Bad devicequery info..

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi I’m testing OpenCL SDK with a 8800GT and so far I have seen only two minor bugs on deviceQuery:

First it reports as CL_DEVICE_EXTENSIONS: cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics
where l_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics must not be since it doesn’t support
shared memory atomics (cuda jargon)

also Mhz tab in Linux reports 0 or 1Mhz where in Windows correctly reports the speed (1500mhz)…

2

I’m effected by the second issue as well under Linux.

It’s running Ubuntu 9.04 with a GeForce 9800 GT (BFG’s OC edition).

Can’t speak for Windows at the moment.

3

My card is 9800gtx.
When I run the basic test “oclVectorAdd” in openCL SDK, the OpenCL context on a GPU device can not be created correctly.

There are following failed asserts:
NVAPI: bLhThunkInit: failed assert: lhThunk.pfnGetDisplayConfigBufferSizes
NVAPI: bLhThunkInit: failed assert: lhThunk.pfnQueryDisplayConfig
NVAPI: bLhThunkInit: failed assert: lhThunk.pfnSetDisplayConfig
NVAPI: bLhThunkInit: failed assert: lhThunk.pfnDisplayConfigGetDeviceInfo
NVAPI: bLhThunkInit: failed assert: lhThunk.pfnDispalyConfigSetDeviceInfo

How can I resolve above issues? :wacko: