I'm sure it is spectacularly fast but right now with my 2010 Mac Pro 8 core, it is already faster than I am. I'm pretty efficient when it comes to multitasking. I usually have around five CS applications open at the same time and switching between them constantly. If I am rendering a movie, I can still work on other stuff at the same time. In a single task environment, if movie rendering is the only important thing, then brute power is important, but in my routine, I'm usually juggling ten balls in the air at once and my current Mac Pro has not let me down yet.
I have the same MP and it has never let me down either. These MP's are simply they best, and I have always bought those (PowerMacs, Performa's et cetera before that).
Pricing it out on the Dell's website tonight ends with a $8,019 quote for:
1300W. Wow. What's up whit this '85& efficiency'?
No Energy Star? Will Greenpeace go after Dell as well? edit: it's a $0.00 option you can check.
Windows 8 not mandatory? Good. Can you also choose Linux? edit, nope, didn't find one. What a God awful website they have. The summary for your config uses 6 different colours.
If so, why would Apple need to do it online, except to "leak" the machine's performance to the public. However, judging from the unremarkable results, that's not very likely.
Huh? This is comparing a single 2.7ghz 12 core xeon cpu to the 2010 mac pro with dual 3.07ghz 6 core xeon cpus, and it still beat it by 10%. Not to mention, ~1/8th the size?
It's still comparing a 3-year old top-of-the-line computer with a brand new top of the line computer. Honestly, people with Mac Pros already have enough space that the size won't make a difference; I mean the Mac Pro is already sitting there. A 10% increase in power over 3-year-old tech is not exactly mindblowing. I mean every year a new Macbook is released, Apple calls it "50% more powerful than last year's" and things like that - not "10% more powerful than 3 generations ago".
It's still comparing a 3-year old top-of-the-line computer with a brand new top of the line computer. Honestly, people with Mac Pros already have enough space that the size won't make a difference; I mean the Mac Pro is already sitting there. A 10% increase in power over 3-year-old tech is not exactly mindblowing. I mean every year a new Macbook is released, Apple calls it "50% more powerful than last year's" and things like that - not "10% more powerful than 3 generations ago".
And I think this trend will continue; the increases will be less and less more than they've been. Many articles on this very subject out there; the tech simply won't increase twofold every year. Not anymore.
most of math operation within application are done on 32 integers and floats and this is what is important for CPU performance. 64-bit memory addressing is another story.
You folks are delusional if you think Intel's Xeon processors have jumped in performance to trounce the 2010 Mac Pro. They haven't. Intel has hit a wall and is already rearchitecting their CPU systems to follow AMD's path.
You folks are delusional if you think Intel's Xeon processors have jumped in performance to trounce the 2010 Mac Pro. They haven't. Intel has hit a wall and is already rearchitecting their CPU systems to follow AMD's path.
I guess we'd expect 'upto' x2 performance like Apple says. Whatever that means...and on what app and on what benchmark etc.
AMD's Opterons look pretty good bang for buck... I guess it's Intel though. And it should bring appreciable performance cpu gains over teh '3 year old' model. (Sniggers derisively at last year's 'upgrade...')
I think the move to gpu computational emphasis with dual GPUs is long over due and definitely in sync with the push to make Open CL adoption all pervasive. It's a smart move. For those task that can take full advantage we can get an order of magnitude improvement in certain benches/apps. Also, dual GPUs can certainly throw around what's in your painting and modelling window very quickly if the Mari/Pixar' demo is anything to go by.
Fast cpu, memory and bus, even faster x2 gpus with faster internal storage. Anything you need more than that? Go external.
You folks are delusional if you think Intel's Xeon processors have jumped in performance to trounce the 2010 Mac Pro. They haven't. Intel has hit a wall and is already rearchitecting their CPU systems to follow AMD's path.
I guess we'd expect 'upto' x2 performance like Apple says. Whatever that means...and on what app and on what benchmark etc.
