-
Notifications
You must be signed in to change notification settings - Fork 80
Closed
Description
What's the issue, what's expected?:
- GPU STREAM (
superbenchmark/superbench/benchmarks/micro_benchmarks/gpu_stream/gpu_stream.cu) fails to compile withe CUDA >= v13.0 . ThememoryClockRateproperty incudaDevicePropis no longer supported. Probably need an#ifdefto use the correctcudaDevicePropproperties based on the CUDA version.
Log message or shapshot?:
.../superbenchmark/superbench/benchmarks/micro_benchmarks/gpu_stream/gpu_stream.cu(121): error: class "cudaDeviceProp" has no member "memoryClockRate"
std::cout << " Memory: " << prop.memoryClockRate / 1000.0 << "MHz x " << prop.memoryBusWidth
^
.../superbenchmark/superbench/benchmarks/micro_benchmarks/gpu_stream/gpu_stream.cu(122): error: class "cudaDeviceProp" has no member "memoryClockRate"
<< "-bit = " << (prop.memoryClockRate / 1000.0) * (prop.memoryBusWidth / 8) * 2 / 1000.0 << " GB/s PEAK ";
^
2 errors detected in the compilation of ".../superbenchmark/superbench/benchmarks/micro_benchmarks/gpu_stream/gpu_stream.cu".
gmake[2]: *** [CMakeFiles/gpu_stream.dir/build.make:105: CMakeFiles/gpu_stream.dir/gpu_stream.cu.o] Error 2
gmake[1]: *** [CMakeFiles/Makefile2:83: CMakeFiles/gpu_stream.dir/all] Error 2
gmake: *** [Makefile:136: all] Error 2
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2025 NVIDIA Corporation
Built on Wed_Aug_20_01:57:39_PM_PDT_2025
Cuda compilation tools, release 13.0, V13.0.88
Build cuda_13.0.r13.0/compiler.36424714_0
Additional information:
- Compiling for GB300
- Note using Docker
Metadata
Metadata
Assignees
Labels
No labels