wait 2 years for a new box. Haswell is going to be better than Ivy bridge... so how about waiting 3 years. etc. etc. etc. etc.
Eventually fall into same bucket as the folk who post about how their old workstation is going to be desupported next week and it is the ultimate doom because don't have anything to move to from their 5-7 year old machine. Relative to the 5-7 year machine the current offering is not "old". What is ancient is what you're riding on now.

[Digital Skunk]

See, I knew you'd change the argument a third time, you did not fail to disappoint. :D So it has morphed from "Intel won't make any USB 3 chips until 2012" to "haha, I told you so, Intel is going to be making USB 3 chips in 2011", to "show me where that chip (that I, just a moment ago, dismissed as not even existing) will sit on the reference boards for Sandy Bridge". This is pointless. But since you asked I'll give you one last dog bone: Read the article, or at least the quote, it mentions that Intel will provide the all-in-one USB 3 / SATA III chip in conjunction with the Sandy Bridge launch, so that OEMs will be able to easily implement USB 3 and SATA III on the new Sandy Bridge motherboards.

Good night all!

I read all of the threads and Eidorian is right, and your articles only support what he's said.

I don't know what all the fanfare is for.

[iMacmatician]

The majority of folks here can only afford the low and middle versions so won't be that huge of a change. It is extremely misguided to set any sort of expectation that 8 core package set-ups will be common place in the Sandy Bridge era. I do not expect 8-core Sandy Bridge to be more widespread than 6-core is with Westmere (likely less).

[Eidorian]

I do not expect 8-core Sandy Bridge to be more widespread than 6-core is with Westmere (likely less).I'm expecting to see 6 cores make it into the $284 price point under Sandy Bridge.

The Core i7 970 ($561) doesn't appear to have a counterpart on the Xeon side that I know of. For a single socket system you're still looking at ~$1,000 3.33 GHz Gulftown 6-core Xeon.

8-cores is going to be the premium like 6-cores is on Xeon. As mentioned before the die size is only going to make matters worse when it comes to pricing. It's all still 32nm. Not that the entities purchasing an 8-core processor will care all that much.

On the consumer side the next Extreme processor is going to once again be at ~$1,000 like its many predecessors. It'll probably be 8 cores too.

http://att.macrumors.com/attachment.php?attachmentid=228308&d=1275313400

[rhett7660]

If they would have released Mac Pros, it would have been during the keynote or on the Tuesday the day after, that's how they usually do it.

PS: The new release of Xcode is simply that, a new version of Xcode, with the Interface Builder integrated into the workflow, that's it, there's no hidden meaning. ;)

Um. wasn't the 08 MP released right before Mac world 08 with very little if any mention at Mac World?

[Umbongo]

Um. wasn't the 08 MP released right before Mac world 08 with very little if any mention at Mac World?

Yup. Only the 06 model was released at a keynote. 07 was a week before NAB and 09 was press release only.

[DoFoT9]

nice read :)

so ~mid/late 2011 for the 8 core CPUs, thats a nice advancement - roughly a year away. great, ill start saving for my MP now.

the main question (i think) is what sort of improvement % wise are we talking? if it is just going to be the usual 20%-30% then i wont be that excited. i guess its too early to speculate :(

[Hellhammer]

I'm expecting to see 6 cores make it into the $284 price point under Sandy Bridge.

That would be logical and could fight against AMD's hexa-cores (in price)

The Core i7 970 ($561) doesn't appear to have a counterpart on the Xeon side that I know of. For a single socket system you're still looking at ~$1,000 3.33 GHz Gulftown 6-core Xeon.

It isn't even out yet, right? I'm sure Intel will deliver at least deliver UP version (same chip with ECC support, Xeon W3xxx) of it for same price, just like they did with Core i7 X980 (Xeon W3680). No need for DP version really as the price difference between X5670 and X5680 is only 220$ plus Intel would be running out of numbers (X5675?) :D Not really even need for it

On the consumer side the next Extreme processor is going to once again be at ~$1,000 like its many predecessors. It'll probably be 8 cores too.

