New Intel Sandy Bridge Xeon chips available for potential Mac Pro update
Intel's new Sandy Bridge update for its Xeon line of high end CPUs is due next week, suggesting the potential for Apple to refresh the Mac Pro, which hasn't changed since the middle of 2010.
The availability of Intel's latest Xeon E5 workstation-class chips next week was reported by British site The Inquirer.
The new Xeon E5 chips incorporate the Sandy Bridge micro architecture that first appeared in MacBook Pros and iMacs early last year, followed by a mobile variant used by Apple in the MacBook Air last summer.
Apple's latest Mac Pro models currently use Intel Xeon Bloomfield or Gulftown processors based on the Nehalem and closely related Westmere microarchitectures.
The latest release of OS X 10.7.3 Lion included support for AMD's high end Tahiti graphic cards, which are expected to arrive in the market around the same time as Intel's new Xeon chips.
However, people famliar with the matter have said that Apple's management, as far back as last May, were in limbo over whether to put any additional resources toward the Mac Pro product line.
Internal discussions at Apple were said to focus on the fact that sales of the high-end Mac Pro workstations have dropped off so considerably that the desktop machines are no longer particularly profitable for the company.
[ View article on AppleInsider ]
Comments
So yeah I hope Apple uses this chip, I just hope that it is in a modern implementation that completely leverages the technology they have available these days. Oh and please do it at a reasonable price point.
I personally think Apple would be making a big mistake not offering their Pro customers and upgrade path to Sandy Bridge Xeon. There are a lot of creative professionals in Music, Video, Graphics, etc. that would likely migrate to PC platforms...
Of course Apple has to be careful about what they replace the Mac Pro with. Performance is paramount but the architecture of the current Mac Pro isn't really suitable for further development. To support things like TB, Solid State Storage, optical interconnects and the like requires a bit of rethinking about what a workstation is.
For example supporting SATA in the base unit might be seen as a transitional have to have but long term they will need a port to support very fast solid state storage methods. Not all of these solid state initiatives are flash based so it really needs to be some sort of advanced PCI-Express 3 port. The same thing goes for Thunderbolt, it doesn't look like the requirement for motherboard integration will go away and it will only need to b e faster in the future.
One might figure that with Apple's profitability they could afford to carry one piece of hardware that doesn't make a huge profit. I'd sure hate to see the Mac Pro line die, although I don't have much use for one since the top-of-the-line iMac is enough for me.
Until Apple addresses a few issues with the iMac I have no need for one. Likewise I have no need for the Mac Pro in its high price configurations. I would go for a redesigned machine if there was one option available at a reasonable price that had respectable performance.
This is in fact one of the Mac Pros greatest failings, there isn't really a viable low cost option. If you are just looking for a bit of expandability in a modern desktop Apple doesn't have one.
It's just that the Mac Pro seems like such a wonderful design. I suppose that isn't enough reason to keep the Mac Pro around if not enough people are buying it.
It was at one time a wonderful design and still is if you have a need for the top end configurations and can justify the cost. Most people can't though, thus low sales.
What a shame if Apple discontinues it. Now that Apple is getting into the enterprise, it might have some chance of getting better sales.
There is no shame. Keeping the Pro around makes as much sense as keeping the Apple 2 around. The Apple 2 was an outstanding little machine in its day, but has been well eclipsed buy technology far better. The problem with the Pro is that it is quickly begin eclipsed by the advent of new technologies that make it a questionable investment for Apple. Apple can do much better with a platform that can cover the needs of a wider array of users.
Also the usual stuff, 2,499.00 for 3GB is just flinging poop at pro mac users, 6GB should be standard.
Wow, it really has been a while since Nehalem, and once the SB Xeons are ready Ivy Bridge on the consumer side will be almost ready, too bad Intel doesn't release them at the same time.
I wonder how Thunderbolt would work on a desktop too, for video out from the video card it would need to be connected there, but it uses a lot of bandwidth, could the GPU pass all that through?
The Mac Pro is really outdated technology for a workstation. I'd much rather see Apple take something line Sandy Bridge E and make a thoroughly modern and forward looking desktop computer. The Mac Pro was the right approach ten years ago but it isn't the product to base the next ten years of development upon.
So yeah I hope Apple uses this chip, I just hope that it is in a modern implementation that completely leverages the technology they have available these days. Oh and please do it at a reasonable price point.
"The Mac Pro was the right approach ten years ago but it isn't the product to base the next ten years of development upon."
