Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Mac Pro D500 Unit ready for Testing. What would you like me to run?
- Thread starter Mgmx
- Start date
- Sort by reaction score
The D700 standard 3d Mark test it's here. MacFormat tested
http://www.youtube.com/watch?v=TdLOh8MdU20
Thank you
http://www.youtube.com/watch?v=TdLOh8MdU20
Thank you
Last edited:
Here is what macworld found but not at the same resolution sadly...
http://www.macworld.com/article/2082568/lab-tested-new-mac-pro-is-the-speedster-weve-been-waiting-for-finally.html?page=2
Also on Mac X the opengl seems to be depend on the computer CPU speed a little bit on most games when it comes to FPS if using the same video cards. So the computer with the higher CPU speeds will get the better FPS even if they have the same video card.
as an example:
http://www.barefeats.com/gpu680v4.html
I should have said non-realtime graphical calculations. For 3D renders I think the double precision advantage will make a difference. For games or benchmarks like these, as we've found out in this thread, there is effectively no difference between the D300 and D500.
Plus in those benchmarks, and games they're still using OpenGL 3.2, and just a single AMD Dxx graphics card.
The NVIDIA cards have always had better OpenGL performance, I wonder if the software will be updated to be able to take advantage of both GPUs, and the newer OpenGL versions.
At the moment it seem the D700's are the best all rounders.
Yeah, which means in some OpenGL 3d apps like Mudbox, it will be worse than 5870 ... But it all can change because of better Pcie link speed...
Thats why i'm still trying to get real world experiment by just loading as many polygons as possible in Maya and Mudbox will reveal true power of this card.
It's FAR from useless with 3d work.
It outperforms a Titan with Mari at the moment.
MAri is a ONE of a kind beast they was able to utilize double gnu power, can we compare d500 in maya to gtx 680 4gb ? please?
Just create a cube subdivide it to roughly 90 000 ploys, and duplicate it as much as you can until you can't spin viewport. iMac 680MX started to lagg at 48 000 000 polygon.
----------
osx mac pro 5,1 gtx 680 4gb flashed Heven score if anyone interested:
I'll test this maya bit quickly but keep in mind Maya will only run one of the GPUs in 2014 SP3 and this will also very much about total system memory as well.
Yes of course, so as other apps too, and Autodesk won't utilize those 2 gpus in near future, they just don't care.
Lets be real here everyday 3d Artist / video editor won't use true power of those dual gpus, we just want to test already existing software which is most likely developers won't invest into optimizing for new mac pro... in very near future, lets leave all those TFLOPS to scientists with their own soft.
Thanks for testing.
In maya viewport (default rendering)
at Verts 70288790, still moving..seems to get useless around 84000000 (skippy but not destroyed)
----------
Didn't save one.
----------
Autodesk has full plans to implement OpenGL with multiple gpus in the near future.
at Verts 70288790, still moving..seems to get useless around 84000000 (skippy but not destroyed)
----------
Didn't save one.
----------
Autodesk has full plans to implement OpenGL with multiple gpus in the near future.
Any realtime rendering will use the GPU's to huge effect. e.g. VRAY RT on 3dsmax + other OpenCL accelerated viewport display.
In theory it's all pretty yes, but chaps group already gave up on trying to make RT work on AMD OpenCL
They tried new mac pro still crashes...
I don't have a request...
... just wanted to say thanks to Mgmx for indulging all of us with the benchmarks. Great stuff.
... just wanted to say thanks to Mgmx for indulging all of us with the benchmarks. Great stuff.
Still going actually.....
And honestly, on the multitude of extremely high budget films I've worked with I've never had the need to put this much geometry in a viewport.
In theory it's all pretty yes, but chaps group already gave up on trying to make RT work on AMD OpenCL
Are you sure? I know Chaosgroup gave up on Vray 1.5x when RT was a separate tool but improved it in 2.0 and again in 3.0 beta (which I really want to use myself).
yeah man just talked the other week to Vlado, no go on OpenCL on AMD, but we have speedster CPU here so we good with V-RAY RT CPU, in future maybe plugging tesla through thunderbolt can gives us V-RAY RT CUDA.
Or Chaos will be total geniuses and make AMD work.
You're welcome! I'm going to have to get back to actual work after this, so I'm unfortunately not going to be able to do much more in the way of benchmarking.