Jiong Gong

Propose code owners from Intel

5

This PR proposes a list of CPU-related PyTorch modules that Intel is willing to own or co-own.

Incorrect result from first-run autotune function

5

Repro: ```python import torch from torch._C import _cuda_getCurrentRawStream as get_cuda_stream import triton import triton.language as tl @triton.autotune( configs=[ triton.Config({'XBLOCK': 1, 'RBLOCK': 64}), triton.Config({'XBLOCK': 1, 'RBLOCK': 64}), ], key=['xnumel', 'rnumel',], )...

[inductor][cpp] move vectorized type conversion to aten/cpu/vec

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #119979 * #119734 * #119655 * #119654 cc @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @voznesenskym @penguinwu @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10...

What GELU algorithm is used for LIBXSMM_MELTW_TYPE_UNARY_GELU?

3

Is it tanh approximation?

[inductor][cpp] disable masked load for non-fp data types

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #120558 Fix https://github.com/pytorch/pytorch/issues/120377. We disable the masked load for non-fp data types for now. The complete support of masks will be added...

[inductor][cpp] GEMM template

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #124021 * #125159 * #125152 cc @voznesenskym @penguinwu @EikanWang @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @aakhundov...

[inductor] autotune benchmark support for cpu

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #124021 * __->__ #125159 * #125152 cc @ezyang @msaroufim @bdhirsh @anijain2305 @chauhang @voznesenskym @penguinwu @EikanWang @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy...

[inductor][cpp] move some common cpp utils to cpp_utils.py

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #124021 * #125159 * __->__ #125152 cc @ezyang @msaroufim @bdhirsh @anijain2305 @chauhang @voznesenskym @penguinwu @EikanWang @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy...

I found the model is different from what was proposed form the original StyleTransfer paper: https://arxiv.org/abs/1508.06576 The original model has been implemented with Caffe: https://github.com/fzliu/style-transfer May I know how did...

[inductor][cpp] support bf16/fp16 gemm template epilogue fusion

1

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #126545 cc @voznesenskym @penguinwu @EikanWang @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang