[GPU Tool] GPU Caps Viewer 1.8.9 with OpenGL 4.1 and Better OpenCL Support
GPU Caps Viewer 1.9.0 with OpenGL 4 tessellation demo is available HERE.
A new update of GPU Caps Viewer is available. This new version adds the support of OpenGL 4.1 and improves the OpenCL support.
You can download GPU Caps Viewer 1.8.9 here (Win32 installer):
Webmasters: hotlinking is not allowed, please use the post url as download link.
The ZIP version is available here (no installation required):
If you have several GPUs in your system, it’s now possible to choose the OpenCL device for running the OpenCL demos. Just go on the OpenCL panel and you’ll see the device selection boxes:
For example, my dev system has a GTX 480 as main graphics card and a GT 240 as secondary card. By default, the OpenCL device is set to 0. So the GTX 480 is used to run the OpenCL kernels:
Now when the device 1 is selected, the GT 240 is used to process the OpenCL kernels:
On my test bench, GPU Caps Viewer 1.8.8 didn’t find the OpenCL support for my Radeon HD 5870 + Catalyst 10.7 + ATI Stream v2.1. GPU Caps 1.8.9 fixes that bug… (don’t really know if it was a bug, I certainly changed something but I don’t know what). Anyway it works now 😉
What is GPU Caps Viewer?
GPU Caps Viewer is an information utility focused on the OpenGL / OpenCL / CUDA API level support of the main graphics card. Even if you have several graphics cards in your system, only one is considered as the main OpenGL or Direct3D device. For OpenCL and CUDA, GPU Caps Viewer details the API support of each capable device. More information about GPU Caps Viewer can be found HERE.
GPU Caps Viewer 1.8.9 changelog
- New: added the support of OpenGL 4.1
- New: added in OpenCL panel, the selection of the current device index for the demos.
- New (PRO version only): a command line option allows to select the OpenCL device on which running the OpenCL kernels. This option is useful if you have several OpenCL capable GPUs. By default the first device is used (OpenCL device 0). The command line param is /cl_device_index=n where n ranges from 0 to # of GPU/CPU – 1.
- New: added NVIDIA Quadro 5000 to the database.
- Update: minor changes in OpenCL related code.