[QNNPACK:Sparsity] Add A matrix pretransformed based sparse kernels for FC (#50587)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/50587
This diff introduces two kernesl. One is to pretransform A to do block
wise transforms.
And then the kernel that directly works on top pretransformed weights.
Test Plan:
./build/local/q8gemm-sparse-test
./build/local/fully-connected-sparse-test
Imported from OSS
Reviewed By: AshkanAliabadi
Differential Revision: D25925504
fbshipit-source-id: 9b02819405ce587f20e675b154895dc39ecd1bad
