Intel Launches Haswell Processors Ahead of WWDC Mac Updates

[flux73]

I wasn't really expecting a huge jump in performance but the GPU performance was something I was waiting to hear about, that sucks due to the hype that was built around it. Now I'm not even sure on battery life just due to how much power the display takes from the rMBP. I mean I'm sure will see an improvement but not the huge jump everyone was expecting. They need to improve battery tech or find a screen that has better power consumption.

Well, I think the big thing is supposed to be better power management. From what I've read, computers are idle for the vast majority of the time and there are a lot of wasted CPU cycles during times like for example, just simply writing this post. It takes an extraordinarily small amount of the CPU's processing power to allow you to type and most of the CPU's processing is wasted (thus wasting energy) during these moments.

Ultimately, the real world test is what is going to matter. If a Haswell Macbook Air or Pro can last 8-9 hours with real life usage, I think it would be considered a success. 10 hours would be astounding and I would be snapping up some Intel stock.

 

[Oletros]

Yes, so you agree that ARM designs could ramp up the performance to equal or surpass Intel. Now factor in a couple of other facts:

And equaling or surpassing Intel power comsumption defeating the ARM purposes

 

[milo]

An E3 based "Mac Pro" likely would not be simply called "Mac Pro", if "Mac Pro" at all. It would likely be significantly smaller and more limited box.

I'm merely pointing out the new chip that's the new equivalent to the one they're using in the base MP right now. The current quad has limitations compared to the dual socket version, and assuming they do ship a single socket MP again, I'd expect it again would be more limited than the dual.

Maybe they would do two different boxes, but there's a good argument that they should have been doing that all along...and they haven't been. While there are some good reasons to do two different cases, I doubt it will happen.

Are the current Mac Pros slouches in some way? I know they haven't been updated in a long time, but I'd figured the 6-core Xeon chips in them would still be blazing fast. No?

With the 6 and 12 cores, CPU isn't so bad, but they are slouches in other respects, mainly IO like SATA, TB, and USB. With the quads, they are outperformed by the high end mac mini.

 

[deconstruct60]

So you don't believe Apple is testing OS X on ARM in their labs in anticipation for future ARM chips that might be powerful enough to meet their needs?

Probably not. All that testing would cover is the true operating system portion of OS X and the minimal set of build chain tools. Well, 70-90% of that is already shared with iOS. The core of the OS is already ported to ARM. Cocoa versus Cocoa Touch testing isn't going to show a whole lot.

It makes almost no sense at all to sink a ton of testing in to a minor variant of the OS on the same architecture you already have to build for production anyway.

When Apple was building x86 variants while running OS X in production on PPC Apple had NO operating systems on x86. None, zip, nada. Same thing during the 68000 to PPC transition.

Throw on top of that that ARM has so far demonstrated no ability to match Intel in the performance spectrum and it makes zero sense to waste alot of time on this. Intel is only now finished turning its full competitive attention to competing with ARM.

It might not happen anytime soon, but if you don't think they're considering all the possibilities several years down the road, you're the one who's crazy:

Far more likely would is that OS X would just end. iOS would take over as being the sole survivor. It is already the case that there are more iOS devices sold than Mac sold ever (over its whole lifetime back into the 80s). Apple might stick a "classic box" on iOS but OS X would be dead ender. Like Apple GS/OS. And like OS 9.

If Intel starts to fail at competing yes. It isn't just possibility they should explore as it is more probable they would need it. It is possible to port to MIPS. What the reason to porting to MIPS? ( not alot). Same issue with ARM in the performance space where Mac are in now. If ARM was closing the gap on Intel, which they are not, then sure. So the probability is quite small right now.

 

[wildmac]

With the 6 and 12 cores, CPU isn't so bad, but they are slouches in other respects, mainly IO like SATA, TB, and USB. With the quads, they are outperformed by the high end mac mini.

Regarding the MacPros here..

The other thing is the antiquated video card solutions available.

No one wants to upgrade to these outdated systems, so they are sitting on much older systems waiting for something reasonable to upgrade to.