"Consumer" :p But yeah, at least what is currently known, there will be 8-core "consumer" chip, i.e. the successor of i7 X980, likely same chip that is used in UP Xeons

[Maurier]

I honestly don't understand why people are arguing over this.
I think a lot of people are trying to justify why they bought a 2009 Mac pro:D
I just bought one 2 days ago and for my needs ( video editing ,watching movies and playing wow from time to time) i'm more then happy and will not need to upgrade until late 2011 or 2012.
Who cares? If you really need those 6/12 cores then buy one at the end of the month if not buy a 2009 now and be done with it.
This is getting silly , and even a little nerdy :p

Either 90% of the people here are all high end professional doing amazing things with their mac's and need the latest and greatest or just a lot of tech geeks.
Seriously look what Steve Job has done to you guys lol

I just finished watching WWDC 2010 and i felt really nerdy after haha , all those geeks applouding at a phone WTF?
What's this world coming to? Some people need a reality check and this is coming from a guy who plays rpg's 5 hours a day.

[surferfromuk]

I honestly don't understand why people are arguing over this.
I think a lot of people are trying to justify why they bought a 2009 Mac pro:D
I just bought one 2 days ago and for my needs ( video editing ,watching movies and playing wow from time to time) i'm more then happy and will not need to upgrade until late 2011 or 2012.
Who cares? If you really need those 6/12 cores then buy one at the end of the month if not buy a 2009 now and be done with it.
This is getting silly , and even a little nerdy :p

Either 90% of the people here are all high end professional doing amazing things with their mac's and need the latest and greatest or just a lot of tech geeks.
Seriously look what Steve Job has done to you guys lol

I just finished watching WWDC 2010 and i felt really nerdy after haha , all those geeks applouding at a phone WTF?
What's this world coming to? Some people need a reality check and this is coming from a guy who plays rpg's 5 hours a day.

Interesting. I completely agree! However, I doubt its high end professionals as they just get on with buying for paying jobs, and if they need extreme power they have the money to buy more Mac Pros ( regardless of cost) or put jobs out onto to a cheap PC render farm (Hackintosh unlikely!).

There is, however, definitely some kind of effect here (team effort I'd say!) that is in some ways attempting to erode the Mac Pro's professional industry userbase by prematurely creating a notion of an second rate and/or overdue product where in reality it's an amazing piece of kit and the update is still not technically due till September. Sure it's expensive - top end Apple kit is - always has been, but equally never been cheaper!

Arguably, some of these efforts could even (legally?) be construed as a form of deliberate industrial and commercial sabotage! It's certainly gone beyond rumor and way past idle speculation - some people are almost aggressively vitriolic against potential buyers.

In fact I find this current thread utterly laughable - the OP is actually now proposing people avoid even the NEXT as yet unreleased Mac Pro in favor of some entirely mythic future 2012 Mac Pro...Quite bizarre

[Eidorian]

It isn't even out yet, right? I'm sure Intel will deliver at least deliver UP version (same chip with ECC support, Xeon W3xxx) of it for same price, just like they did with Core i7 X980 (Xeon W3680). No need for DP version really as the price difference between X5670 and X5680 is only 220$ plus Intel would be running out of numbers (X5675?) :D Not really even need for itIt's scheduled for later this year. The Core i7 990X (http://www.techpowerup.com/124334/Intel_to_Release_Core_i7_990X_Extreme_Edition_Six-Core_Processor_in_Q4_2010.html) appears to be possible as well. It will be interesting to see it with a higher BCLK than the current 133 MHz.


"Consumer" :p But yeah, at least what is currently known, there will be 8-core "consumer" chip, i.e. the successor of i7 X980, likely same chip that is used in UP Xeons"Enthusiast" would have been a better choice. It's still in the single socket market without ECC.

[johnnymg]

The reason for this years focus on iOS should be obvious.

Apple has three lines of business that are all approx. $10B business units:
- Mac
- iPod
- iPhone

Their Mac market is growing at a tremendous rate, almost all by itself (see my post on why iDevices are good for the Mac).

The iPod market is somewhat stagnant, and to some extent being cannibalized by Apple's own more sophisticated iOS devices. It's a market in decline.

The iPhone market is under assault by Google and it's critical that Apple focus on competing strongly here.

Finally, Apple is trying to develop a new line of business around tablets that they hope will be a fourth $10B business, and it too is already in the sights of many competitors. If they want to achieve $10B like revenues in this area, it's going to require a massive effort.

To have done anything other than focus this years WWDC on catering to iOS developers would have been a bad decision for their two business units which are most vulnerable.

You guys need to think with your Apple business hat on. Presumably, all of the new customers Apple is bringing on board, is great news for the pros who frequent this forum... it means more customers to consume whatever you are developing.

