search

NVIDIA / cutlass

CUDA Templates for Linear Algebra Subroutines
Other
5.3k stars 892 forks source link

[BUG] Convolution examples fail to compile #1591

Open ogoidmatos opened 2 months ago

ogoidmatos commented 2 months ago

Describe the bug When compiling the sample code for examples/16_ampere_tensorop_conv2dfprop/ampere_tensorop_conv2dfprop.cu, it fails with the following error message. Any other example for convolution forward propagation gives the same error.

nvcc conv.cu -gencode=arch=compute_86,code=\"sm_86,compute_86\" -std=c++17 -I/home/user/cutlass-3.4.1/include -I/home/user/cutlass-3.4.1/tools/util/include -c -o objs/conv.o

/home/user/cutlass-3.4.1/include/cutlass/conv/device/implicit_gemm_convolution.h(215): error: namespace "cutlass::conv" has no member "implicit_gemm_tensor_c_size"
          size_t(cutlass::conv::implicit_gemm_tensor_c_size(kConvolutionalOperator, args.problem_size)) *
                                ^

1 error detected in the compilation of "conv.cu".

Steps/Code to reproduce bug Compile the example code with a RTX3080 GPU.

Expected behavior Compilation without errors.

Environment details (please complete the following information):

(if needed more information, please do ask! Thank you!)

github-actions[bot] commented 1 month ago

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.