Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
It's about time, I've been ready to buy a new machine for a while but they haven't updated in ages.

Well you're going to have to wait longer. Intel just pushed back everything a full quarter, this includes the 2socket tylersburg systems which effects Mac Pros unless Apple is getting some special deal which is possible. (i.e. 2socket X58s, so theoretically apple could use desktop parts on the pros now with this board :O )

Intel to push out Everything in 2009


Basically they're pushing everything out because the economy is so bad and slow and cant afford the upgrade/need the upgrade, not cause they aren't ready or bugged.




Edit: Just realized this...This is going to be fun to watch. Apple actually has a choice this time. They have a choice between Beckton or Gainstown..The difference being Beckton's use FBDIMMs like the last MacPros but Intel this is intels last FBDIMM product (they're trying to get it away from it bick time!!) And Gainstown is the exact same chip but just uses reg DDR3...(EDIT GAINSTOWN HAS 8 MB L3 CACHE BECk has 24mb, WOW that might make a difference?) Interested to see what Apple ultimately chooses to put in the pros. If they choose gainstown they wont have any performance loss, and it will be a lot cheaper because the chips themselves will be, and regular ddr3 is A LOT cheaper then fb dimms...
 
Mac Pro is a workstation

Apple has positioned the Mac Pro as a workstation. It will likely remain that way and IMHO it should. Since it is workstation class it will use a chipset, processor, and memory designed for a workstation. It is highly unlikely Apple will do otherwise (of course they could bring out a lower end tower with desktop parts).
 
Configured with a pair of 2.8GHz Nehalem EP chips... and 24GB of DDR3 memory (not the fastest either presumably, at 1,066MHz).

It would be a bit of a stir if a finalised 3.2GHZ model could be within shouting distance on the SPECfp of an AMD 4S Shanghai rig.

Think having core optimisation on top.... "nice".
 
Xeon's architecture are identical to the Core (desktop parts). The differences are very minor. Last generation, Core2, the difference being the socket because they used FB-DIMM. Sure special versions, typically lower power or MP versions will have larger or smaller cache, depending on what the chip is made for, but again these chips are identical. Intel release's desktop chips for UNI processor systems, then rebrands/rebins those same chips, and calls them Xeons. There are a few cases where you can run 2 desktop chips on a special board (more so last generation than this) but thats the main distinction. Desktop = single socket
Xeon = Dual+

The main architectural difference this generation is that the Xeons have pins/die space for TWO QPI connects which allows it to be placed in system with 2 or more. The Desktop chips only have one active QPI link (the other is disabled) which only allows to be used by itself.

Sooooo in conclusion. YOU CANT HAVE THE "DESKTOP" chips that are out now in your precious Mac towers. And the chips that will be coming out for your mac towers in less then two months WILL BE IDENTICAL performance wise.

Smile everyone :)
Thanks for doing this for me. :D
 
Edit: Just realized this...This is going to be fun to watch. Apple actually has a choice this time. They have a choice between Beckton or Gainstown..

Beckton is for servers not workstations. A 2.4GHz Beckton processor will probably start at around $3,000. Apple can do a 2.66GHz 8 core Mac Pro for that.
 
All the reports so far including from someone I know works at intel with these chips is a 30% - 40% increase, not an amazing 60% which this article is claiming.

The only numbers they gave were for SPECfp. I'm not surprised at all that the new CPUs can crunch floating point numbers much faster. The article said nothing about general purpose performance on "normal" tasks.

For most users these processors will be like fitting a 400 horse power engine into your car. It will not get you home from work any faster. Most of the time if you watch the Activity Meter you see the CPUs are not running at 100% On most Macs the computers spends most of it's time waiting for the user to click the mouse
 
For most users these processors will be like fitting a 400 horse power engine into your car. It will not get you home from work any faster. Most of the time if you watch the Activity Meter you see the CPUs are not running at 100% On most Macs the computers spends most of it's time waiting for the user to click the mouse

Great analogy. The weak link really is the user, followed by the hard drive. All I know it thise CPUs will be fast, and coupled with GPU-power for the floating point stuff and Snow Leopard, this is gonna be nuts! :eek:
 
You can't put dual Core 2's into a system, Intel wont let you.

