MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)

[singhs.apps]

The learning curve is a steep one but a lot of that is because Apple has handled so much of that for us. I am amazed at how many things are accessible through Properties

That's where OSX shines..

For psd and other adobe thumbnails, you might want to check :
http://www.ardfry.com/psd-codec/

Another one :
http://www.fastpictureviewer.com/codecs/

 

[fuchsdh]

The learning curve is a steep one but a lot of that is because Apple has handled so much of that for us. I am amazed at how many things are accessible through Properties.

There are some frustrations such as Keywords (tags) not being displayed or searchable in Explorer for PSD and PDF files. If you save images as TIF instead of PSD, then you see the tags (and search for them) and there is, from what I understand, a third party add-on that will display tags for PDF in Explorer. Also, there is no image preview for PSD or TIF files. Another oddity is that Adobe Bridge also appears to use the Windows search engine so no searching by tags with PSD there either but it does display previews. Easiest answer is to use Lightroom which does it all itself. None of these are deal breakers and I can't say that I really miss my Mac to any real degree.
[doublepost=1485116739][/doublepost]
The other favorite light up toy is the keyboard. OMG, I had a hard time finding a mechanical keyboard that didn't have lights that made it look like the Fourth of July. Sometimes, if you didn't know better, you'd swear that users just want to play games on PC's based on all the lit computers and keyboards.

If I could have gotten my Das Keyboard with a backlight, I woulda'. As is, since it's long-past warranty, next time I break it open to clean it I think I'll do a quick-and-dirty run of LED strip lights in the cavity of the thing and then pipe the power through the USB (not handy with soldering so I'll just run another cable out the back.)

 

[Stacc]

The problem with Apple's simple aluminium vs chip solution is that the heat can only dissapate as fast as the aluminium physically allows. Eg you have maybe 10 square centimeters of chip surface that actually touches the aluminium. Due to physics you will have much higher local temps there than in the fins that help dissipate the heat eventually. Hot spots are a problem for chips, one that gets solved very well by heatpipes as you essentially achieve a higher volume of cooling medium (in this case metal and the coolant/gas in the heatpipes, vs. just the touching aluminium) vs the same die surface.

If Apple had designed the thermal core with heatpipes to distribute heat quicker across a similar volume of metal, they would have at least achieved better thermal balance for when only the cpu or say one gpu are under load and if they had made the core mushroom shaped with a big cooling head on top with a fan extracting air through it, they might have creased the total thermal potential effectively without changing the dimensions of the case much, maybe increasing its height by an inch.

Heatpipes allow you to determine where you want to get rid of the excess heat. Instead Apple chose where they wanted the heat to be produced. The result is a physically very limiting if clever design that is 100% dependant on the thermal conductivity of aluminium and the volume of metal used in the triangular thermal core.

It's a pricey way of solving a problem and you tie yourself in to a very inflexible platform. And that is what I think lies at the heart of this 4 year wait for a new nMP. They designed a new case for the next ten years and then failed to find another configuration of gpus and cpus and peripherals that they could actually cram into the existing limits of the new design.

Sure, but your assumption is that the limitation is transferring the heat from the chip to the heat spreader. The limitation is the size and rotational speed of the fan. Even if you transfer the heat much more efficiently from the chips to the heat spreader, convection is still needed to remove it from the case. In all your cases of improving the heat soak of the cooler still requires a fan to be larger/rotate faster to dissipate that heat.

I think the argument we are really having is that we disagree with the power/thermal constraints apple chose. If the mac pro could consume 800 W, obviously a much different cooling system would be needed. One that would likely increase the size of the chassis and perhaps could include heat pipes.

If we imagine Apple sat down in 2010 and said, "we want the fastest, quietest, and smallest (if possible) computer we can get within a 450 W power envelope that can drive a thunderbolt display," the current mac pro is a pretty good realization of that. You get two mid sized GPUs that offer more compute power than the fastest single GPU at the time and it has the benefit of being easier to cool due to the fact that the heat generated by the GPUs is better distributed about the thermal core.

