pytorch/caffe2/perfkernels
Hector Yuen 26db46b324 change the epilogue of SLS to match the simd section (#21439)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/21439

this bug got exposed after testing accuracy on shapes not multiples of 8

Reviewed By: jspark1105

Differential Revision: D15684759

fbshipit-source-id: 2950f2bd87ee1d8e539148285a14c755f606b3a7
2019-06-05 18:41:55 -07:00
..
__init__.py re-enable copy of python files, but be careful that the copy is only … (#14982) 2018-12-11 16:54:08 -08:00
adagrad_avx.cc use fp16<->fp32 intrinsic (#17496) 2019-03-07 02:23:07 -08:00
adagrad.cc more careful use of inline/template function in perfkernels (#15388) 2019-01-30 22:49:37 -08:00
adagrad.h more careful use of auto in sparse operations (#17958) 2019-03-14 22:10:42 -07:00
CMakeLists.txt Resolve errors in perfkernel for Windows (#16031) 2019-01-16 21:51:00 -08:00
common_avx.cc Remove Apache headers from source. 2018-03-27 13:10:18 -07:00
common_avx2.cc Remove Apache headers from source. 2018-03-27 13:10:18 -07:00
common_avx512.cc include avx512vl to avx512 code path (#14733) 2018-12-05 00:50:51 -08:00
common.h Resolve errors in perfkernel for Windows (#16031) 2019-01-16 21:51:00 -08:00
cvtsh_ss_bugfix.h use fp16<->fp32 intrinsic (#17496) 2019-03-07 02:23:07 -08:00
embedding_lookup_avx2.cc change the epilogue of SLS to match the simd section (#21439) 2019-06-05 18:41:55 -07:00
embedding_lookup_fused_8bit_rowwise_avx2.cc change the epilogue of SLS to match the simd section (#21439) 2019-06-05 18:41:55 -07:00
embedding_lookup.cc remove dependency to caffe2::math and eigen (#21169) 2019-05-31 11:55:16 -07:00
embedding_lookup.h Improve embedding_bag add kernel (#19329) 2019-04-19 19:16:24 -07:00
fused_8bit_rowwise_embedding_lookup.cc remove dependency to caffe2::math and eigen (#21169) 2019-05-31 11:55:16 -07:00
fused_8bit_rowwise_embedding_lookup.h more careful use of inline/template function in perfkernels (#15388) 2019-01-30 22:49:37 -08:00
hp_emblookup_codegen.py change the epilogue of SLS to match the simd section (#21439) 2019-06-05 18:41:55 -07:00
math_cpu_avx2.cc more careful use of inline/template function in perfkernels (#15388) 2019-01-30 22:49:37 -08:00
math_cpu_base.cc Resolve errors in perfkernel for Windows (#16031) 2019-01-16 21:51:00 -08:00
math.h Resolve errors in perfkernel for Windows (#16031) 2019-01-16 21:51:00 -08:00
typed_axpy_avx.cc remove dependency to caffe2::math and eigen (#21169) 2019-05-31 11:55:16 -07:00
typed_axpy_avx2.cc remove dependency to caffe2::math and eigen (#21169) 2019-05-31 11:55:16 -07:00
typed_axpy.cc remove dependency to caffe2::math and eigen (#21169) 2019-05-31 11:55:16 -07:00
typed_axpy.h Remove Apache headers from source. 2018-03-27 13:10:18 -07:00