Intel does let you - via their Skulltrail platform (wintel) - however no system exists for Mac yet.

Who knows - you might get lucky as Intel will be revisiting skulltrail early next year for Core i7
 
Intel does let you - via their Skulltrail platform (wintel) - however no system exists for Mac yet.

Who knows - you might get lucky as Intel will be revisiting skulltrail early next year for Core i7
You know that Skulltrail is currently a Xeon LGA771 platform, right?
 
For most users these processors will be like fitting a 400 horse power engine into your car. It will not get you home from work any faster. Most of the time if you watch the Activity Meter you see the CPUs are not running at 100% On most Macs the computers spends most of it's time waiting for the user to click the mouse

Is it not potentially possible that by having that power, you get emergent properties - more is not the same, but different as Shirky might say?

If you knew you had that power, you might be tempted to go drag racing, maybe take it out on a tour with some friends, take it to the Nuremberg ring, to stretch the analogy.

For one, if this was game compatible, you could use some of that power, if Intel opened up their cloaking for example.

I agree with you in the main, but I think having the capability, even from consumers getting the desktop Core i7 benefits as opposed to server class would be used. Today's consumer desktop CPUs are equivalent to yesteryears server class (the 920 or above in comparison to a Skulltrail rig would be one example i'd imagine).

Anyone seen what Clarkson has been doing with a V8 recently? All sorts of practical uses for that 400bhp CPU engine... As MrCrowbar points out - think of what a decent bunch of Apple engineers could get up to with this sort of con/prosumer CPU power!
 
It's not at all clear to me that SPECfp is a representative benchmark for most things people will be using a Mac Pro for. Nice numbers though.
 
No reason to yell or scream at me.

He did neither. He used capitol letters...

However, Kabunaru I don't know why you keep pushing for the removal of server parts in favor of desktop parts, what's your reason? If it's just so you can have a cheaper Mac Pro that's ridiculous, then it would be less Pro. That's akin to me asking for an iMac, but with the computer bits taken out so I can afford it, just so I can say I have an iMac.
 
Unless you're running a server, you don't really need a server class chip. You don't need to worry about stressing it out or wearing it out either. A server class CPU isn't going to last any longer than a normal CPU, they're not made "tougher" or anything of the sort.

I think there are more reasons to have a server class chip than just running a server. If you are doing any sort of computational work, having a server class chip severely reduces the errors that are created during computation that wouldn't matter in the same way as errors created in a program surfing the internet would. Feel free to correct me if I'm wrong, but I think that many jobs that are run on a server could be just as easily run on a desktop and in those cases you would benefit from a server processor.
 
Given the clear bias in the article. I would give odds of 2 to 1 that the author is taking it up the chuff from Intel.

All the reports so far including from someone I know works at intel with these chips is a 30% - 40% increase, not an amazing 60% which this article is claiming.

They are clearly faster then penryn but not as fast as claimed in this article. As you guys will note if any of you crunch at Seti. The penryn chips are still faster when it comes to the short tasks that Who?'s 8 Core Nehalem machine.
Core-for-core, I've heard it's a 15%~20% increase. Uncore parts make up the extra % increase. Also, the 60% is for floating-point performance, which I've heard Nehalem improves significantly on. Other benchmarks and "real-world" tests will likely show a smaller performance increase.
 
Desktop *and* Workstation, not Desktop *or* Workstation

The Mac Pro IS a workstation, hello???

That's why Apple needs to *ADD* a mini-tower that uses desktop parts to the product line.

The Mac Pro is humonguous, and expensive.

The Mac Mini isn't much more than a toy.

The Imac's an all-in-one - 'nuff said.

Time to *ADD* a mini-tower to the lineup....
 
It's the on-die memory controller

Also, the 60% is for floating-point performance, which I've heard Nehalem improves significantly on. Other benchmarks and "real-world" tests will likely show a smaller performance increase.

SPECfp is as much a memory test as a floating point test - it takes a lot of memory bandwidth to do well on SPECfp. You need to stream the FP data in and out of the CPUs.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.