AMD's Opterons look pretty good bang for buck... I guess it's Intel though. And it should bring appreciable performance cpu gains over teh '3 year old' model. (Sniggers derisively at last year's 'upgrade...')
I think the move to gpu computational emphasis with dual GPUs is long over due and definitely in sync with the push to make Open CL adoption all pervasive. It's a smart move. For those task that can take full advantage we can get an order of magnitude improvement in certain benches/apps. Also, dual GPUs can certainly throw around what's in your painting and modelling window very quickly if the Mari/Pixar' demo is anything to go by.
Fast cpu, memory and bus, even faster x2 gpus with faster internal storage. Anything you need more than that? Go external.
Lemon Bon Bon.
Apple never mentioned integer performance or geek bench scores...
Three years later, why the f**k not around 40K? I hope it's because it was running the benchmark in 32 bit mode because otherwise it's f**king pathetic.
In 64 bit mode the Mid 2012 Mac Pro jumps to 25K. That's not enough of a bump between 32 bit and 64 bit so I hope that this test MacPro is gimped somehow and we do see benchmarks in the 30K+ range.
<td style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Processor floating point performance</td>
[/TR]
</thead><tfoot>[TR]
<td colspan="4" style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Geekbench 2.4.0 for Windows x86 (64-bit)</td>
<td style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Processor floating point performance</td>
[/TR]
</thead><tfoot>[TR]
<td colspan="4" style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Geekbench 2.4.2 for Windows x86 (64-bit)</td>
<td style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Processor floating point performance</td>
[/TR]
</thead><tfoot>[TR]
<td colspan="4" style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Geekbench 2.4.0 Pro for Mac OS X x86 (64-bit)</td>
<td style="padding-top:6px;padding-right:6px;padding-bottom:6px;padding-left:6px;line-height:1.2em;text-align:left;border-top-width:1px;border-top-style:solid;border-top-color:rgb(221,221,221);vertical-align:top;">Processor floating point performance</td>
Look at your data. There's no way in the world that the Windows machines are over 100 times as fast in memory as the Mac Pro. Clearly, there's a problem with the benchmark. Until that's fixed, there's not much point in comparing machines cross-platform.
Better to rely on apps that have actually been optimized for both platforms. Or, even better, for benchmarks on the apps that you use. If you use Photoshop exclusively, then you should be looking at Photoshop performance - regardless of how well it's optimized. If you run Geekbench all day, then that's the benchmark you should be using.
The only info I could find about the Xeon E5 2697 was from an Intel PDF that came up in a Google search. It referenced an 8-core processor, not a 12-core.
I don't expect it to be cheaper either, but since we know that it's only going to be a single core and not dual core, I'm not surprised that other workstations will beat it, in terms of geekbench scores.
If it isn't cheaper, then that will be sad given that it also the least customizable Mac Pro to-date. And don't give me that out of the box crap -- I've done my homework there too and unless the cost drops considerably it is NOT an option right now. Of course, none of us know WHAT THEY ACTUALLY WILL RELEASE; but, I priced a Dell the other day at just over 5K (with dual current generation XEONs) that has benchmarks similar to the HP that meets my needs WAY better…
I just went to Intel's site and couldn't find the E5-2697 listed anywhere. Is this a processor that's not been announced?
There's a list of upcoming Ivy Bridge EP processors here - some have the same model numbers as Sandy Bridge but with v2 after it (2011 refers to the socket not the year):
The E5-2697 v2 is listed as one of the highest models. I expected them to go with the E5-2687W v2 as it's clocked higher but the 2687W had the same core-count as the others. Perhaps the 2697 will be the only one with 12-cores and the others will have 10.
This doesn't seem right given the mid 2010 Mac Pro comes in at 21980. Perhaps Geekbench tests are not suitable for this machine else this is a new Mac Pro with half its cores shut down!
It may not have been running at maximum performance as Wizard69 mentioned, it could be an early CPU without turbo enabled, which sometimes happens. These CPUs dynamically clock themselves higher or lower. The E5-2687W clocks from 3.1-3.8GHz. It's still a realistic result but with the turbo enabled and 2 CPU generations, I'd expect 50% improvement at the same core count and frequency.