 

[G-News]

The fan and its speed only accelerate the effect thermal convection has naturally. They don't change the local hotspots, even if a higher air flow will likely reduce the temperatures overall.

Low noise is achieved by keeping air pressure and fan speed low. The bigger the fan is, the more air it can move while rotating slower, thus keeping air pressure low. There are additional tricks like asymetric fan blades, blade pitch etc. that affect air flow and noise levels. In that regard Apple did a great job of having just one big fan handle it all.

I guessit boils down to what you said. Apple picked a number and went by that. Sadly they picked 450W and 8L of volume, which has proven to have rather short legs compared to pretty much any other design not by Apple and including their own previous Mac Pro, which had/s very, very long legs in comparison.

 

[Mago]

A more efficient heat transfer dont need more fan speed, you'll notice in an hypothetical 800W nMP the fan fill rotate the same speed but the air will be heated twice.

With heatpipes/peltier elements its plausible to cool down with the same fan speed even far beyond 800W.

also the nMP 'vertical chimmny' like design eases to transfer 2x heat to the environment.

 

[G-News]

One thing we haven't touched upon yet is how little surface area the thermal core provides. Because airflow can't be too restricted and because it's extruded from a single piece, the fins are few in number and quite thick. A heatpipe and mushroom top design could have a much higher surface area heatsink below or above the fan, with finer fins and more numerous allowing for more heat extraction by air.

Looking at my 2000W ceramic room heater, the actual heater unit is not much bigger than the palm of my hand and not much thicker either, yet manages to dissipate nearly 2000W of heat. But the structure is very dense and has a large surface area.

Apple likes these chunky but inefficient aluminium heatsinks, the PowerMac G4 MDD had them too.

 

[JimGoshorn]

For psd and other adobe thumbnails, you might want to check :
http://www.ardfry.com/psd-codec/

Thanks, I'll have a look
[doublepost=1485123540][/doublepost]

If I could have gotten my Das Keyboard with a backlight, I woulda'.

Have a look at the Code keyboard by WASD:

https://codekeyboards.com/

Backlit and available with cherry switches. I got a full keyboard with the cherry clear switches and love it.

 

[AidenShaw]

If the mac pro could consume 800 W, obviously a much different cooling system would be needed.

I wonder how much would be gained by making the thermal core out of copper....

 

[JesperA]

Sorry for the sidetracking guys, this is the last post about this, i promise, i just thought it was incredible weird that someone claim that something does not exist even though.....yeah....it obviously does...and even showing evidence of the existens of that "something"; the person still insist that it does not exist, what is that about?

you may find few server racks with on-chip liquid cooling (as i mentioned in NY are more probable), but datacenters use forced air with liquid cooling not on chip, bt at the bottom or side of the racks to avoid leak damage and easy blade inspection/removal..

diy you check new Cray XC line ? http://www.cray.com/sites/default/files/Cray-XC50-Product-Brief.pdf

Yeah, you are a weird dude, i showed you an offering from Cray with on chip water cooling, then you for some reason refuse to believe Crays own information (they should know, they manufacture their on-chip liquid cooling offerings), i listed several other offerings with on-chip liquid cooling, you still refute it and claim no supercomputer uses liquid cooling even though there is several liquid cooled supercomputers on the Top500 list.

https://www.top500.org/news/aquila-takes-wraps-off-liquid-cooled-hpc-server/
http://www.asetek.com/press-room/ne...stallations-in-the-top500-supercomputer-list/
https://www.hpcwire.com/off-the-wire/rsc-tornado-72-core-intel-phi-achieves-1-41-pflops-per-rack/

The last one even breaking the world record for power density for supercomputers with its on-chip liquid cooling (which you time and time claim such a system does not exist...)

You can moan all your want about liquid cooled supercomputers does not exist, if you have a fetish for false information then fine, if that makes you happy. In the real world though, liquid cooled on-chip servers, datacenters and supercomputers exist and are fairly common, end of discussion, pointless to discuss this further, I'm going back to the Mac Pro...

 

[fuchsdh]

Thanks, I'll have a look
[doublepost=1485123540][/doublepost]
Have a look at the Code keyboard by WASD:

https://codekeyboards.com/

Backlit and available with cherry switches. I got a full keyboard with the cherry clear switches and love it.

