diff options
| author | Michael Tyler <michael.tyler@arm.com> | 2024-06-04 15:47:37 +0100 |
|---|---|---|
| committer | Michael Tyler <michael.tyler@arm.com> | 2024-06-25 09:10:13 +0000 |
| commit | fc94f4d23abd4bc427b701f54ad85282e9ec7872 (patch) | |
| tree | 5e2980599256e2b2f4374e5beb61596fc95c9d5a /src/BUILD.bazel | |
| parent | c2237ec4094c7824f8f7e61bc89504d01c5b59ff (diff) | |
| download | ComputeLibrary-fc94f4d23abd4bc427b701f54ad85282e9ec7872.tar.gz | |
Update CPU kernels and add mixed sign GEMM support
- Add support for mixed sign quantized convolution.
- Add support for mixed sign dequantized GEMM.
- Add SME FP16 GEMV kernel.
- Change SME vector length function to use RDSVL instead of static variable.
- Add GEMM dilation support internally (not exposed yet).
- Remove unused "get_default_activation_values" functions.
- Add SVE fixed format interleaved BF16 DOT kernel.
- Updates and optimizations to assembly kernels.
Resolves COMPMID-6926
Change-Id: I227f502502611d4cc4111c89e30c53ce94079544
Signed-off-by: Michael Tyler <michael.tyler@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11570
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/BUILD.bazel')
| -rw-r--r-- | src/BUILD.bazel | 13 |
1 files changed, 13 insertions, 0 deletions
diff --git a/src/BUILD.bazel b/src/BUILD.bazel index f270824ab4..8cb84b3533 100644 --- a/src/BUILD.bazel +++ b/src/BUILD.bazel @@ -248,6 +248,7 @@ filegroup( "core/NEON/kernels/arm_conv/pooling/kernels/sve_u8q_nhwc_max_generic_depthfirst/generic.cpp", "core/NEON/kernels/arm_gemm/interleave_indirect-sve.cpp", "core/NEON/kernels/arm_gemm/kernels/sme2_gemv_bf16fp32_dot_16VL/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp16_mla_16VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp16fp32fp16_dot_16VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp32_mla_16VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sme2_gemv_fp32bf16fp32_dot_16VL/generic.cpp", @@ -280,6 +281,7 @@ filegroup( "core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32_mla_6x4VL/a64fx.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32_mla_6x4VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_ffhybrid_fp32bf16fp32_mmla_4x6VL/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_bf16fp32_dot_8x3VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_bf16fp32_mmla_8x3VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_fp16_mla_8x3VL/a64fx.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_ffinterleaved_fp16_mla_8x3VL/generic.cpp", @@ -304,6 +306,9 @@ filegroup( "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_s8s32_mmla_6x4VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8qa_dot_4x4VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8qa_mmla_4x4VL/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8s8qa_dot_4x4VL/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8s8qa_mmla_4x4VL/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8s8s32_mmla_6x4VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8u32_dot_6x4VL/a64fx.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8u32_dot_6x4VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_hybrid_u8u32_mmla_6x4VL/generic.cpp", @@ -317,6 +322,7 @@ filegroup( "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_s8s32_dot_8x3VL/a64fx.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_s8s32_dot_8x3VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_s8s32_mmla_8x3VL/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8s8s32_mmla_8x3VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_dot_8x3VL/a64fx.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_dot_8x3VL/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/sve_interleaved_u8u32_mmla_8x3VL/generic.cpp", @@ -521,9 +527,11 @@ filegroup( "core/NEON/kernels/arm_gemm/gemm_fp32.cpp", "core/NEON/kernels/arm_gemm/gemm_int16.cpp", "core/NEON/kernels/arm_gemm/gemm_int8.cpp", + "core/NEON/kernels/arm_gemm/gemm_q8_mixed.cpp", "core/NEON/kernels/arm_gemm/gemm_qint8.cpp", "core/NEON/kernels/arm_gemm/gemm_quint8.cpp", "core/NEON/kernels/arm_gemm/gemm_s8fp32.cpp", + "core/NEON/kernels/arm_gemm/gemm_u8s8fp32.cpp", "core/NEON/kernels/arm_gemm/gemm_uint16.cpp", "core/NEON/kernels/arm_gemm/gemm_uint8.cpp", "core/NEON/kernels/arm_gemm/interleave-8way.cpp", @@ -574,6 +582,10 @@ filegroup( "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8qa_dot_4x16/a55.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8qa_dot_4x16/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8qa_mmla_4x16/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8s8qa_dot_4x16/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8s8qa_mmla_4x16/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8s8s32_dot_6x16/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8s8s32_mmla_6x16/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_dot_6x16/a55.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_dot_6x16/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_hybrid_u8u32_mmla_6x16/generic.cpp", @@ -583,6 +595,7 @@ filegroup( "core/NEON/kernels/arm_gemm/kernels/a64_interleaved_bf16fp32_mmla_8x12/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/a510.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_interleaved_s8s32_mmla_8x12/generic.cpp", + "core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8s8s32_mmla_8x12/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/a510.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_interleaved_u8u32_mmla_8x12/generic.cpp", "core/NEON/kernels/arm_gemm/kernels/a64_sgemm_8x12/a53.cpp", |