I took it to the Apple Store Monday. OMG what a pig. The current version is just way too big, especially if these need to be moved at all (filming, live audio, etc w/e). I'm no potato chip eater either .
Yeah, it's not a problem until you have to lift it somewhere. I think many IT personnel will appreciate the smaller design.
I do wonder, will we be able to daisy chain Mac Pros via Thunderbolt 2?
I think it should be possible even if only via IP over Thunderbolt. They can be connected over ethernet but I'd like to see Apple-provided special sauce in the OS that allows zero-config and the possibility of processes just automatically being able to scale across machines e.g copying the data and binaries over temporarily and sending the result back.
Three years later, why the f**k not around 40K? I hope it's because it was running the benchmark in 32 bit mode because otherwise it's f**king pathetic.
There is a large difference between 32-bit and 64-bit tests but they wouldn't get 40k with it being a single CPU. What you have to remember is the price.
Previously, they sold 2x $1440 CPUs in the 12-core model at 35% markup so in the $6200 Mac Pro, you'd pay $3888 just for the CPUs for that performance.
Now, you'd be paying say $2000 at 35% markup so for 10% higher performance, you'd pay $2700.
So you pay nearly $1200 less and get 10% more power. If you leave the GPUs at the base spec (dual W5000), you'd probably get the top 12-core for $5000.
As I've said many times, Apple has never used two of the top-end processors so I don't know where the comparison is coming from that they could have put in two 12-cores. Two 12-cores at $2k each would cost $5400 just for the CPUs so that model would cost $7712. It's fair enough to say that getting double the speed for 55% more money is good value but the volume of users that pay that much money is extremely low. What they are doing is giving the best value to the largest audience at the lower price points and for OpenCL tasks, much better value.
If so, why would Apple need to do it online, except to "leak" the machine's performance to the public. However, judging from the unremarkable results, that's not very likely.
They put prototypes out for testing to 3rd parties (Pixar/Foundry), which may explain the model number.
There's another Geekbench test supposedly with two of these E5 CPUs clearly with a false result:
This happens quite often with Geekbench. I'm still confident they'll manage a 50% performance improvement over the last model but as I say, the single CPU will be significantly cheaper than the dual CPU so worst case, you'll get slightly higher performance for a lot less money.
I wonder if the next Mac mini will be 1/8th the size of the current model?
They can probably shrink it in height until it reaches the thickness of the laptops. Moving to PCIe SSD would be a nice move but it comes with cost increases. They could replace dual 2.5" drives with a single 2.5" and a PCIe SSD but on the lowest-end config, leave out the SSD. They'd save some height by soldering in RAM too but would potentially cap the RAM at 8GB and the RAM would be more expensive than 3rd party memory.
There's no need for it to be 1/8th the size. The Mac Pro had such a dramatic saving because it was much larger than it needed to be for the performance. The Mini is a fairly ideal size for its purpose.
There's a list of upcoming Ivy Bridge EP processors here - some have the same model numbers as Sandy Bridge but with v2 after it (2011 refers to the socket not the year):
The E5-2697 v2 is listed as one of the highest models. I expected them to go with the E5-2687W v2 as it's clocked higher but the 2687W had the same core-count as the others. Perhaps the 2697 will be the only one with 12-cores and the others will have 10.
It may not have been running at maximum performance as Wizard69 mentioned, it could be an early CPU without turbo enabled, which sometimes happens. These CPUs dynamically clock themselves higher or lower. The E5-2687W clocks from 3.1-3.8GHz. It's still a realistic result but with the turbo enabled and 2 CPU generations, I'd expect 50% improvement at the same core count and frequency.
You don't get 64-bit benchmarks in the free version:
Yeah, it's not a problem until you have to lift it somewhere. I think many IT personnel will appreciate the smaller design.
