[cherry-pick] Optimize sparse kernel and fix some bug#50118

Merged

lanxianghit merged 7 commits intoPaddlePaddle:release/2.4from zkh2016:cherry-sparse

Feb 2, 2023

+1,322-197

Commits on Jan 31, 2023

[Sparse]Optimize BatchNorm1D forward in test mode (PaddlePaddle#47736 )
zhangkaihuo
committed
[Sparse]optimize sparse convolution and fix MaskHelper bug (PaddlePaddle#47703 )
zhangkaihuo
committed
cherry-pick PaddlePaddle#47604
zhangkaihuo
committed
summer-ospp 2022: 飞桨PaddlePaddle Sparse Conv开发和优化: gather-gemm-scatter fuse (PaddlePaddle#46679 )

umiswing
authored and
zhangkaihuo
committed
group the index in not cutlass mode (PaddlePaddle#48439 )
zhangkaihuo
committed
[Sparse]Optimize performance of sparse conv on T4 (PaddlePaddle#49009 )
zhangkaihuo
committed
cherry-pick PaddlePaddle#49734
zhangkaihuo
committed