My nMP Review (vs my old 4,1 Mac Pro)

[sirio76]

It's not certified yet in Windows Professional apps. This is what makes it a "workstation" card. It may happen eventually, it may not.

As far as the benchmarks -- I'm sure they'll turn up eventually

So once again there's no real fact to support your speculation. Otherwise, please show us some source.
I'm not saying you are wrong, but it's always better to have some accurate information before post something.

 

[slughead]

So once again there's no real fact to support your speculation. Otherwise, please show us some source.
I'm not saying you are wrong, but it's always better to have some accurate information before post something.

It's not speculation... The Dx00 series haven't been certified in those apps yet.

As far as the performance, I never said one way or another how they would behave. I'm skeptical, but I am waiting for the benchmarks.

 

[sirio76]

The Dx00 series haven't been certified in those apps yet.

My question is simple, and require a simple answer. Where I can read about this fact?

 

[wheelhot]

So once again there's no real fact to support your speculation. Otherwise, please show us some source.
I'm not saying you are wrong, but it's always better to have some accurate information before post something.

Well if only somebody would run the test that....I dunno....specviewperf? That I've been mentioning repeatedly in over different forum topics

P.S: It even has its own topic called for those nMP who installed Windows

 

[sirio76]

Well if only somebody would run the test that....I dunno....specviewperf? That I've been mentioning repeatedly in over different forum topics

P.S: It even has its own topic called for those nMP who installed Windows

So, some more speculations. I promise that when I'll get my nMP I'll run that test just for you

 

[Anim]

So, some more speculations. I promise that when I'll get my nMP I'll run that test just for you

Same here. After windows 8.1 it will be the first app I install and run. Unless there is good reason nobody is running it yet

 

[JordanNZ]

I don't know why it's labeled like that. I too am waiting for these results.

Before units started shipping, there were tons of posts talking about how great a deal this was because it was a "FirePro"... now we're asking for proof and there's nothing but crickets.

FireGL Drivers on the windows side. √

And the fact that the cards are capable of outputting 10-bit colour finally rules out consumer cards/drivers.

http://www.digitalartsonline.co.uk/...re/apple-mac-pro-2013-review-with-benchmarks/

 

[wheelhot]

So, some more speculations. I promise that when I'll get my nMP I'll run that test just for you

Same here. After windows 8.1 it will be the first app I install and run. Unless there is good reason nobody is running it yet

Thanks Guys!

 

[kazmac]

Not a MP user but

VirtualRain your review was one of the most comprehensive, excellent computer reviews I've ever read.

Reading it made me wish I stayed with design and the like so I could justify getting one.

Thank you for such an excellent read.

 

[ZnU]

Some interesting benchmarks of DaVinci Resolve here. Notable because Resolve uses OpenCL and dual GPUs (although it's not clear if it was set to use the display GPU for additional processing here). Numbers from a 12 core 2010 Mac Pro with a GTX 780 are provided for reference. My read is that the 6C/D500 test system gets slightly edged out by the 2010 for CPU-limited tasks (though it's impressively close given that the 2013 has half the number of cores), and comes out ahead on GPU-based image processing tasks. Pretty respectable showing for the D500.

 

[slughead]

FireGL Drivers on the windows side. √

And the fact that the cards are capable of outputting 10-bit colour finally rules out consumer cards/drivers.

http://www.digitalartsonline.co.uk/...re/apple-mac-pro-2013-review-with-benchmarks/

Nope. Apparently Radeons support 10-bit color, it just happens to be disabled in the drivers. (here is a hack to enable it). I'm reading this thread about a few people who have done it on some older Radeons, they took screenshots and everything.

Makes sense, they are using the firepro drivers after all--but will they get the firepro performance benefits?

----------

My question is simple, and require a simple answer. Where I can read about this fact?

Well you're welcome to look them up one by one. Here is the autodesk list of certified cards, not there, not even in Maya in OS X or windows.

I'm not looking up every one, but I would doubt any of these major professional Windows apps have it.

 

[JordanNZ]

Nope. Apparently Radeons support 10-bit color, it just happens to be disabled in the drivers. (here is a hack to enable it). I'm reading this thread about a few people who have done it on some older Radeons, they took screenshots and everything.

