- 8674629 Use QS8 GEMM WAsm SIMD microkernels with FP32 requantization in the benchmark by Marat Dukhan · 3 years, 3 months ago
- 0ff7989 Use FP32 requantization for extended-weights QS8 GEMM microkernels on x86 by Marat Dukhan · 3 years, 3 months ago
- ec47958 Prune redundant NEON GEMM/IGEMM microkernels with GEMMLOWP requantization by Marat Dukhan · 3 years, 3 months ago
- 3c5e662 Initialize QU8 VMUL[C] microkernels for pre-NEON ARM by Marat Dukhan · 3 years, 3 months ago
- 2dac7bb Unify on wasm_f64x2_spalt(0.0) to materialize zero SIMD vector in WAsm by Marat Dukhan · 3 years, 4 months ago
- d4db6af Replace wasm_i32x4_lt(vzero, vXX) with wasm_i32x4_shr(vXX, 31) by Marat Dukhan · 3 years, 4 months ago
- ebb6207 QU8 4x16 IGEMM remove push for X21 register by Frank Barchard · 3 years, 4 months ago
- 8a211a3 Check parameter initialization functions for non-NULL before calling by Marat Dukhan · 3 years, 4 months ago
- e145d56 Fix incompatibilities with AArch64 gcc in FP16 microkernels by Marat Dukhan · 3 years, 4 months ago
- eca1ea9 Fix typo in QU8 VMUL[C] NEON microkernels by Marat Dukhan · 3 years, 4 months ago
- 1e6fc21 Fix incompatible pointer type in QU8 DWCONV NEON microkernels by Marat Dukhan · 3 years, 4 months ago
- 1d90101 Fix GCC incompatibility in QS8/QU8 NEON microkernels by Marat Dukhan · 3 years, 4 months ago
- 8431a06 Include intrinsics polyfill on NEONV8 QS8/QU8 VMUL[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- 2c6d196 Q8 4x16 and 1x16 Neon GEMM/IGEMM quantize using V0-V3 by Frank Barchard · 3 years, 4 months ago
- fbe0c6f Q8 4x16 Neon IGEMM quantize using V0-V3 by Frank Barchard · 3 years, 4 months ago
- f479a1c Initialize QU8 4x16 Neon assembly microkernel for each ARM CPU. by Frank Barchard · 3 years, 4 months ago
- 18f32f5 Expose quantized Multiply operator in Subgraph API by Marat Dukhan · 3 years, 4 months ago
- 0853b8a QS8/QU8 Multiply ND operators by Marat Dukhan · 3 years, 4 months ago
- fb3a94f QU8 4x16 Neon assembly microkernel for Cortex A75 by Frank Barchard · 3 years, 4 months ago
- 4a7b70f QS8/QU8 VMUL[C] microkernels in NEON implementation by Marat Dukhan · 3 years, 4 months ago
- 7999341 QS8/QU8 VMUL[C] microkernels in scalar implementation by Marat Dukhan · 3 years, 4 months ago
- a962f1e Enable QU8 4x16 Neon assembly microkernel by Frank Barchard · 3 years, 4 months ago
- 86a1618 QU8 Neon params replace pad with duplicated zero_point by Frank Barchard · 3 years, 4 months ago
- 59ed1da QU8 4x16 Neon assembly microkernel by Frank Barchard · 3 years, 4 months ago
- 661ea6d QS8/QU8 VMUL[C] microkernels in WAsm SIMD implementation by Marat Dukhan · 3 years, 4 months ago
- a212eac QS8/QU8 VMUL[C] microkernels in SSE2/SSE4.1/AVX implementation by Marat Dukhan · 3 years, 4 months ago
- bea849a QS8 Deconvolution operator by Marat Dukhan · 3 years, 4 months ago
- 6967eb0 Add a rewind variable for params. - no impact on code, just simplified source by Frank Barchard · 3 years, 4 months ago
- eb3cff3 LD128 versions of QS8/QU8 VADD[C] NEON microkernels by Marat Dukhan · 3 years, 4 months ago
- 01debd9 Optimize QS8 VADD[C] microkernel selection on ARM/ARM64 by Marat Dukhan · 3 years, 4 months ago
- 60bb7ec Accumulate in 16 bits once in AVX2 MUL16 VPUNPCK QS8/QC8 DWCONV before extending to 32 bits by Marat Dukhan · 3 years, 4 months ago
- 793c8da QS8 igemm comment for zero use int8_t* instead of float* by Frank Barchard · 3 years, 4 months ago
- 881ab02 AVX2 MUL16 QS8/QC8 DWCONV microkernels using VPUNPCK instructions to extend the product by Marat Dukhan · 3 years, 4 months ago
- 2848059 Optimize QC8 DWCONV microkernel selection on AVX and XOP by Marat Dukhan · 3 years, 4 months ago
- 02f06e3 Fix QS8 DWCONV microkernel selection for XOP processors by Marat Dukhan · 3 years, 4 months ago
- caa7fc7 Optimize selection of QU8 DWCONV microkernel on AVX processors by Marat Dukhan · 3 years, 4 months ago
- 73a899a QU8 DWCONV NEON microkernels with RNDNU requantization by Marat Dukhan · 3 years, 4 months ago
- 173661d QU8 GEMM/IGEMM NEON microkernels with RNDNU requantization by Marat Dukhan · 3 years, 4 months ago
- ab1127f docs: spelling grammar by slowy07 · 3 years, 4 months ago
- 510b8e0 Code generator for RNDNU quantization mode on neon-mull-addw-dup microkernel by Frank Barchard · 3 years, 4 months ago
- 0966856 Accumulate in 16 bits once in SSE2/SSE4/AVX/XOP MUL16 QS8/QC8 DWCONV before extending to 32 bits by Marat Dukhan · 3 years, 4 months ago
