CUDA Nvidia GPU Benches computes mandelbrot and nbody (galaxies) on these Nvidia GPUs: 8xxx/9xxx/GTX285.
mandelbrot shows FPS
nbody shows GigaFlops
nbody (Galaxies) in that CUDA bench runs faster that the OpenCL Galaxies (Apple) Bench because GPU code part was optimized. GTX 285 got 450 GigaFlops! (Galaxies OpenCL got 300 GigaFlops.)
Requirements
Intel, Nvidia GPU 8xxx/9xxx/GTX285, Nvidia OS X CUDA driver (Links in the read me).
Be the first to recommend a similar software title.
I don't know if your reply came before the message that some library may be missing. In any case the LIBCUDART library is in the nBody folder and I surely did the folder change (CD command) before starting Mandelbrot.
from dev: Sorry, if cuda tools sdk wasnt installed, some .dylib are missed. I and the other beta testers all have that sdk installed so i havent seen that problem.
Will be updated soon (in a few hours) with that dylibs.
Thanks for problem report here.
Information from dev for MacPro / XServe Users:
The actual CUDA driver (available from Nvidia, Info + DL Links in the readme) can´t work in the 64 Bit KernelMode of MacPro / XServe.
But Nvidia will updated soon. If you have an MacPro wich is using the 64 Bit KernelMode by default, you can try that Benches by booting in 32 Bit KernelMode ( Apps can run in 64 Bit independend from 32 Bit KernelMode)
Happy CUDA Benching !!!
No way to have neither Mandelbrot nor Nbody to work. Procedure has been done per included Readme file.
Error message is:
dyld: Library not loaded: @rpath/libcudart.dylib
Referenced from: /Users/majortom/Desktop/CUDA_Nvidia_GPU_Benches/Mandelbrot_/Mandelbrot
Reason: image not found
Trace/BPT trap
Both in 10.5.8 and 10.6.1 on iMac 2009 - C2D 2,66 - nVidia 9400.
Please login or create a new MacUpdate Member account to use this feature
Watch Lists are available to MacUpdate Desktop Members Upgrade Now
Download and auto-install
using MacUpdate Desktop. Save
time moving folders and cleaning-up.
CUDA Nvidia GPU Benches computes mandelbrot and nbody (galaxies) on these Nvidia GPUs: 8xxx/9xxx/GTX285.
mandelbrot shows FPS
nbody shows GigaFlops
nbody (Galaxies) in that CUDA bench runs faster that the OpenCL Galaxies (Apple) Bench because GPU code part was optimized. GTX 285 got 450 GigaFlops! (Galaxies OpenCL got 300 GigaFlops.)
+73
Will be updated soon (in a few hours) with that dylibs.
Thanks for problem report here.
+73
The actual CUDA driver (available from Nvidia, Info + DL Links in the readme) can´t work in the 64 Bit KernelMode of MacPro / XServe.
But Nvidia will updated soon. If you have an MacPro wich is using the 64 Bit KernelMode by default, you can try that Benches by booting in 32 Bit KernelMode ( Apps can run in 64 Bit independend from 32 Bit KernelMode)
Happy CUDA Benching !!!
Error message is:
dyld: Library not loaded: @rpath/libcudart.dylib
Referenced from: /Users/majortom/Desktop/CUDA_Nvidia_GPU_Benches/Mandelbrot_/Mandelbrot
Reason: image not found
Trace/BPT trap
Both in 10.5.8 and 10.6.1 on iMac 2009 - C2D 2,66 - nVidia 9400.
+73
start terminal
cd ....path to the folder of the app you want to test (DRAG&DROP) ENTER
Now you can start the app (nbody or mandelbrot) in terminal.
If you try to start the app before you changed the directory to the folder of the app, the app cant find any .dylib and will fail !
So the change directory (cd command) is essential .