Makes sense, they are using the firepro drivers after all--but will they get the firepro performance benefits?

Considering the performance benefits come from the cards being able to use the FireGL drivers, you'd hope so.

 

[JordanNZ]

Well you're welcome to look them up one by one. Here is the autodesk list of certified cards, not there, not even in Maya in OS X or windows.

I'm not looking up every one, but I would doubt any of these major professional Windows apps have it.

From the FAQ

"12. What hardware and monitor configurations does Autodesk use for testing?
Unless otherwise noted, Autodesk hardware certification tests are run on systems containing a single video card with a single monitor attached. Autodesk does not currently run certification tests on systems with multiple graphics cards installed or multiple monitors."

 

[slughead]

Considering the performance benefits come from the cards being able to use the FireGL drivers, you'd hope so.

It'd take me a while to find again, but attempts at soft-modding 7970's to us FirePro drivers did not yield improved results in SpecPerf. There are performance benefits disabled on the hardware level with Radeons which give the FirePros an advantage.

The mere fact that they are identified as FirePro and using the drivers does not make them run like FirePro cards. We already know they're using Crossfire X instead of Pro--what other differences are there?

 

[JordanNZ]

It'd take me a while to find again, but attempts at soft-modding 7970's to us FirePro drivers did not yield improved results in SpecPerf. There are performance benefits disabled on the hardware level with Radeons which give the FirePros an advantage.

The mere fact that they are identified as FirePro and using the drivers does not make them run like FirePro cards. We already know they're using Crossfire X instead of Pro--what other differences are there?

Hopefully we will find out in a week or so, when everyone's machines start arriving.

By the way, the hack to enable 10-bit support was for the mobility series cards only. No one got it to work on Desktop cards.

 

[VirtualRain]

Can you try out my monitoring app, XRG, to see if that will show the temperatures for you?

http://www.gauchosoft.com/xrg/

Thanks for the nice write-up. Can't wait for my own nMP to be delivered…

Hey, I tried out your monitoring app, it looks like it's functional... (unlike iStat Menus)

I've attached a couple of screenshots.

The first one is while running Geekbench Stress Test and the other is after a couple of Heaven Extreme loops.

Observations:

CPU: there are two "CPU Proximity" temps and one is about 5-deg higher than the other. It's not clear what these are actually referring to.

GPU: The GPU Die 0 seems to be tied to the GPU in slot 1 which I think is dedicated to compute. The GPU Die 1 seems to be the GPU in slot 2 which does display/OpenGL duty. As you might expect, the GPU doing display duty is warmer than the idle GPU during regular usage. GPU2 climbed considerably during the Heaven benchmark, but as you can see, the Heaven benchmark appears to be reporting the idle GPU temperature... so the Heaven benchmark temperature reporting is wrong. I can get one or the other GPU temps to climb more than the other selecting different GPUs in LuxMark.

Disk: The disk indicator in the first screen shot captured a Blackmagic benchmark.

Other Notes:

- Note that due to the nature of the heat sink, temps for the CPU and both GPUs will rise in tandem, even if only one of them is working hard.

- The temps seem to fall as quickly as they rise, which means once the load is removed from the CPU or GPU it cools quickly.

 

[sirio76]

Well you're welcome to look them up one by one. Here is the autodesk list of certified cards, not there, not even in Maya in OS X or windows.

I'm not looking up every one, but I would doubt any of these major professional Windows apps have it.

This list is not updated and the latest AMD and Nvidia cards are simply missing, just like the latest version of OSX(but I'm sure many pro are succesfully using this with or without Autodesk certification). About pc apps, why would Autodesk reccomend an Apple specific hardware for Windows tasks? This will never happen.
Anyway we are "Mac" people on a Mac forum, most of us use the software almost exclusively in OSX, so the Windows drivers things is much overrated IMO.
I'm personally interested in how those cards perform in OSX, and they look very good in both benchmarks and users experience. I do not care if the heaven benchmark score is the slowest on earth as long as the Cinema4D OpenGl score is quite good, because C4D is part of my workflow and I make money with it, while I'm sure nobody is making money with heaven benchmark

 

[ozbimmer]

Nope. Apparently Radeons support 10-bit color, it just happens to be disabled in the drivers. (here is a hack to enable it). I'm reading this thread about a few people who have done it on some older Radeons, they took screenshots and everything.