EXCELLENT comments!

In a perfect world (1) your comments would be a sticky and (2) macrumors would limit new members to only read status for a month.

cheers
JohnG

[iMacmatician]

Presumably, all of the new customers Apple is bringing on board, is great news for the pros who frequent this forum... it means more customers to consume whatever you are developing.Never thought of it that way…very good point.

[VirtualRain]

EXCELLENT comments!

In a perfect world (1) your comments would be a sticky and (2) macrumors would limit new members to only read status for a month.

cheers
JohnG

LOL... thanks. We can dare to dream! :p

[xgman]

Engineering Sample in case anyone is interested:

http://forum.coolaler.com/showthread.php?t=240578

[jjahshik32]

Its a shame that apple is behind the desktop compared to PCs. I bet most people would just pick up a mac mini with an nvidia 320m to hold them over until the new mac pro is released.. but then again the mac mini is taking forever to update as well!

[bearcatrp]

After reading these posts, which bring good information, and steve jobs email, something tells me that just maybe steve has something up his sleeve to blow us away. He still has a good relation with intel. Why do an upgrade when sandy bridge is so close to being released. Apple has been able to produce products before anyone else was able to get there hands on it. So it could be feasible for apple to get sandy bridge early and hold off just a little longer to come out with a new Mac pro with sandy bridge. I just hate to see what the cost will be though. My 2008 Mac pro is doing just fine. I can wait for sandy bridge. My guess is we will see it in the fall.

[Icaras]

I can wait for sandy bridge. My guess is we will see it in the fall.

You're joking, right?

[the editor]

wow...people are talking about when apple is going to probably release MacPro's with sandy bridge processors....2010 is almost over and Apple is stil claiming the 2009 "pro" :rolleyes: machine to be the "newest and latest technology"...in the mean time, people are still waiting for apple to release an update that is long overdue and here we are talking about waiting for a release in Q2 of 2011...this is getting a bit...silly

[Umbongo]

wow...people are talking about when apple is going to probably release MacPro's with sandy bridge processors....2010 is almost over and Apple is stil claiming the 2009 "pro" :rolleyes: machine to be the "newest and latest technology"...in the mean time, people are still waiting for apple to release an update that is long overdue and here we are talking about waiting for a release in Q2 of 2011...this is getting a bit...silly

Almost over? We aren't even half way through the year.

[Digital Skunk]

Its a shame that apple is behind the desktop compared to PCs. I bet most people would just pick up a mac mini with an nvidia 320m to hold them over until the new mac pro is released.. but then again the mac mini is taking forever to update as well!

Agreed, and I am tempted to pull the trigger on the base model one just so I can have an extra machine to work on while I render, rip, encode and upload.

I don't even mind that they haven't been updated yet, the big problem for me is that one would cost the same price as a used 1,1 Mac Pro.

wow...people are talking about when apple is going to probably release MacPro's with sandy bridge processors....2010 is almost over and Apple is stil claiming the 2009 "pro" :rolleyes: machine to be the "newest and latest technology"...in the mean time, people are still waiting for apple to release an update that is long overdue and here we are talking about waiting for a release in Q2 of 2011...this is getting a bit...silly

What's silly is that this thread is all about the OP suggesting that Apple may just wait for SB, and skip Westmere altogether, which is just plain absurd.

[gglockner]

I think the original point is that the only significant difference between Gainestown and Westmere is whether it has 4 or 6 cores on die. Both are Nehalem architecture and should give comparable performance at the same clock speed.

The bigger difference will come when Sandy Bridge is released, which is a new architecture.

Bottom line is that if you're on the fence about buying a Mac Pro today, buy it if you need it. A 2009 Mac Pro is a huge upgrade from a G5 or 1st generation Intel Mac Pro. If you've got a 2008 or 2009 Mac Pro, then take a deep breath and wait until Sandy Bridge is released. If you've got money burning a hole in your pocket, buy some RAM and an SSD. In most cases, those will make a bigger difference in performance.

Finally, if you absolutely must be on the bleeding edge of technology, Apple is not for you. There are plenty of other choices.

[Icaras]

Its a shame that apple is behind the desktop compared to PCs. I bet most people would just pick up a mac mini with an nvidia 320m to hold them over until the new mac pro is released.. but then again the mac mini is taking forever to update as well!

Agreed, and I am tempted to pull the trigger on the base model one just so I can have an extra machine to work on while I render, rip, encode and upload.