Eh, no sense in buying more junk when what I've got is serviceable. I mean, that's why I'm still on my cMP—until Adobe gets off its ass and optimizes its software better I've got diminishing returns to upgrade to anything, Mac or PC.

 

[ManuelGomes]

August for the -X, -W might come soon after:
http://wccftech.com/intel-skylake-x-kaby-lake-x-core-i7-august-2017-launch/
Still convinced, I see the new nMP being announced by mid year and coming as Late 2017 nMP.

 

[H2SO4]

I wonder how much would be gained by making the thermal core out of copper....

LOTS, (of money for Apple).

 

[ETN3]

Anyone know what these 22 core MacPro 6,1 Geekbench results are? Hackintosh, Fake or the real deal?

http://browser.primatelabs.com/v4/cpu/1357458


Single-Core Score Multi-Core Score
4279 38395
Geekbench 4.0.3 Tryout for Mac OS X x86 (64-bit)
Result Information
Upload Date December 17 2016 11:28 AM
Views 60
System Information
MacPro6,1
Operating System macOS 10.12.2 (Build 16C67)
Model MacPro6,1
Processor Intel Xeon E5-2696 v4 @ 2.20 GHz
1 processor, 22 cores, 44 threads
Processor ID GenuineIntel Family 6 Model 79 Stepping 1
Processor Codename
Processor Package
L1 Instruction Cache 32 KB x 22
L1 Data Cache 32 KB x 22
L2 Cache 256 KB x 22
L3 Cache 56320 KB
Motherboard Apple Inc. Mac-F60DEB81FF30ACF6 MacPro6,1
Northbridge
Southbridge
BIOS Apple Inc. MP61.88Z.0116.B04.1312061508
Memory 131072 MB 2400 MHz DDR4

More results here: http://browser.primatelabs.com/v4/cpu/search?dir=desc&q=macpro6,1&sort=multicore_score

 

[Stacc]

Anyone know what these 22 core MacPro 6,1 Geekbench results are? Hackintosh, Fake or the real deal?

http://browser.primatelabs.com/v4/cpu/1357458


Single-Core Score Multi-Core Score
4279 38395
Geekbench 4.0.3 Tryout for Mac OS X x86 (64-bit)
Result Information
Upload Date December 17 2016 11:28 AM
Views 60
System Information
MacPro6,1
Operating System macOS 10.12.2 (Build 16C67)
Model MacPro6,1
Processor Intel Xeon E5-2696 v4 @ 2.20 GHz
1 processor, 22 cores, 44 threads
Processor ID GenuineIntel Family 6 Model 79 Stepping 1
Processor Codename
Processor Package
L1 Instruction Cache 32 KB x 22
L1 Data Cache 32 KB x 22
L2 Cache 256 KB x 22
L3 Cache 56320 KB
Motherboard Apple Inc. Mac-F60DEB81FF30ACF6 MacPro6,1
Northbridge
Southbridge
BIOS Apple Inc. MP61.88Z.0116.B04.1312061508
Memory 131072 MB 2400 MHz DDR4

More results here: http://browser.primatelabs.com/v4/cpu/search?dir=desc&q=macpro6,1&sort=multicore_score

Hackintosh. Mac identifiers are easy to spoof. You can tell this is a hack because while it identifies as the existing mac pro, it uses Intel's newest Broadwell based Xeons. Once in awhile an unreleased mac will pop up in geekbench results, but they can be hard to identify because they are so easy to spoof.

 

[ETN3]

Hackintosh. Mac identifiers are easy to spoof. You can tell this is a hack because while it identifies as the existing mac pro, it uses Intel's newest Broadwell based Xeons. Once in awhile an unreleased mac will pop up in geekbench results, but they can be hard to identify because they are so easy to spoof.

Yes, but is it possible to spoof the prosessor and operation system? Is it possible to run a 22 core E5 V4 on a Hackintosh? Because otherwise it might be Apple spoofing the version number of the machine to not draw attention. I know they have been busted on Geekbench before. Plausible or a long shot?