 

[The Deepness]

With the 6 and 12 cores, CPU isn't so bad, but they are slouches in other respects, mainly IO like SATA, TB, and USB. With the quads, they are outperformed by the high end mac mini.

Yikes! Now I understand why you Mac Pro folks have been screaming for an update so much. Hope you get a good announcement at WWDC.

 

[flux73]

And equaling or surpassing Intel power comsumption defeating the ARM purposes

For tech companies needing a CPU, licensing ARM chips will always have a cost benefit vs buying Intel. It also gives you flexibility in choosing your foundry and can modify their design to suit your needs.

It's true that Intel has the clear edge in manufacturing - both in scale and tech - but I don't think that will matter as much as the bottomline: cost. I don't see Intel ever providing a better value proposition. Don't forget they were once involved in building ARM chips with their XScale business which they sold to Marvell. The reason was said to be because they couldn't make (enough) money in it.

 

[smakdown61]

Hmmm, interesting differences between the "H" and "M" variants for the MQ processors.

Looks like the chips with the better iGPU (Iris 5200 pro) trades off significant CPU performance; the fastest chips, the 3.0GHz, have only the HD4600, which means that the Haswell MBP will probably still need a dGPU.

I wonder which ones the mac mini will get?

I was hoping they could have the entry 15" rmbp use only the iris5200 with no dgpu and drop the price to $2k or slightly under. Then have the more expensive versions have the faster cpu with dgpu. Although after looking at the price of that chip I don't think that will happen. Most likely will see all models with the hd4600 + 750m as predicted.

Sucks because I can certainly live with 650m performance and 10 hour battery life (potentially) for $2k or under.

 

[Pentad]

So you don't believe Apple is testing OS X on ARM in their labs in anticipation for future ARM chips that might be powerful enough to meet their needs?

Moving to ARM would be a huge disaster in many ways. Here are a few:

1. ARM is a great power sipping CPU but lacks the architecture to compete with x86 in performance. Having used both, there is just no comparison. Anything that is CPU demanding will crush ARM. It just wasn't designed to do that kind of work. It is a light-weight portable CPU.

I'm not degrading ARM but it is what it is. My lawn mower's engine is awesome but I'm not going to put it in my car.

2. An ARM CPU would break all current OS X applications. Sure, you could do a Rosetta deal but talk about SLOW. A slow CPU doing instruction set translations on the fly. Yeah, that will be fast.

3. Whether you like it or not, it is a Windows world. Windows runs on x86. When Apple moved to Intel, many people (including me) jumped on the bandwagon because I can still run Windows apps via VM or BC. Moving to ARM would kill that. The amount of people that would jump ship would really hurt Apple's bottom line.

4. Intel knows that if they can get an x86 CPU to sip power like an ARM they will rule the market (price being equal). An x86 tablet that can run Windows apps as long as an iPad would dominate the marketplace.

5. Intel is smart and they will have a power sipping CPU well before ARM can boost the performance of ARM. Intel has the technology, the people, and the experience.

Also, you can replace ARM with Apple CPU and it all stays the same.

People think Microsoft is stupid because they didn't do an iPad before Apple. It is much more complicated than that. If you do x86 you get a wealth of applications that run from the start but you will loose on battery life.

If you do ARM you loose the wealth of applications but gain battery life.

So you have Surface RT (ARM) which is a failure and Surface Pro (x86) which is some-what of a failure.


BTW: I am not an Intel fan boy, I'm just trying to put this into perspective here. I'm not sure everyone here truly realizes the depth of saying "Apple should move to ARM!"

-P

 

[tgi]

Agile Remote Missile.

Were you the victim of an ARM?

 

[WestonHarvey1]

I get a decent 6 hours on my 2010 MBA playing music and surfing/using pages?

Get a job sir

 

[deconstruct60]

I'm merely pointing out the new chip that's the new equivalent to the one they're using in the base MP right now.

It is NOT equivalent. Just because it says Xeon E3 doesn't means it is equivalent to the old 3500 and 3600 Xeon series. It is not in any way shape or form.

Intel has other lower number 3000 numbered Xeon that were based on mainstream desktop implementations. Mac Pro never used any of those.

