Purpose of these kernels

Accelerated Computing GPU-Accelerated Libraries
1

I have been profiling a ML workload and have been coming across kernels named nvjet_hsh_128x128_64x6_1x1_h_bz_splitK_NTN and nvjet_hsh_192x192_64x3_1x2_h_bz_coopB_NTN and so on. This is a new kind of kernel names I am seeing. May I know what are nvjet kernels in general?

Purpose of these kernels
What is this kernel 'nvjet_tst_112x64_64x9_1x2_h_bz_bias_TNN'? Which cuda api do i need?
Switch from "sm90_xmma_gemm.._cublas"/ "void cutlass::Kernel<cutlass_80_tensorop_.." kernels with CUDA-12.1 to "nvjet_tst..." kernels with CUDA-12.8
2

Hi, thanks for your question! In general, the nvjet kernels are another one of NVIDIA’s internal kernel backends.

3

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.