 

[Stacc]

Yes, but is it possible to spoof the prosessor and operation system? Is it possible to run a 22 core E5 V4 on a Hackintosh? Because otherwise it might be Apple spoofing the version number of the machine to not draw attention. I know they have been busted on Geekbench before. Plausible or a long shot?

Nope, don't get your hopes up. Xeon based hackintoshes tend to be more difficult, but are certainly possible. For example here is a broadwell based Xeon hackintosh.

 

[Jack Burton]

don't get your hopes up

For anyone just now looking at this thread, this sums up everything, LOL

 

[Stacc]

Because I think about this stuff too much, I made a fun graph to show the relative compute efficiency of some recent GPUs.

First, remember that TDP is somewhat arbitrary, so its not quite an apples to apples comparison. Unsurprisingly GP104/Nvidia GTX 1080 finishes first here. I was most surprised that Polaris 11 makes a fairly strong case for being the choice in the 15" MacBook pro. It is much more efficient than GP107, which was only released a couple weeks ago.

Given the weakness of Polaris 10, I hope Apple uses GP104 in the next mac pro or if they plan on using Vega it will have to be a fairly significant jump in efficiency to be competitive.

 

[PhiLLoW]

My 12-core D700 1TB Mac Pro idles at 22C and rarely goes past 30C under sustained loads.
But, I built a wind tunnel shelf, in addition to steps mentioned above.

My Mac Pro setup is ranked #1 for single-core & #3 for multi-core performance for 12-core Mac Pro 6,1.
See my Geekbench scores: Online Benchmark Results (user dread64)

I really want to see how this wind tunnel and your setup looks like.

 

[Daniel Reed]

I really want to see how this wind tunnel and your setup looks like.

heh, right now its rather rough and not the prettiest thing to look at.

The 30C temp quote was for CPU only loads.
CPU+GPU loads currently doesn't seem to pass 35C

The wind tunnel shelf is in prototyping stage as I tweak it for better performance, looks, and functionality.
Think lots of plywood, cardboard and gaffers tape

I have a couple ideas I plan to try out that should further drop the temps.
And provide a better way to access the nMP inputs.

Once I settle on the design i'll have it built by a carpenter/contractor guy I work with.

I'll most likely add links to it in my signature - it'll be at least a few months w/ my existing workloads.

Another idea is to restore the vertical look by taking off the aluminum tube and placing it inside a larger diameter clear acrylic vertical wind tunnel resting on a pedestal with a hole cut out on the bottom. This design would have a large fan at both the top and bottom and connected to a fan speed control dial

as an aside, has anyone tried just removing the outer tube and point two or 3 side fans at the nMP?
[doublepost=1485286999][/doublepost]

Anyone know what these 22 core MacPro 6,1 Geekbench results are? Hackintosh, Fake or the real deal?

http://browser.primatelabs.com/v4/cpu/1357458


Single-Core Score Multi-Core Score
4279 38395
Geekbench 4.0.3 Tryout for Mac OS X x86 (64-bit)
Result Information
Upload Date December 17 2016 11:28 AM
Views 60
System Information
MacPro6,1
Operating System macOS 10.12.2 (Build 16C67)
Model MacPro6,1
Processor Intel Xeon E5-2696 v4 @ 2.20 GHz
1 processor, 22 cores, 44 threads
Processor ID GenuineIntel Family 6 Model 79 Stepping 1
Processor Codename
Processor Package
L1 Instruction Cache 32 KB x 22
L1 Data Cache 32 KB x 22
L2 Cache 256 KB x 22
L3 Cache 56320 KB
Motherboard Apple Inc. Mac-F60DEB81FF30ACF6 MacPro6,1
Northbridge
Southbridge
BIOS Apple Inc. MP61.88Z.0116.B04.1312061508
Memory 131072 MB 2400 MHz DDR4

More results here: http://browser.primatelabs.com/v4/cpu/search?dir=desc&q=macpro6,1&sort=multicore_score

i suspect they're Hackintoshes

 

[goMac]

First, remember that TDP is somewhat arbitrary, so its not quite an apples to apples comparison. Unsurprisingly GP104/Nvidia GTX 1080 finishes first here. I was most surprised that Polaris 11 makes a fairly strong case for being the choice in the 15" MacBook pro. It is much more efficient than GP107, which was only released a couple weeks ago.