I think it should be possible even if only via IP over Thunderbolt. They can be connected over ethernet but I'd like to see Apple-provided special sauce in the OS that allows zero-config and the possibility of processes just automatically being able to scale across machines e.g copying the data and binaries over temporarily and sending the result back.
There is a large difference between 32-bit and 64-bit tests but they wouldn't get 40k with it being a single CPU. What you have to remember is the price.
Previously, they sold 2x $1440 CPUs in the 12-core model at 35% markup so in the $6200 Mac Pro, you'd pay $3888 just for the CPUs for that performance.
Now, you'd be paying say $2000 at 35% markup so for 10% higher performance, you'd pay $2700.
So you pay nearly $1200 less and get 10% more power. If you leave the GPUs at the base spec (dual W5000), you'd probably get the top 12-core for $5000.
As I've said many times, Apple has never used two of the top-end processors so I don't know where the comparison is coming from that they could have put in two 12-cores. Two 12-cores at $2k each would cost $5400 just for the CPUs so that model would cost $7712. It's fair enough to say that getting double the speed for 55% more money is good value but the volume of users that pay that much money is extremely low. What they are doing is giving the best value to the largest audience at the lower price points and for OpenCL tasks, much better value.
They put prototypes out for testing to 3rd parties (Pixar/Foundry), which may explain the model number.
There's another Geekbench test supposedly with two of these E5 CPUs clearly with a false result:
This happens quite often with Geekbench. I'm still confident they'll manage a 50% performance improvement over the last model but as I say, the single CPU will be significantly cheaper than the dual CPU so worst case, you'll get slightly higher performance for a lot less money.
Sorry, but the tradeoff with the Single CPU is we finally get Dual FirePro GPGPUs and together those retail for > $3k. Even if AMD cuts the price by $500 you still have $2500 in GPGPUs now.
This Mac Pro will most certainly cost much more than today's current and underpowered single socket Model at $2500+. Instead, this 12 core without the FirePro would be around the midpoint between the $2500 and $3800 models with old GPGPU technology. In short, $3150 + the new GPGPUs which will put this product well past $5k.
What you will get is continued expansion of power and life of a system because more and more of the OS and Application installed base is moving to leveraging OpenCL and GCD across the board. Adobe has recognized that in their Premiere Pro and CS Suite by working closely with AMD using OpenCL 1.2.
Nvidia doesn't have OpenCL 1.2 drivers. I know. They are crappily stuck on 1.1 partial. Nvidia is pushing a dead end technology: CUDA with version 5.5 that was just released.
Intel is following AMD into the HSA arena [by necessity not choice as the rest of the industry is getting on-board with the HSA 3.0 architecture] and the core tech behind that is OpenCL.
The only product keeping Apple tethered to Intel is LightPeak [Thunderbolt/Lightning] technology. Depending on the legalities of Thunderbolt, eventually Intel is going to have to open it up to allow AMD to license it. When that happens Apple can pick either vendor to suit its needs.
Right now: Apple is stuck with Intel for the CPU and can choose between 3 vendors for GPGPU tech. It wisely chose AMD. Both AMD and Apple are at the forefront of OpenCL commitment. AMD dumped it's own proprietary solution and is betting the company on it.
PS4 is all in with OpenCL. XBox One is a mixed-bag of hurt. AMD jointly working with ARM is a brilliant move on their part. They are going to own the low to mid-tier server market spaces, the console market space, and the commodity APU laptop space. Intel can't compete with them in those arenas.
Apple can wait while it perfects its current offerings and decide which gives them the best value, at the time, to collaborate on future projects. There are no best friends in Silicon Valley.
Comments
I have the same MP and it has never let me down either. These MP's are simply they best, and I have always bought those (PowerMacs, Performa's et cetera before that).
1300W. Wow. What's up whit this '85& efficiency'?
No Energy Star? Will Greenpeace go after Dell as well? edit: it's a $0.00 option you can check.
Windows 8 not mandatory? Good. Can you also choose Linux? edit, nope, didn't find one. What a God awful website they have. The summary for your config uses 6 different colours.