I'm curious (not trying to be sarcastic), what would be the "right" Mac Pro to base the next 10 years upon?
I honest think they will have to integrate the GPU onto the motherboard. This might result in fixed GPU memory, but if Apple was smart they would put the GPU memory on a separate card. The other option would be multiple Mac models varying in GPU implementation.
The more that I think about this the more I believe that the Pro is due for a total make over. This is just one technical issue that Apple will be running into.
I wonder what the baseline GPU will be? Tahiti is the top end Radeon 7970, nothing in the Mac Pros history suggests they would ship with something that expensive as a baseline. Todays uses the 5770, so my guess would be the 7770, but nothing in the OS code showed Pitcairn support, did it? Something with Graphics Core Next would be nice. And for a "Pro" machine 1GB video memory is just cutting corners, the upgrade to 2GB card-wise doesn't cost much and in professional applications like Mudbox 1GB slows you down, and historically the only way you got more VRAM was switching up to the far more expensive workstation class cards which not everyone needs, so I hope they have 3GB non-workstation options this time.
Also the usual stuff, 2,499.00 for 3GB is just flinging poop at pro mac users, 6GB should be standard.
Wow, it really has been a while since Nehalem, and once the SB Xeons are ready Ivy Bridge on the consumer side will be almost ready, too bad Intel doesn't release them at the same time.
I wonder how Thunderbolt would work on a desktop too, for video out from the video card it would need to be connected there, but it uses a lot of bandwidth, could the GPU pass all that through?
The Mac Pro is really outdated technology for a workstation. I'd much rather see Apple take something line Sandy Bridge E and make a thoroughly modern and forward looking desktop computer. The Mac Pro was the right approach ten years ago but it isn't the product to base the next ten years of development upon.
So yeah I hope Apple uses this chip, I just hope that it is in a modern implementation that completely leverages the technology they have available these days. Oh and please do it at a reasonable price point.
I'm expecting some potentially updated workstation designs seeing as most of them much like Apple haven't updated much since 2009. Some of the Sandy Bridge chips are up to 130W if I recall correctly. I'm not sure how the next two revisions look on power appetite. Quadro variants on kepler will also be a big deal given the relative dominance of NVidia in the workstation market segment. The quadros are mostly driver based adjustments rather than raw hardware, but it makes a very significant difference as long as the drivers are well tuned and tested on the application. Nothing sucks more than lag (aside from perhaps crashes.
HP and Lenovo came out with more compact designs. The HP was an all in one like the imac, but with full access panel. The hardware starts with something consumer level but has a lot of options for workstation gpus, unfortunately they're mobile workstation gpus, and other things. I don't what the display is like. If they have an option that comes close to their dreamcolor displays (even if it's an extremely expensive upgrade) they could actually have something there. Those never caught on in volume, but they are very good. Before anyone mentions it, those ones cater to a different market than the TB display.
All of the really compact models have been pretty expensive thus far. i'm not sure if they required exotic cooling solutions or anything. The point there is that smaller case doesn't necessarily translate to cheaper. With the top end market items, you always pay a much heftier premium, but it does seem to be becoming a much wider price gap. Apparently they do have a dreamcolor option on the display. That display alone sells for $2500 normally for a 24" display, and roughly $3k set up fully for proper calibration and profiling. If the final price point and warranty on the rig makes sense as a whole, it could be okay. I think it's likely that an impressive configuration will be really really expensive early on. The still using a mobile workstation gpu variant thing makes me a little sad.
I seriously doubt any of the 130W cpus will make it into any of these. Last time we saw those kinds of figures in workstation cpus, oems waited it out. The same might be true here. Obviously intel hasn't been that focused on this segment either.
"The Mac Pro was the right approach ten years ago but it isn't the product to base the next ten years of development upon."
I'm curious (not trying to be sarcastic), what would be the "right" Mac Pro to base the next 10 years upon?
Well, since you're responding to wizard69, I"m sure he's thinking of something like this:
http://geekhack.org/showthread.php?1...o-you-have-one
"The Mac Pro was the right approach ten years ago but it isn't the product to base the next ten years of development upon."
I'm curious (not trying to be sarcastic), what would be the "right" Mac Pro to base the next 10 years upon?
First; we have t think about Thunderbolt (TB) and how Apple will go about supporting that and the associated video channels. The first possibility is that they punt and simply don't support the video multiplexing. I don't think that will happen. They could go with a special purpose video card connector but again I don't think that ill happen. I suspect that they will integrate the chip right on the motherboard. This is the right long term solution especially as video hardware becomes more closely coupled to the CPU hardware and allows them to support TB as they already have been. Like it or not this is one reason why I think the Pro is dead.