Given the weakness of Polaris 10, I hope Apple uses GP104 in the next mac pro or if they plan on using Vega it will have to be a fairly significant jump in efficiency to be competitive.

Polaris 10 being an absolute efficiency failure, to me, seems like the reason Apple is probably moving on to Vega. I can't imagine Polaris 10 being a good fit for either the iMac or the Mac Pro. Totally didn't live up to it's promises.

 

[iMcLovin]

The learning curve is a steep one but a lot of that is because Apple has handled so much of that for us. I am amazed at how many things are accessible through Properties.

There are some frustrations such as Keywords (tags) not being displayed or searchable in Explorer for PSD and PDF files. If you save images as TIF instead of PSD, then you see the tags (and search for them) and there is, from what I understand, a third party add-on that will display tags for PDF in Explorer. Also, there is no image preview for PSD or TIF files. Another oddity is that Adobe Bridge also appears to use the Windows search engine so no searching by tags with PSD there either but it does display previews. Easiest answer is to use Lightroom which does it all itself. None of these are deal breakers and I can't say that I really miss my Mac to any real degree.
[doublepost=1485116739][/doublepost]
The other favorite light up toy is the keyboard. OMG, I had a hard time finding a mechanical keyboard that didn't have lights that made it look like the Fourth of July. Sometimes, if you didn't know better, you'd swear that users just want to play games on PC's based on all the lit computers and keyboards.

Im with you, nothin in particular I miss from Mac. But I've gotten so much more with my PC.
But Quick View and Thumbnails are something that was important to me. For all kinds of files.

I use MysticThumbs, it gives me preview for all kinds of file formats. Such as all kind of image files with or without alpha if you want. For psd, psb, pdf, tga, ai, eps, etc etc etc....I recommend it. THe UI is ugly but it does the job perfectly.

Also for quick view I use Seer. Its not as slick as the integrated one in OSX but in many ways its better after Apple decided to make theirs support less codecs. In Seer I can play any video Ive tested so far.

 

[Stacc]

Polaris 10 being an absolute efficiency failure, to me, seems like the reason Apple is probably moving on to Vega. I can't imagine Polaris 10 being a good fit for either the iMac or the Mac Pro. Totally didn't live up to it's promises.

I agree. Its too bad Apple has moved away from Nvidia. Sticking the notebook versions of the GTX 1070/1080 into the iMac and mac pro would make for some killer machines.

 

[koyoot]

Radeon Pro WX 5100. Polaris 10LE. 1792 GCN cores. 1.086 GHz core clock. 75W TDP.

51 GFLOPs/watt. Is it really a failure in efficiency?

And yes, it is 75W GPU, and yes, it has 1.086 GHz. I have bought that GPU for tests, for my companies purposes. Polaris was optimized for lower power, thats where it gets tremendous perf/watt. Increasing the core clocks puts the architecture out of its comfort zone, end results in pretty bad efficiency, compared to for example GTX 1080.

End-off top.

 

[Stacc]

Radeon Pro WX 5100. Polaris 10LE. 1792 GCN cores. 1.086 GHz core clock. 75W TDP.

51 GFLOPs/watt. Is it really a failure in efficiency?

And yes, it is 75W GPU, and yes, it has 1.086 GHz. I have bought that GPU for tests, for my companies purposes. Polaris was optimized for lower power, thats where it gets tremendous perf/watt. Increasing the core clocks puts the architecture out of its comfort zone, end results in pretty bad efficiency, compared to for example GTX 1080.

End-off top.

Huh, interesting. If Vega can maintain that efficiency and fill out the high end of AMD's lineup then it too would make a killer mac pro. Its a shame its taking them so long...

If we trust the rumors, Vega is rumored to be 12 TFLOPS at 225 W. This puts it at 53 GFLOPS/W.