Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / google/XNNPACK issues and pull requests
#5951 - Fix enable QD8_F32_QC8W 5x8C8 AVX VNNI microkernels before avx512skx
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5950 - Fix QS8-QC8W-GEMM variable names
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5949 - Enable QD8_F32_QC4W 7x8C8 AVX512VL VNNI microkernels for mobile
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5948 - average pool zero buffer should be aligned
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5947 - Add an XNNPACK delegate for the `Rsqrt` node in TFLite.
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5946 - QS8 E2E GEMM benchmark add AVXVNNI
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5945 - Explicitly disable RISC-V vector extensions on Android
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5944 - Get xnnpack.h back into list of installed files
Pull Request -
State: open - Opened by iskunk 10 months ago
#5943 - Enable QD8-F32-QC8W GEMM/IGEMM AVXVNNI microkernels
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5942 - Update amalgamated microkernels with unary `rsqrt`.
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5941 - Add an operator for the reciprocal square root.
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5940 - Add f32 rsum RVV implementation microkernels, tests and config changes
Pull Request -
State: open - Opened by KaustubhIMG 10 months ago
- 15 comments
#5939 - Add f32 rsum RVV implementation microkernels, tests and config changes
Pull Request -
State: closed - Opened by KaustubhIMG 10 months ago
- 1 comment
#5938 - QS8/QD8 C8 AVXVNNI IGEMM microkernel
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5937 - Add has_udot for ARM disable dot product on linux kernels older than 6.7
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5936 - QD8-AVXVNNI C4 GEMM microkernel use same store code QD8-AVXVNNI C8 GEMM
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5935 - [QD8_F32_QC4W] Issue with odd number of input_channels
Issue -
State: closed - Opened by digantdesai 10 months ago
- 3 comments
#5934 - Add benchmarks for the scalar `f32` `vrsqrt` microkernels.
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5933 - Disable dot product in armv7 on linux kernels older than 4.7
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5932 - Add tests for the scalar `f32` `vrsqrt` microkernel.
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5931 - Add microkernels for scalar `f32` `rsqrt`.
Pull Request -
State: open - Opened by copybara-service[bot] 10 months ago
#5930 - Fix bazel OS builds
Pull Request -
State: closed - Opened by copybara-service[bot] 10 months ago
#5929 - reshape concatenation operators output
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5928 - reshape Batch Matrix Multiply output
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5927 - Simplify AVX_VNNI hardware config.
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5926 - Conditionally compile avxvnni code
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5925 - re generate tests and benches
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5924 - Automatically generate spmm benches
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5923 - avx512 GEMM/IGEMM kernels switch from vsra to vsrl
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5922 - Fix white space on PPC expression
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5921 - Remove OOB declarations from RVV
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5920 - Add F16-VCLAMP-RVV for RISC-V clamping
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5919 - Fix xnn_pack_deconv_goki_w_fn signature for CFI
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5918 - VCLAMP ASAN fix for F32 and F16 Neon
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5917 - F32-VCLAMP Neon read remainder with 2 or 1 float to avoid asan overread
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5916 - Disable subgraph shape inference tests until the design is finalized
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5915 - QD8_F32_QC4W VNNI GEMM and IGEMM microkernels remove masking on remainder
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5914 - Reshape output tensor for average pooling 2d
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5913 - Add xnn_define_concatenate for 5 inputs
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5912 - Add support for RVV x32-transpose microkernels.
Pull Request -
State: closed - Opened by phoebesv 11 months ago
- 1 comment
#5911 - Move channels, input_pixel_stride and output_pixel_stride from create to reshape so that these dimension may be dynamically reshaped.
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5910 - [Dynamic Shapes] Fully-connected op
Pull Request -
State: closed - Opened by digantdesai 11 months ago
- 2 comments
#5909 - Fix CFI issue with XNNPack deconv
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5908 - Enable AVX-VNNI 5x8c8 GEMM microkernel for QD8_F32_QC4W
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5907 - Look up weights cache before doing packing in fully-connected
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5906 - Fix OS builds:
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5905 - Refactor average pooling 2d to remove duplicate code.
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5904 - AVX-VNNI MRx8c8 GEMM microkernels use vpermq immediate
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5903 - Reshape output tensor for static transpose
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5902 - Add RVV softmax benchmark & add f32 rmax, rminmax, and raddstoreexpminusmax to config
Pull Request -
State: closed - Opened by bhbruce 11 months ago
- 1 comment
#5901 - Only update output shape if it has changed
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5900 - Fix x86 OS builds
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5899 - Make clamp on empty ranges valid
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5898 - AVX-VNNI MRx8c8 GEMM microkernels.
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5897 - Add oneDNN v3 API support in XNNPACK softmax benchmark.