Second; SATA is pretty much dead as the primary storage channel for forward looking PC hardware. Instead I'm expecting that a base machine would rely upon solid state storage implemented on an advanced plug in card. This card would use the latest PCI Express interface electrical standards but might vary in physical dimensions. This would allow Apple to implement very fast solid state arrays for the primary secondary storage implementation ( in other words the boot and app drive). I'm careful to say solid state here as i don't expect flash to be around for long, however memory on a card is the long term play. I could see Apple implementing two or three slots dedicated to high speed system storage.
Third; I see a move away from dual socket systems. Things like Sandy Bridge E will improve rapidly and some of the leak benchmarks are indeed impressive. If we do see dual socket in the future one of those sockets will be for the GPU at which point it will be an equal to the CPU on the memory bus. Interestingly I see AMD as the more interesting play here with their GCN initiative. In any event I suspect all future machines will have one CPU socket
Fourth; far more TB ports. I wouldn't be surprised at all to see at least four on a Pro replacement machine. They will of course be used to hook up the video screens but I see extended usage for connection to disk arrays, and for the creation of clusters of Macs. TB all also start to support a lot of novel hardware so the more ports the better.
Fifth; if the new machine provides for any sort of magnetic drive capability it will probably be with small notebook sized drives. The preference for mass storage will be TB connected disk arrays. The Pros replacement would signal the end of magnetic media in desktop machines.
Sixth; Even though the machine will be far more compact I expect that it will still have expansion slots. I just don;t think they will be capable of supporting graphics cards.
Seventh; the same chassis that supports an advanced machine running Sandy Bridge E or what ever workstation processor will also have a mid level performance machine. This would use a Fusion or Ivy Bridge like processor to deliver something better than Mini performance. Think of this as a mid range volume machine where GPU performance can be more modest.
I'm expecting some potentially updated workstation designs seeing as most of them much like Apple haven't updated much since 2009. Some of the Sandy Bridge chips are up to 130W if I recall correctly. I'm not sure how the next two revisions look on power appetite. Quadro variants on kepler will also be a big deal given the relative dominance of NVidia in the workstation market segment. The quadros are mostly driver based adjustments rather than raw hardware, but it makes a very significant difference as long as the drivers are well tuned and tested on the application. Nothing sucks more than lag (aside from perhaps crashes.
I'm not convinced that NVidia will be around for the long haul. Honestly I'd rather see AMD GPUs in these new workstations.
However this is a Sandy Bridge E thread so let's worry about those. I noticed that 130 watts concerns you but consider that this might not be the issue you think it is. Why, single socket workstations. I could see many punting when it comes to dual socket workstations.
HP and Lenovo came out with more compact designs. The HP was an all in one like the imac, but with full access panel. The hardware starts with something consumer level but has a lot of options for workstation gpus, unfortunately they're mobile workstation gpus, and other things. I don't what the display is like. If they have an option that comes close to their dreamcolor displays (even if it's an extremely expensive upgrade) they could actually have something there. Those never caught on in volume, but they are very good. Before anyone mentions it, those ones cater to a different market than the TB display.
More compact workstations will be the order of the day. As will be matching storage arrays. I could see a decent workstation half he size of the current Mac Pro. It is just a matter of dropping all of the legacy stuff and concentrating on what is needed.
All of the really compact models have been pretty expensive thus far. i'm not sure if they required exotic cooling solutions or anything. The point there is that smaller case doesn't necessarily translate to cheaper.
Well the case wouldn't. Most of your cost is tied up in other components. The issue with the cases is that it takes up room it doesn't have to take up. Much of that room is devoted to legacy hardware, delete that and the case will shrink fast.
With the top end market items, you always pay a much heftier premium, but it does seem to be becoming a much wider price gap.
Much of that is due to excellent Intel pricing. Let's face it Intel has gotten away with some serious price increases. This is another reason to favor a smaller case though because it becomes a platform for a midrange desktop machine. Swap out the high performance motherboard for one with a Fusion or Ivy Bridge processor and many desktop needs are satisfied. You might also need a powersupply swap but the mechanical and even some of the electrical architecture would serve two classes of performance.
