alankelly

Results 49 comments of alankelly

Xnnpack still builds with `+dotprod` and `+fp16` with `-DXNNPACK_ENABLE_ARM_DOTPROD=OFF -DXNNPACK_ENABLE_ARM_FP16_SCALAR=OFF -DXNNPACK_ENABLE_ARM_FP16_VECTOR=OFF`

Can you try adding -DXNNPACK_ENABLE_ASSEMBLY=OFF?

Xnnpack still builds with `+dotprod` and `+fp16` with `-DXNNPACK_ENABLE_ARM_DOTPROD=OFF -DXNNPACK_ENABLE_ARM_FP16_SCALAR=OFF -DXNNPACK_ENABLE_ARM_FP16_VECTOR=OFF`

Ok, we know what the problem is now. The solution is to get the update-microkernels script to split the assembly files into ones with and without arm V8 and to...

Why is Signal 7 reporting an error on the armv7a platform TEST (F16_VCMUL_NEONFP16ARITH_U8, batch_lt_8) ?

Thanks for the report. This was fixed back in March in https://github.com/google/XNNPACK/commit/e5c870c1309a7857ceb9e4f5a2be30087d485874

unsupported instruction `vpdpbusd'

This looks similar to https://github.com/google/XNNPACK/issues/5892 Did you compile your compiler yourself? "From the Pytorch thread, the user is using CentOS 7 which is now quite old. gcc 10.2 was locally...

unsupported instruction `vpdpbusd'

Can we close this?

Add QS8_QC8W GEMM/IGEMM microkernels for Wasm Relaxed Unsigned and Signed …

Thanks for this. Improved WASMSIMD performance is desperately needed on Intel. Can you please send a follow up updating the gemm-config?

Add f32 prelu RVV microkernels, tests and config.

@fbarchard can you please review?

QB4W MLAL GEMM Kernels

Importing.

Update test and benchmark generation for blockwise kr>2

@gonnet Can you please review?

Update test and benchmark generation for blockwise kr>2

Please rebase and re-run test generation. This PR will be difficult to land since it touches the frequently changing generated bench/test files so after you re run generation, verify that...