The 3500/5500 and 3600/5500 had shared implementations (both implementation architecture and socket). Intel cleaned up the number scheme when shifted to Ex XXXX nomenclature so that the related series are now E5 1600 and 2600 . The '1' and '2' standing for how many sockets. Those are equivalent to the 3500/5500 and 3600/5600 pairings of before.

The Intel scheme in general.

http://communities.intel.com/commun...n-processor-numbering-more-than-just-a-number


Xeon E3 means significant step down for a Mac Pro. Not evolutionary at all in terms of performance. [ In may be evolutionary in terms of matching workload growth leveling off among a high fraction of Mac Pro buyers. ]

Maybe they would do two different boxes, but there's a good argument that they should have been doing that all along...and they haven't been.

More than just maybe. The Xeon E3 is meant for entirely different TDP profile. It would be a huge waste of space and effort to put it into a box sized to take dual E5 26000 and four hefty TDP PCI-e cards.

White box clone vendors do it because it cheaper to slap the same ATX sized board into the same ATX case they bought in bulk .... but it is highly doubtful Apple would do that.

Likewise Dell/HP/Lenovo (all the Tier 1 workstation vendors ) have different size boxes for their line ups as vary across E3 , E5 1600 , and E5 2600 ( e.g., HP z220 , z620 , z820 or Dell 3600 , 5600 , 7600 )

While there are some good reasons to do two different cases, I doubt it will happen.

I doubt so too.... which is why the Xeon E3 is likely out. ( Unless the focus is primarily on Thunderbolt.... then it is in because the implementation is way easier. ). The natural evolutionary path for a Mac Pro is E5 1600 and 2600 v2 which aren't due to be released until Fall. ( Apple could have done E5 1600 /2600 v1 in Jan-Feb but appears were completely not ready. )

There implicit presumption there though is that Apple still wants to track the dual customer market segment. Given that it appears that killed off Mac Pro R&D for a long while ( hence could not release anything earilier this year) that has a decent chance of happening.

If just trim down to just single package and put a high Thunderbolt focus on it ..... downshifting to E3 could be their move. They get USB 3.0 "for free" with the chipset ( just like the other Macs). iGPU to make TB easy to implement. If just a couple SATA 6Gb/s bays... done with chipset functionality.

With the quads, they are outperformed by the high end mac mini.

Not really as the speeds are a bit better for the E3. Frankly the quad 1600 will outstrip any mainstream Core i that Apple would use. The primarily reason why quad Mac Pro don't do well now is that they are still largely on the 3500 series which is almost two tick-tock Intel cycles behind. With E5 1600 last year (and upcoming) Intel stopped kneecapping the quad implementation on clock speed. They are now one of the fastest GHz in the E5 line up ( as they should be, as the core count is low and the TDP is allocated for alot more cores. )

 

[smakdown61]

Moving to ARM would be a huge disaster in many ways. Here are a few:

1. ARM is a great power sipping CPU but lacks the architecture to compete with x86 in performance. Having used both, there is just no comparison. Anything that is CPU demanding will crush ARM. It just wasn't designed to do that kind of work. It is a light-weight portable CPU.

I'm not degrading ARM but it is what it is. My lawn mower's engine is awesome but I'm not going to put it in my car.

2. An ARM CPU would break all current OS X applications. Sure, you could do a Rosetta deal but talk about SLOW. A slow CPU doing instruction set translations on the fly. Yeah, that will be fast.

3. Whether you like it or not, it is a Windows world. Windows runs on x86. When Apple moved to Intel, many people (including me) jumped on the bandwagon because I can still run Windows apps via VM or BC. Moving to ARM would kill that. The amount of people that would jump ship would really hurt Apple's bottom line.

4. Intel knows that if they can get an x86 CPU to sip power like an ARM they will rule the market (price being equal). An x86 tablet that can run Windows apps as long as an iPad would dominate the marketplace.

5. Intel is smart and they will have a power sipping CPU well before ARM can boost the performance of ARM. Intel has the technology, the people, and the experience.

Also, you can replace ARM with Apple CPU and it all stays the same.

People think Microsoft is stupid because they didn't do an iPad before Apple. It is much more complicated than that. If you do x86 you get a wealth of applications that run from the start but you will loose on battery life.