I don't even mind that they haven't been updated yet, the big problem for me is that one would cost the same price as a used 1,1 Mac Pro.

I see I am not the only one that's thought about this route. Honestly, I thought this might have been a silly solution to begin with but I see some people think it's practical too. If the Mac Pro weren't so "imminent", I probably would pull the trigger too now.

[Digital Skunk]

...

Agreed

I see I am not the only one that's thought about this route. Honestly, I thought this might have been a silly solution to begin with but I see some people think it's practical too. If the Mac Pro weren't so "imminent", I probably would pull the trigger too now.

Oh yeah bro, I've been toying with the idea for a few weeks now. May just toss in 8GB of RAM and a 640GB HDD just for kicks. The sweetest part is that it'll run two 23" ACDs, so when the Mac Pro finally comes, I just clone, connect, and have 22x the speed.

[Icaras]

Oh yeah bro, I've been toying with the idea for a few weeks now. May just toss in 8GB of RAM and a 640GB HDD just for kicks. The sweetest part is that it'll run two 23" ACDs, so when the Mac Pro finally comes, I just clone, connect, and have 22x the speed.


Ah, very tempting! Well I suppose I'll wait until June passes and see if I may want to go that way too.

I'd say that config you mentioned is a $1000+ rig already?

[sandylp]

Thanks for starting this thread. I just sold my Mac Pro with 2x 2.8 Quad processors in order to get as much money as I could before it became too outdated. I had planned on buying a new MP, but now I think I'll just use my laptop (less than a year old) with a new 1TB drive until Sandy Bridge comes out. It seems as though I've always been on Intel's Tick schedule instead of the Tock. My goal now is to get on the Tock schedule. :)

[Digital Skunk]

Ah, very tempting! Well I suppose I'll wait until June passes and see if I may want to go that way too.

I'd say that config you mentioned is a $1000+ rig already?

Yeah, I'd buy it with the student/teacher discount and it'll be about $550 for the Mini, and another $250 for the RAM, then $150 for the drive. Round-a-bouts, I don't have the numbers in front of me.

The hard part is that I've seen many 1,1 Mac Pro 2.66 Quads stock going for around $1000. So I have to see if I want to break tradition and get a brand new Mac Pro, go stick with what's save me thousands and get a used machine.

[sandylp]

I think the original point is that the only significant difference between Gainestown and Westmere is whether it has 4 or 6 cores on die. Both are Nehalem architecture and should give comparable performance at the same clock speed.

The bigger difference will come when Sandy Bridge is released, which is a new architecture.

Bottom line is that if you're on the fence about buying a Mac Pro today, buy it if you need it. A 2009 Mac Pro is a huge upgrade from a G5 or 1st generation Intel Mac Pro. If you've got a 2008 or 2009 Mac Pro, then take a deep breath and wait until Sandy Bridge is released. If you've got money burning a hole in your pocket, buy some RAM and an SSD. In most cases, those will make a bigger difference in performance.

Finally, if you absolutely must be on the bleeding edge of technology, Apple is not for you. There are plenty of other choices.


I found this on the internet, which only confirms the rationale for waiting for Sandy Bridge:

The current dual-core "Westmere" processors aren't exactly ideal: the cores and L3 cache are on a 32-nm die, but the memory controller, GPU, and PCI-E controller are on a separate 45-nm die. This design does hurt memory bandwidth and latency. "Sandy Bridge" will bring all of these components (CPU cores, L3 cache, memory controller, GPU, and PCI-E controller) unto a single monolithic 32-nm die, which by itself should improve performance across the board.

So, if you can hold out until Sandy Bridge, you'll be rewarded with a major upgrade over whatever you're using now. :)

[Eidorian]

I found this on the internet, which only confirms the rationale for waiting for Sandy Bridge:

The current dual-core "Westmere" processors aren't exactly ideal: the cores and L3 cache are on a 32-nm die, but the memory controller, GPU, and PCI-E controller are on a separate 45-nm die. This design does hurt memory bandwidth and latency. "Sandy Bridge" will bring all of these components (CPU cores, L3 cache, memory controller, GPU, and PCI-E controller) unto a single monolithic 32-nm die, which by itself should improve performance across the board.

So, if you can hold out until Sandy Bridge, you'll be rewarded with a major upgrade over whatever you're using now. :)That only really matters on Clarkdale and Arrandale.