If so, why would Apple need to do it online, except to "leak" the machine's performance to the public. However, judging from the unremarkable results, that's not very likely.
Quote:
Originally Posted by emig647
It's not out yet. Supposedly shipping Q3 2013.
Huh? This is comparing a single 2.7ghz 12 core xeon cpu to the 2010 mac pro with dual 3.07ghz 6 core xeon cpus, and it still beat it by 10%. Not to mention, ~1/8th the size?
It's still comparing a 3-year old top-of-the-line computer with a brand new top of the line computer. Honestly, people with Mac Pros already have enough space that the size won't make a difference; I mean the Mac Pro is already sitting there. A 10% increase in power over 3-year-old tech is not exactly mindblowing. I mean every year a new Macbook is released, Apple calls it "50% more powerful than last year's" and things like that - not "10% more powerful than 3 generations ago".
And I think this trend will continue; the increases will be less and less more than they've been. Many articles on this very subject out there; the tech simply won't increase twofold every year. Not anymore.
Quote:
Originally Posted by ifij775
32-bit metrics? Isn't everything 64-bit any more.
most of math operation within application are done on 32 integers and floats and this is what is important for CPU performance. 64-bit memory addressing is another story.
Quote:
Originally Posted by mdriftmeyer
You folks are delusional if you think Intel's Xeon processors have jumped in performance to trounce the 2010 Mac Pro. They haven't. Intel has hit a wall and is already rearchitecting their CPU systems to follow AMD's path.
I guess we'd expect 'upto' x2 performance like Apple says. Whatever that means...and on what app and on what benchmark etc.
AMD's Opterons look pretty good bang for buck... I guess it's Intel though. And it should bring appreciable performance cpu gains over teh '3 year old' model. (Sniggers derisively at last year's 'upgrade...')
I think the move to gpu computational emphasis with dual GPUs is long over due and definitely in sync with the push to make Open CL adoption all pervasive. It's a smart move. For those task that can take full advantage we can get an order of magnitude improvement in certain benches/apps. Also, dual GPUs can certainly throw around what's in your painting and modelling window very quickly if the Mari/Pixar' demo is anything to go by.
Fast cpu, memory and bus, even faster x2 gpus with faster internal storage. Anything you need more than that? Go external.
Lemon Bon Bon.
September/fall will come soon enough...and the Mac Pro will be upon us...with another surprise at launch?
Who knows...maybe a 4k display?
Lemon Bon Bon.
Apple never mentioned integer performance or geek bench scores...
Look at your data. There's no way in the world that the Windows machines are over 100 times as fast in memory as the Mac Pro. Clearly, there's a problem with the benchmark. Until that's fixed, there's not much point in comparing machines cross-platform.
Better to rely on apps that have actually been optimized for both platforms. Or, even better, for benchmarks on the apps that you use. If you use Photoshop exclusively, then you should be looking at Photoshop performance - regardless of how well it's optimized. If you run Geekbench all day, then that's the benchmark you should be using.
I would think so. They made a (relatively) big deal of the three x 4k monitors at the keynote.
It'll be $$$$
Quote:
Originally Posted by Apple ][
I don't expect it to be cheaper either, but since we know that it's only going to be a single core and not dual core, I'm not surprised that other workstations will beat it, in terms of geekbench scores.
If it isn't cheaper, then that will be sad given that it also the least customizable Mac Pro to-date. And don't give me that out of the box crap -- I've done my homework there too and unless the cost drops considerably it is NOT an option right now. Of course, none of us know WHAT THEY ACTUALLY WILL RELEASE; but, I priced a Dell the other day at just over 5K (with dual current generation XEONs) that has benchmarks similar to the HP that meets my needs WAY better…
There's a list of upcoming Ivy Bridge EP processors here - some have the same model numbers as Sandy Bridge but with v2 after it (2011 refers to the socket not the year):
http://www.techpowerup.com/185624/ivy-bridge-ep-based-xeon-e5-processors-specifications-leaked.html
The E5-2697 v2 is listed as one of the highest models. I expected them to go with the E5-2687W v2 as it's clocked higher but the 2687W had the same core-count as the others. Perhaps the 2697 will be the only one with 12-cores and the others will have 10.