Apparently they do have a dreamcolor option on the display. That display alone sells for $2500 normally for a 24" display, and roughly $3k set up fully for proper calibration and profiling. If the final price point and warranty on the rig makes sense as a whole, it could be okay. I think it's likely that an impressive configuration will be really really expensive early on. The still using a mobile workstation gpu variant thing makes me a little sad.
I'm not a graphics pro so I drift off when talk turns to calibrated monitors. However there is one other reality manufactures have to deal with these days. That is the Green movement, so that mobile GPU may be there simply to lower the machines overall power profile as a selling point. Or it simply was needed for thermal reasons.
I seriously doubt any of the 130W cpus will make it into any of these. Last time we saw those kinds of figures in workstation cpus, oems waited it out. The same might be true here. Obviously intel hasn't been that focused on this segment either.
A 130 watt CPU is a bargain if you are trying to get away from dual socket machines.
Well, since you're responding to wizard69, I"m sure he's thinking of something like this:
http://geekhack.org/showthread.php?1...o-you-have-one
Even a half-compromise would work. Selling the Mac Pro just as parts, so that people with existing models can swap out the Motherboard, CPU, and RAM if they want a base unit upgrade, or swap out the GPU's if they just need faster GPU's. Or bring the Mac Pro into an Apple store and have the parts swapped out.
New PC motherboards support EFI, another compromise could be just to have Pegatron (Which used to be part of ASUStek) build OS X compatible Socket 2011 boards, and Apple could sell a OS X Workstation for roughly the same price as Windows 7 Complete that only installs on licensed boards. This gives an out to those that actually NEED OS X on high end equipment without opening the market to cheap clones.
But for the short term I still expect Apple to release a Socket 2011 Mac Pro
If there is any company on this planet that could do this, it would be Apple.
First; we have t think about Thunderbolt (TB) and how Apple will go about supporting that and the associated video channels. The first possibility is that they punt and simply don't support the video multiplexing. I don't think that will happen. They could go with a special purpose video card connector but again I don't think that ill happen. I suspect that they will integrate the chip right on the motherboard. This is the right long term solution especially as video hardware becomes more closely coupled to the CPU hardware and allows them to support TB as they already have been. Like it or not this is one reason why I think the Pro is dead.
TB hasn't shown full performance with displays attached. Given the potential to decouple the chip from displays there and offer its full bandwidth potentially without the need to route through integrated graphics, this could be a decent path forward.
Second; SATA is pretty much dead as the primary storage channel for forward looking PC hardware. Instead I'm expecting that a base machine would rely upon solid state storage implemented on an advanced plug in card. This card would use the latest PCI Express interface electrical standards but might vary in physical dimensions. This would allow Apple to implement very fast solid state arrays for the primary secondary storage implementation ( in other words the boot and app drive). I'm careful to say solid state here as i don't expect flash to be around for long, however memory on a card is the long term play. I could see Apple implementing two or three slots dedicated to high speed system storage.
SATA isn't going to disappear overnight given the cost of high density solid state storage. I think it'll still take some time for that to mature. It's still pretty costly per gigabyte and you can fully saturate a storage bus with newer SATA drives, so the cost: benefit ratio isn't quite there yet even if it is the likely direction.
Third; I see a move away from dual socket systems. Things like Sandy Bridge E will improve rapidly and some of the leak benchmarks are indeed impressive. If we do see dual socket in the future one of those sockets will be for the GPU at which point it will be an equal to the CPU on the memory bus. Interestingly I see AMD as the more interesting play here with their GCN initiative. In any event I suspect all future machines will have one CPU socket
Intel has priced dual socket higher and higher, so it has been becoming in some ways a niche market. Sometimes it can still make sense, but the situations are limited. Part of what has made a dual socket configuration unattractive is the number of applications that do not gain much from high core counts. 8 core has been available for years and yet many many applications don't scale past 2-4.
Fourth; far more TB ports. I wouldn't be surprised at all to see at least four on a Pro replacement machine. They will of course be used to hook up the video screens but I see extended usage for connection to disk arrays, and for the creation of clusters of Macs. TB all also start to support a lot of novel hardware so the more ports the better.
I'd have to look up how many are supported per chip and how many total PCI lanes this derives from. Given the upcoming PCIe 3 squeezing out more thunderbolt ports might not be that difficult.
Fifth; if the new machine provides for any sort of magnetic drive capability it will probably be with small notebook sized drives. The preference for mass storage will be TB connected disk arrays. The Pros replacement would signal the end of magnetic media in desktop machines.