Makes sense, they are using the firepro drivers after all--but will they get the firepro performance benefits?

----------



Well you're welcome to look them up one by one. Here is the autodesk list of certified cards, not there, not even in Maya in OS X or windows.

I'm not looking up every one, but I would doubt any of these major professional Windows apps have it.

I can't find the screenshot that demonstrates 10-bit output on screen. Also apparently the hack doesn't not work for some...

 

[vhp3]

Imac Benchmark

Probably not much. But its about a 5 mins test. I did my system (mp 1,1) and wanted to compare. Plus there are some numbers from other reviews (digital arts) that show 8 core and D700 to compare against the 6 and D500. I would be curious to see the 680 there too. Mostly because GPU tests seem so inconsistent, I think you need several to get a good picture.

The nmp 8 core scores 1225 CPU, and D700 scores 87.74 FPS GPU

I'm gonna guess the nmp 6 core scores 830 CPU and 66.0 FPS GPU.

I'm not sure if this helps with any discussion here, but I am also thinking about getting a nMP. My choice would be 6 core / D700. Here is a benchmark from my late 2013 iMac Fully maxed out.

 

[wheelhot]

I'm not sure if this helps with any discussion here, but I am also thinking about getting a nMP. My choice would be 6 core / D700. Here is a benchmark from my late 2013 iMac Fully maxed out.

Wow, your iMac is relatively new, any particular reason why you feel the need to upgrade to a nMP top spec?

 

[Gav Mack]

It'd take me a while to find again, but attempts at soft-modding 7970's to us FirePro drivers did not yield improved results in SpecPerf. There are performance benefits disabled on the hardware level with Radeons which give the FirePros an advantage.

The mere fact that they are identified as FirePro and using the drivers does not make them run like FirePro cards. We already know they're using Crossfire X instead of Pro--what other differences are there?

Here's the inf file for the FirePro V nMP drivers if you want to take a look..

 

[VirtualRain]

I mentioned in my original post that my CF card reader was taking awhile to mount when connected to the nMP (like 10-15 seconds). Well, I found out there was a firmware update for my Transcend USB3 card reader, and that seems to have cured it. It now mounts instantly!

Performance with a 1000x UDMA7 card is awesome...

 

[slughead]

This list is not updated and the latest AMD and Nvidia cards are simply missing, just like the latest version of OSX(but I'm sure many pro are succesfully using this with or without Autodesk certification). About pc apps, why would Autodesk reccomend an Apple specific hardware for Windows tasks? This will never happen.
Anyway we are "Mac" people on a Mac forum, most of us use the software almost exclusively in OSX, so the Windows drivers things is much overrated IMO.
I'm personally interested in how those cards perform in OSX, and they look very good in both benchmarks and users experience. I do not care if the heaven benchmark score is the slowest on earth as long as the Cinema4D OpenGl score is quite good, because C4D is part of my workflow and I make money with it, while I'm sure nobody is making money with heaven benchmark

After multiple posts of you asking for a source (that this isn't certified in Pro Windows Apps), I post one, then you say you don't care. Why'd you make such a big deal about it then??

I'm still waiting for SpecPerf benchmarks, the guy who tested SolidWorks on his D300 seemed impressed--that's a good sign, but we still need numbers. It runs about twice as fast on the W7000 as 7970GE.

 

[spaz8]

I'm not sure if this helps with any discussion here, but I am also thinking about getting a nMP. My choice would be 6 core / D700. Here is a benchmark from my late 2013 iMac Fully maxed out.

So looking at your imac score. (my 6core D700 is on order)

Looks like the D700 is not even 10% better than the 780M in OpenGL, it will crush it at compute though - and I guess if crossfireX ever makes it into OSX.

On the CPU side, as expected the 6-core beats the i7 by about 37%.

 

[Coyote2006]

I decided to upgrade my good old MacPro 4.1 to give it some more years to live. It's still a very fast machine (for webdesign and programming) and with the new CPU, the GTX 680, SSDs and more RAM it's even great for gaming. I never had a PC that lived that long ...

With the money I saved from upgrading instead of buying a nMP I bought a fully loaded MacBook Pro 15".

I'm happy with my decision

Next hardware: a new monitor...