It may not have been running at maximum performance as Wizard69 mentioned, it could be an early CPU without turbo enabled, which sometimes happens. These CPUs dynamically clock themselves higher or lower. The E5-2687W clocks from 3.1-3.8GHz. It's still a realistic result but with the turbo enabled and 2 CPU generations, I'd expect 50% improvement at the same core count and frequency.
You don't get 64-bit benchmarks in the free version:
http://www.primatelabs.com/geekbench/editions/
Yeah, it's not a problem until you have to lift it somewhere. I think many IT personnel will appreciate the smaller design.
I think it should be possible even if only via IP over Thunderbolt. They can be connected over ethernet but I'd like to see Apple-provided special sauce in the OS that allows zero-config and the possibility of processes just automatically being able to scale across machines e.g copying the data and binaries over temporarily and sending the result back.
There is a large difference between 32-bit and 64-bit tests but they wouldn't get 40k with it being a single CPU. What you have to remember is the price.
Previously, they sold 2x $1440 CPUs in the 12-core model at 35% markup so in the $6200 Mac Pro, you'd pay $3888 just for the CPUs for that performance.
Now, you'd be paying say $2000 at 35% markup so for 10% higher performance, you'd pay $2700.
So you pay nearly $1200 less and get 10% more power. If you leave the GPUs at the base spec (dual W5000), you'd probably get the top 12-core for $5000.
As I've said many times, Apple has never used two of the top-end processors so I don't know where the comparison is coming from that they could have put in two 12-cores. Two 12-cores at $2k each would cost $5400 just for the CPUs so that model would cost $7712. It's fair enough to say that getting double the speed for 55% more money is good value but the volume of users that pay that much money is extremely low. What they are doing is giving the best value to the largest audience at the lower price points and for OpenCL tasks, much better value.
They put prototypes out for testing to 3rd parties (Pixar/Foundry), which may explain the model number.
There's another Geekbench test supposedly with two of these E5 CPUs clearly with a false result:
http://browser.primatelabs.com/geekbench2/2057228
This happens quite often with Geekbench. I'm still confident they'll manage a 50% performance improvement over the last model but as I say, the single CPU will be significantly cheaper than the dual CPU so worst case, you'll get slightly higher performance for a lot less money.
I wonder how long till Apple adds a 6-core or 4-core configuration to the Mac Pro line up?
They can probably shrink it in height until it reaches the thickness of the laptops. Moving to PCIe SSD would be a nice move but it comes with cost increases. They could replace dual 2.5" drives with a single 2.5" and a PCIe SSD but on the lowest-end config, leave out the SSD. They'd save some height by soldering in RAM too but would potentially cap the RAM at 8GB and the RAM would be more expensive than 3rd party memory.
There's no need for it to be 1/8th the size. The Mac Pro had such a dramatic saving because it was much larger than it needed to be for the performance. The Mini is a fairly ideal size for its purpose.
Quote:
Originally Posted by Marvin
There's a list of upcoming Ivy Bridge EP processors here - some have the same model numbers as Sandy Bridge but with v2 after it (2011 refers to the socket not the year):
http://www.techpowerup.com/185624/ivy-bridge-ep-based-xeon-e5-processors-specifications-leaked.html
The E5-2697 v2 is listed as one of the highest models. I expected them to go with the E5-2687W v2 as it's clocked higher but the 2687W had the same core-count as the others. Perhaps the 2697 will be the only one with 12-cores and the others will have 10.
It may not have been running at maximum performance as Wizard69 mentioned, it could be an early CPU without turbo enabled, which sometimes happens. These CPUs dynamically clock themselves higher or lower. The E5-2687W clocks from 3.1-3.8GHz. It's still a realistic result but with the turbo enabled and 2 CPU generations, I'd expect 50% improvement at the same core count and frequency.