I disagree with you here. If you look at other workstation manufacturers, many of them grant the option to use 3.5" drives or more 2.5" drives. Sometimes 3.5" can still make total sense, so the option is nice. You can get a huge amount storage from 3.5" these days, but overall storage needs have still been increasing.
Sixth; Even though the machine will be far more compact I expect that it will still have expansion slots. I just don;t think they will be capable of supporting graphics cards.
Seventh; the same chassis that supports an advanced machine running Sandy Bridge E or what ever workstation processor will also have a mid level performance machine. This would use a Fusion or Ivy Bridge like processor to deliver something better than Mini performance. Think of this as a mid range volume machine where GPU performance can be more modest.
That would be logical assuming a cost effective base chassis, and Apple seems to do well getting a ridiculously good price on things.
I'm not convinced that NVidia will be around for the long haul. Honestly I'd rather see AMD GPUs in these new workstations.
However this is a Sandy Bridge E thread so let's worry about those. I noticed that 130 watts concerns you but consider that this might not be the issue you think it is. Why, single socket workstations. I could see many punting when it comes to dual socket workstations.
There are two things to mention there. Under OSX AMD has done fine, but they've never released a single FirePro model. It only matters at a somewhat extreme end in the sense of those who depend on a fluid OpenGL experience in affected applications. Under Windows and Linux, NVidia has been the preferred card. Both mark up workstation cards heavily over similar hardware on consumer type cards. The differences are drivers and sometimes ram. Higher ram density might drive up manufacturing costs, and the rest is just the cost of developing specialized drivers for a smaller market. Under Windows rather than trending away from a benefit seen from such hardware, some applications have actually become more dependent on it with newer features being implemented. If you can spend thousands on software, the gpu cost doesn't seem quite so horrendous.
NVidia so far has remained a definitively superior option outside of Macs. Most of the time the complaint with AMD seems to be drivers, which happens to be an NVidia issue under OSX with their last Quadro option (unless they fixed that finally).
More compact workstations will be the order of the day. As will be matching storage arrays. I could see a decent workstation half he size of the current Mac Pro. It is just a matter of dropping all of the legacy stuff and concentrating on what is needed.
Well the Lenovo still supported dual socket configurations. Even applications that are built for high core counts do top out somewhere, so it's a matter of how many cores still make sense. Part of the issue with the dual socket workstations and market share has been pricing as I mentioned.
Well the case wouldn't. Most of your cost is tied up in other components. The issue with the cases is that it takes up room it doesn't have to take up. Much of that room is devoted to legacy hardware, delete that and the case will shrink fast.
Cases can still be expensive. My point about compact is if it requires a lot of extra effort to maintain cooling or during assembly, it can cost just as much as the larger one (or more). It depends what is required to make the intended components run within the box is all.
Much of that is due to excellent Intel pricing. Let's face it Intel has gotten away with some serious price increases. This is another reason to favor a smaller case though because it becomes a platform for a midrange desktop machine. Swap out the high performance motherboard for one with a Fusion or Ivy Bridge processor and many desktop needs are satisfied. You might also need a powersupply swap but the mechanical and even some of the electrical architecture would serve two classes of performance.
Part of it is intel, but it seems like the high end is where they've increased prices. Apple has also kept their base pricing similar, but migrated to cheaper hardware for it.
I'm not a graphics pro so I drift off when talk turns to calibrated monitors. However there is one other reality manufactures have to deal with these days. That is the Green movement, so that mobile GPU may be there simply to lower the machines overall power profile as a selling point. Or it simply was needed for thermal reasons.
A 130 watt CPU is a bargain if you are trying to get away from dual socket machines.
The calibrated display thing is commonly misunderstood. Until 2007 or so, it was near impossible to find something that was competitive with a high end crt. The misunderstanding is that whatever display they own, a $130 spyder (I hate datacolor, I use an xrite device on mine) will make it perfect and it will match everything else. It won't. It can't change the uniformity or halt the inherent drift of the hardware. It just describes it to the computer. More sophisticated combinations do a bit more.
I'm thinking of the 2007 mac pro here. Apple took the 130W displays. The other oems passed, but as you mentioned that was a dual socket setup. I think the mobile gpu is for cooling reasons seeing as gpu power can be incredibly useful at times.
Any word on the pricing? With the decline of the desktop and workstation it might finally behoove Apple to lower the entry price a bit.
Decline of the desktop?
Lower prices?!
'He's a witch...BURN HIMMMMMM!'
Lemon Bon Bon.