CUDA Information Utility
for Windows, Linux and Mac OSX
Last Release: 0.7.181/0.7.184
This program was born as a parody of another Z-utilities such as CPU-Z and GPU-Z. CUDA-Z shows some basic information about CUDA-enabled GPUs and GPGPUs. It works with nVIDIA Geforce, Quadro and Tesla cards, ION chipsets.
CUDA-Z shows following information:
- Installed CUDA driver and dll version.
- GPU core capabilities.
- Integer and float point calculation performance.
- Performance of double-precision operations if GPU is capable.
- memory size and bandwidth.
This program also exports collected information to HTML format and plain text file.
This software is distributed under the terms of the GNU General Public License Version 2. So, feel free to redistribute and modify it under the GPL conditions.
CUDA-Z is known to not function with default Microsoft driver for nVIDIA chips. User must install official driver for nVIDIA products to run CUDA-Z.
It's strongly recommended to update your Windows regulary and use antivirus software to prevent data loses and system performance degradation.
CUDA-Z is known not to function with 3rd-party linux driver for nVIDIA chips. You can download the official driver from nVIDIA website if your distribution is not equipped with proper driver.
Note also that x86_64 linux requires 32-bit compatibility libraries package being installed to run CUDA-Z.
sudo apt-get install ia32-libs for Ubuntu x86_64).
In case of problem with available package you can download the source archive and follow instructions in readme.txt.
Mac OSX notes:
CUDA-Z runs in Mac OSX 10.7 Snow Leopard and later. Macs a shipped with proper Geforce driver but CUDA driver is not, so CUDA-Z additionaly requires special CUDA 5.0 Drivers for MacOS being installed in the system.
As far as CUDA 5.0 supports only Mac OSX 10.7 and 10.8 the new version of CUDA-Z is not able to run under Mac OSX 10.6. For those who runs Snow Leo on their Mac's it's recommended to use CUDA-Z 0.6.163 instead.
Please note that CUDA-Z for Mac OSX is in bata stage now and is not acquires heavy testing.
2013.05.11: Release 0.7.184 is out
Hotfix release for Mac OSX.
2013.05.02: Release 0.7.181 is out
- Better support of Windows 8.
- Number of fixes for Linux and MacOSX.
- Multiple bugfixes for all platforms.
- Use CUDA SDK 5.0+.
- Add GK11x support.
2012.10.19: Release 0.6.169 is out
Hotfix release for Linux.
- Fix ~/tmp removal on linux (Bug: 11).
2012.10.14: Release 0.6.163 is out
- Add new Heavy Load Test Mode. The CUDA test will run on all multiprocessors.
- Fix big total memory size bug (Bug: 6).
- Fix host to device bandwidth calculation (Bug: 10).
- Several graphic updates.
- Add driver and runtime version readout.
- Add new informations.
- Add MacOSX release.
- Use IEC 60027 prefixes for memory sizes and other 2^10-based value.
- Use perl instead of GNU tools for SVN build generation.
- Use of CUDA Toolkit 4.2.
- Multiple bugfixes.
Q: What should I do if CUDA-Z finds no CUDA?
A: Either you have no supported nVIDIA hardware or not supported OS or you have no proper driver installed. First check if your hardware is CUDA-enabled. If the hardware is OK, try to update your driver. It might be you have either limited nVIDIA driver or 3rd-party driver that not capable to run CUDA code.
Q: Ooooooops! CUDA-Z crashes.
A: Report a bug. I'll take care of it as fast as I can.
Q: I have a question...
A: Ask it in the forum. You are always welcome! :-)
Q: Gimme code!