You don't get 64-bit benchmarks in the free version:
http://www.primatelabs.com/geekbench/editions/
Yeah, it's not a problem until you have to lift it somewhere. I think many IT personnel will appreciate the smaller design.
I think it should be possible even if only via IP over Thunderbolt. They can be connected over ethernet but I'd like to see Apple-provided special sauce in the OS that allows zero-config and the possibility of processes just automatically being able to scale across machines e.g copying the data and binaries over temporarily and sending the result back.
There is a large difference between 32-bit and 64-bit tests but they wouldn't get 40k with it being a single CPU. What you have to remember is the price.
Previously, they sold 2x $1440 CPUs in the 12-core model at 35% markup so in the $6200 Mac Pro, you'd pay $3888 just for the CPUs for that performance.
Now, you'd be paying say $2000 at 35% markup so for 10% higher performance, you'd pay $2700.
So you pay nearly $1200 less and get 10% more power. If you leave the GPUs at the base spec (dual W5000), you'd probably get the top 12-core for $5000.
As I've said many times, Apple has never used two of the top-end processors so I don't know where the comparison is coming from that they could have put in two 12-cores. Two 12-cores at $2k each would cost $5400 just for the CPUs so that model would cost $7712. It's fair enough to say that getting double the speed for 55% more money is good value but the volume of users that pay that much money is extremely low. What they are doing is giving the best value to the largest audience at the lower price points and for OpenCL tasks, much better value.
They put prototypes out for testing to 3rd parties (Pixar/Foundry), which may explain the model number.
There's another Geekbench test supposedly with two of these E5 CPUs clearly with a false result:
http://browser.primatelabs.com/geekbench2/2057228
This happens quite often with Geekbench. I'm still confident they'll manage a 50% performance improvement over the last model but as I say, the single CPU will be significantly cheaper than the dual CPU so worst case, you'll get slightly higher performance for a lot less money.
Sorry, but the tradeoff with the Single CPU is we finally get Dual FirePro GPGPUs and together those retail for > $3k. Even if AMD cuts the price by $500 you still have $2500 in GPGPUs now.
This Mac Pro will most certainly cost much more than today's current and underpowered single socket Model at $2500+. Instead, this 12 core without the FirePro would be around the midpoint between the $2500 and $3800 models with old GPGPU technology. In short, $3150 + the new GPGPUs which will put this product well past $5k.
What you will get is continued expansion of power and life of a system because more and more of the OS and Application installed base is moving to leveraging OpenCL and GCD across the board. Adobe has recognized that in their Premiere Pro and CS Suite by working closely with AMD using OpenCL 1.2.
Nvidia doesn't have OpenCL 1.2 drivers. I know. They are crappily stuck on 1.1 partial. Nvidia is pushing a dead end technology: CUDA with version 5.5 that was just released.
Intel is following AMD into the HSA arena [by necessity not choice as the rest of the industry is getting on-board with the HSA 3.0 architecture] and the core tech behind that is OpenCL.
The only product keeping Apple tethered to Intel is LightPeak [Thunderbolt/Lightning] technology. Depending on the legalities of Thunderbolt, eventually Intel is going to have to open it up to allow AMD to license it. When that happens Apple can pick either vendor to suit its needs.
Right now: Apple is stuck with Intel for the CPU and can choose between 3 vendors for GPGPU tech. It wisely chose AMD. Both AMD and Apple are at the forefront of OpenCL commitment. AMD dumped it's own proprietary solution and is betting the company on it.
PS4 is all in with OpenCL. XBox One is a mixed-bag of hurt. AMD jointly working with ARM is a brilliant move on their part. They are going to own the low to mid-tier server market spaces, the console market space, and the commodity APU laptop space. Intel can't compete with them in those arenas.
Apple can wait while it perfects its current offerings and decide which gives them the best value, at the time, to collaborate on future projects. There are no best friends in Silicon Valley.