CUDA 3.0がリリースされましたが、今回からOpenCL SDKも同梱されて /Developer
CUDA ("C"ディレクトリの中身)はmakeでsampleが普通にビルドできたのですが、OpenCLの方は以下のようなエラーが出ます。
1 src/oclUtils.cpp: In function ‘void oclPrintDe vInfo(int, _cl_device _id*)’: 2 src/oclUti ls.cpp:282 : error: ‘CL_DEVICE_ COMPUTE_CA PABILITY_M AJOR_NV’ was not declared in this scope 3 src/oclUti ls.cpp:283 : error: ‘CL_DEVICE_ COMPUTE_CA PABILITY_M INOR_NV’ was not declared in this scope 4 src/oclUti ls.cpp:290 : error: ‘CL_DEVICE_ REGISTERS_ PER_BLOCK_ NV’ was not declared in this scope 5 src/oclUti ls.cpp:294 : error: ‘CL_DEVICE_ WARP_SIZE_ NV’ was not declared in this scope 6 src/oclUti ls.cpp:298 : error: ‘CL_DEVICE_ GPU_OVERLA P_NV’ was not declared in this scope 7 src/oclUti ls.cpp:302 : error: ‘CL_DEVICE_ KERNEL_EXE C_TIMEOUT_ NV’ was not declared in this scope 8 src/oclUti ls.cpp:306 : error: ‘CL_DEVICE_ INTEGRATED _MEMORY_NV ’ was not declared in this scope 9 make: *** [obj/releas e/oclUtils .cpp.o] Error 1
これらの定数はSDK同梱のcommon/inc
posted by
genki
on Fri 2 Apr 2010
at 04:16