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5896 - Test. Do not submit.
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5895 - internal config change
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5894 - Fix GTest linkage for new tests from commits 576ba7b and 06ae300
Pull Request -
State: closed - Opened by iskunk 11 months ago
#5893 - Support RVV F32-GEMM
Pull Request -
State: closed - Opened by bhbruce 11 months ago
- 2 comments
#5892 - Request for Legacy CPU Support or Improved Error Handling
Issue -
State: closed - Opened by yewentao256 11 months ago
- 2 comments
#5892 - Request for Legacy CPU Support or Improved Error Handling
Issue -
State: open - Opened by yewentao256 11 months ago
#5891 - f32-raddstoreexpminusmax-rvv-rr2-p6-u4v.c error: no member named 'rvv_rr2_p6' in 'union xnn_f32_expminus_params'
Issue -
State: closed - Opened by fbarchard 11 months ago
- 1 comment
#5890 - Replace riscv64 vector GCC CI with clang toolchain.
Pull Request -
State: closed - Opened by phoebesv 11 months ago
- 1 comment
#5889 - AVX512VNNI GEMM use GFNI to shift QC4W nibbles left 4
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5888 - Remove qs8 GEMM/IGEMM template generators
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5887 - Add AVX512VNNI C8 microkernels and AVX512 1,5,6,7,8x16 kernel sizes
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5886 - Fix OS builds without Ruy
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5885 - Remove QS8 microkernels from gemm & igemm headers
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5884 - Remove deprecated QS8 GEMM & IGEMM microkernels
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5883 - Remove references to deprecated qs8 microkernels
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5882 - qs8-gemm-e2e bench uses qs8-qc8w microkernels
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5881 - Remove deprecated qs8_gemm_config
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5880 - Conv2D QS8 takes equivalent QS8-QC8W path
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5879 - Deconv2D QS8 takes QS8-QC8W path
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5878 - Fully Connected QS8 takes equivalent QS8-QC8W path
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5877 - Add RVV version of vpoc f32-vbinary
Pull Request -
State: open - Opened by bhbruce 11 months ago
- 1 comment
#5877 - Add RVV version of vpoc f32-vbinary
Pull Request -
State: closed - Opened by bhbruce 11 months ago
- 1 comment
#5876 - RVV F32-Softmax Patch 2: Add RVV f32-raddstoreexpminusmax
Pull Request -
State: closed - Opened by bhbruce 11 months ago
- 1 comment
#5875 - Remove unused extended variant from avx512 QS8 GEMM
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5875 - Remove unused extended variant from avx512 QS8 GEMM
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5874 - C8 VNNI use acc2 for MR < 4
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5874 - C8 VNNI use acc2 for MR < 4
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5873 - Addressing TF query analysis errors
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5872 - Add enable_avxvnni to bulldozer targets
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5872 - Add enable_avxvnni to bulldozer targets
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5871 - Use vdupq_n_f16 instead of vld1q_dup_f16 as MSVC doesn't support this instruction
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5871 - Use vdupq_n_f16 instead of vld1q_dup_f16 as MSVC doesn't support this instruction
Pull Request -
State: closed - Opened by copybara-service[bot] 11 months ago
#5870 - QS8 AVX512 GEMM/IGEMM use _mm512_cvtepi32_epi8 to convert int32_t to int8_t
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5869 - RVV F32-Softmax Patch 1: Add RVV reduced min & max & minmax
Pull Request -
State: closed - Opened by bhbruce 12 months ago
- 1 comment
#5868 - Clamp on empty ranges should be valid
Issue -
State: closed - Opened by fdwr 12 months ago
- 2 comments
#5867 - Fix another source of undefined behavior in memcpy
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5866 - Avoid undefined behavior in memcpy call in `xnn_define_static_reshape`
Issue -
State: closed - Opened by huningxin 12 months ago
#5865 - Reshape for all binary elementwise functions
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5864 - Support xN-transpose benchmarking for problem sizes HW128, 256, and 512.
Pull Request -
State: closed - Opened by phoebesv 12 months ago
- 1 comment
#5863 - QS8 AVX512 use vpshufd instead of vpshufb
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5862 - Enable f32-vclamp-test in rvv-ci
Pull Request -
State: closed - Opened by bhbruce 12 months ago
- 1 comment
#5861 - QS8 AVX512SDK MRx16C8 IGEMM microkernels use same store code as VNNI
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5860 - Implementation of reshape output tensor for all unary elementwise ops
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5859 - Unary elementwise op reshape output tensor
Pull Request -
State: closed - Opened by copybara-service[bot] 12 months ago
#5858 - Shape inference for Add
Pull Request -
State: open - Opened by copybara-service[bot] 12 months ago