[deconstruct60]

I found this on the internet, which only confirms the rationale for waiting for Sandy Bridge:

The current dual-core "Westmere" processors aren't exactly ideal: the cores and L3 cache are on a 32-nm die, but the memory controller, GPU, and PCI-E controller are on a separate 45-nm die.


What does this have to do with a Mac Pro? If you haven't noticed the Mac Pro's have 4 cores and the memory controller on the same die. This issue is really only talking about the "Westmere IGP" processors where Intel nuked the on die memory controller to move it over to the IGP die. Those integrated, CPU + GPU , chip packages are targeted toward laptops and mainstream desktop machines; not workstation (i.e.., Mac Pro).

Even the top end iMac don't have this problem. (they are short one less memory controller but it is all on die ).


The big leap that Sandy Bridge does is provide a more plausibly useful CPU + GPU package. The Sandy Bridge IGP will likely be much closer to what the 9400m is/was. That will make it more realistic to offer economy systems that don't have discrete chips to pull up the slack. I'm not sure Apple is going to use any of those. Well maybe a MBA, if it is still around.


The other Sandy Bridge feature that is more oriented toward a workstation better vector engine (AVX). There is also a better bus between the cores on the die.



So, if you can hold out until Sandy Bridge, you'll be rewarded with a major upgrade over whatever you're using now. :)

The "Sandy Bridge" like Xeon offering (5700/3700 ) is not coming till next April-June. It is likely going to slide back from March because:

1. still haven't finished getting the 3600 series out.
2. Now also has an -EX line to launch and ramp up. (more stuff )

If you waiting for relatively short term it is for Intel to finish rolling out the 3600 series.

[DoFoT9]

The big leap that Sandy Bridge does is provide a more plausibly useful CPU + GPU package. The Sandy Bridge IGP will likely be much closer to what the 9400m is/was.
annoying that its already going to be outdated and it hasnt been released yet :(

[nanofrog]

annoying that its already going to be outdated and it hasnt been released yet :(
As he mentioned, the IGP models are aimed at low cost systems. You still want a separate graphics controller for improved performance, and there's models within Sandy Bridge aimed at this market as well (no IGP). But not every part in that family will release at the same time.

[DoFoT9]

As he mentioned, the IGP models are aimed at low cost systems. You still want a separate graphics controller for improved performance, and there's models within Sandy Bridge aimed at this market as well (no IGP). But not every part in that family will release at the same time.
true i guess. its just a CPU, only needs to display images not play games :P

[Ravich]

I would be more than happy to wait if there were something to hold me over in the meantime.


The closest I can get is a dual HDD mac mini.

[deconstruct60]

annoying that its already going to be outdated and it hasnt been released yet :(

It isn't outdated. Most likely it will be equal in performance and pull less power. If graphics costs separated out, it will also cost substantially less than a 9400m's graphics. Performance isn't the only criteria. For 100's of millions of folks who just want any home computer that will be sufficient performance. Not everyone has $700+ to blow on a computer.

Apple thinks you can't make a less than $500 computer that isn't crap. With the more mainstream priced versions of Sandy Bridge you will be more able to do so.

Nor even if you do have a higher budget many folks do not need to run the discrete GPU all the time. You can chop power when not doing high end graphics.

[nanofrog]

Apple thinks you can't make a less than $500 computer that isn't crap. With the more mainstream priced versions of Sandy Bridge you will be more able to do so.
Exactly. Budget systems are a substantial market that Intel's not willing to ignore and leave solely to AMD. System vendors may not make that much on the individual system, but the sales volume is high enough they still turn a decent profit. Then when you consider Intel sells to all of these vendors, they make out quite well. ;)

[DoFoT9]

It isn't outdated. Most likely it will be equal in performance and pull less power. If graphics costs separated out, it will also cost substantially less than a 9400m's graphics. Performance isn't the only criteria. For 100's of millions of folks who just want any home computer that will be sufficient performance. Not everyone has $700+ to blow on a computer.

Apple thinks you can't make a less than $500 computer that isn't crap. With the more mainstream priced versions of Sandy Bridge you will be more able to do so.

Nor even if you do have a higher budget many folks do not need to run the discrete GPU all the time. You can chop power when not doing high end graphics.

outdated may have been the word - but i totally see your point. for end users with little graphics requirements for sure, they just need something to show those little pixels! :D

but aren't we more referring to MacPro users whose requirements are more demanding then that of a normal end user? or am i missing something?