If you do ARM you loose the wealth of applications but gain battery life.

So you have Surface RT (ARM) which is a failure and Surface Pro (x86) which is some-what of a failure.


BTW: I am not an Intel fan boy, I'm just trying to put this into perspective here. I'm not sure everyone here truly realizes the depth of saying "Apple should move to ARM!"

-P

I think we have a better chance of seeing intel chips in the iphone/ipad than we do seeing ARM in a macbook. I believe that was Job's goal back in 2007 with the iphone but someone important threatened to quit his team if they didn't go ARM over intel, which back then would have been a disaster.

 

[Renzatic]

Were you the victim of an ARM?

Victim? No.

Acquitted? Oh yes.

 

[Onimusha370]

Get a job sir

I have 3 whole months of playing video games, then uni before I need to do that!

 

[TMar]

I think your friend picked up the last revision of the latest chip. Haswell hasn't even launched desktop chips as far as I know yet.

They have been out a few days now...

 

[iChrist]

Do you think they will give the boost to the 15" rMBP as well? Is there any reason to believe they would not?

I hope so. The performance of the current model is subpar (screen stuttering). Many issue with USB as well.

 

[deconstruct60]

Moving to ARM would be a huge disaster in many ways. Here are a few:

1. ARM is a great power sipping CPU but lacks the architecture to compete with x86 in performance. Having used both, there is just no comparison. Anything that is CPU demanding will crush ARM. It just wasn't designed to do that kind of work. It is a light-weight portable CPU.

I suspect someone is going to counter that ARM 15 is suppose to bring it up to parity to the "big boys". There is huge problem why that really isn't true. ARM is still 32-bit. OS X just made the transition to being 64-bit top-to-bottom. Why on Earth would Apple go back to 32-bit process address space????????????

ARM 15 added some gimmicks to address larger than 32-bit physical addresses of memory. That is primarily to serve to support the ability to run multiple virtual machine images ( each of which is 32-bit OS) on the same server. [ It all goes to cloud server consolidation. Think 20 VMs running 20 different OS server images of some blogs that get 50,000 hits a day. Which is nothing so the images are largely idle doing nothing most of the day. That's what ARM 15 targeted toward. Single users from a timeslice point of view. ]

Not even 64 bit is basically a show stopper. Yeah sure Apple could add there own custom 64 bit implementation (it has been done by others). Eventually ARM will added in yet another future generation....but until it gets to 64-bit it isn't very viable. Even a Rosetta/Classic emulator is problematical when don't even have the same size process address space. Sure there is some Rube Goldberg hacks that might get around the problem .... but basically Apples and Oranges.

 

[WestonHarvey1]

I have 3 whole months of playing video games, then uni before I need to do that!

Well there you go, let's see what some gaming does to the battery!

 

[Tech198]

Yay... Haswell....!!

Looks like these new chips are primary designed and will be see more benefit for laptop users rather than desktop since all they'll see is CPU increases... and probably graphics too.

 

[nunomoreira10]

Can the 13r get this rumored 28w with iris 5100 in the fall?

http://www.anandtech.com/show/7012/...ell-with-iris-5100-graphics-2560-x-1440-panel

if there is this dual core 28w im sure a custum version for apple with 37w sould be possible.
still no iris pro

 

[ric275]

Im waiting to order a 27" iMac... Any chance of an iMac refresh with these processors anytime soon?

 

[Oletros]

For tech companies needing a CPU, licensing ARM chips will always have a cost benefit vs buying Intel. It also gives you flexibility in choosing your foundry and can modify their design to suit your needs.

It's true that Intel has the clear edge in manufacturing - both in scale and tech - but I don't think that will matter as much as the bottomline: cost. I don't see Intel ever providing a better value proposition. Don't forget they were once involved in building ARM chips with their XScale business which they sold to Marvell. The reason was said to be because they couldn't make (enough) money in it.

I'm not talking about costs, I'm talking about processing power and energy efficiency

 

[torana355]

Nice, i think its time to upgrade my 2011 MBA 13".

 

[toaster64]

Yay, Mac news.