- 26e8378 Reduce register pressure in GEMMLOWP quantization on NEON by Frank Barchard · 3 years, 4 months ago
- 28407b2 Support zeroes in shape dimensions in binary elementwise operators by Marat Dukhan · 3 years, 4 months ago
- ab952f1 Remove multiplication in QS8/QC8 DWCONV MUL16 microkernels for SSE4 by Marat Dukhan · 3 years, 4 months ago
- 5f2939f QS8/QC8 DWCONV NEON MUL8/MLA8 microkernels using 128-bit loads by Marat Dukhan · 3 years, 4 months ago
- caccd8e Accumulate in 16 bits once in NEON QS8/QC8 DWCONV before extending to 32 bits by Marat Dukhan · 3 years, 4 months ago
- 1a2dbe1 RNDNU scalar GEMM/IGEMM microkernel by Frank Barchard · 3 years, 4 months ago
- e76049a AVX512 implementation of QS8/QU8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- efa123d Update Neon code with generators for added comment by Frank Barchard · 3 years, 4 months ago
- 6c7b9e8 Disable MSan in quantized addition microkernels by Marat Dukhan · 3 years, 4 months ago
- 9670626 Support QU8 Fully Connected operator in the Subgraph API by Marat Dukhan · 3 years, 4 months ago
- 09a1f65 Support QU8 Add operator in Subgraph API by Marat Dukhan · 3 years, 4 months ago
- 22f9a9f Enable RNDNU requantization for NEON QS8 GEMM/IGEMM by Frank Barchard · 3 years, 4 months ago
- 3eac69c Optimized QU8 VADD[C] microkernels for SSE4/AVX/XOP/AVX2 by Marat Dukhan · 3 years, 4 months ago
- db007cd QU8 Add ND operator by Marat Dukhan · 3 years, 4 months ago
- 76e78c8 Generalize QS8 VADD[C] templates to cover QU8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- 7679b1e Optimize QS8 VADD[C] microkernels for SSE4/AVX/XOP/AVX2 by Marat Dukhan · 3 years, 4 months ago
- 6691324 Split initialization function for QS8 VADD parameters by Marat Dukhan · 3 years, 4 months ago
- 22fbe77 RNDNU quantized 1x16 and 4x16 Neon lane GEMM/IGEMM microkernels. by Frank Barchard · 3 years, 4 months ago
- 288ecd4 Use function pointer to initialize microkernel parameters in QS8 Addition operator by Marat Dukhan · 3 years, 4 months ago
- 13db60f RNDNU quantized Neon assembly GEMM/IGEMM microkernels. by Frank Barchard · 3 years, 4 months ago
- 8a04565 Use RNDNU requantization in QS8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- c3f69fd Simplify requantization in WAsm SIMD QS8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- 947c298 Simplify requantization in SSE2/SSE4/AVX/XOP/AVX2 QS8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- c0612f0 Simplify requantization in NEON QS8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- 60729d0 4x16c4 RNDNU quantized Neon assembly GEMM/IGEMM microkernel. by Frank Barchard · 3 years, 4 months ago
- a842fef Rename zero_point_product parameter to bias in Quantized Add microkernels by Marat Dukhan · 3 years, 4 months ago
- e6a4805 Simplify requantization in scalar QS8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- f0ebd4b Reduce multiplier precision in quantized addition by Marat Dukhan · 3 years, 4 months ago
- 49d9005 Refactor QS8 VADD[C] parameters by Marat Dukhan · 3 years, 4 months ago
- af5843d Optimize QS8 VADD[C] microkernels for SSE2 by Marat Dukhan · 3 years, 4 months ago
- d4c478b Restrict input-to-output scale ratio in quantized addition by Marat Dukhan · 3 years, 4 months ago
- 076bcfe Refactor argument names in QS8 VADD[C] microkernels by Marat Dukhan · 3 years, 4 months ago
- 6e0fc39 Relax initialization of Quantized Addition microkernel parameters by Marat Dukhan · 3 years, 4 months ago
- 575dfb9 Disable MSan in quantized DWCONV microkernels by Marat Dukhan · 3 years, 4 months ago
- 4ba70b7 QS8/QC8 NEON microkernels using 8x8->16-bit multiplication by Marat Dukhan · 3 years, 4 months ago
- 5c92195 Fix incompatibilities with GCC on ARM by Marat Dukhan · 3 years, 4 months ago
- 2bb448c Fix polyfill for vcvtnq_s32_f32 on AArch32 GCC by Marat Dukhan · 3 years, 4 months ago
- e903dff QS8 GEMM/IGEMM microkernels with RNDNU requantization by Marat Dukhan · 3 years, 4 months ago
- be18f5c QS8 DWCONV microkernels with RNDNU requantization by Marat Dukhan · 3 years, 4 months ago
- f975d7f Fix instruction listings in NEON requantization stubs by Marat Dukhan · 3 years, 4 months ago
- d3d818c Fix requantization stubs for Ruy requantization schema by Marat Dukhan · 3 years, 4 months ago
- 7b1aeb9 Evaluation stubs for Ruy requantization schema by Marat Dukhan · 3 years, 4 months ago
- 2837e8b Remove 0 offset from loads. by Frank Barchard · 3 years, 4 months ago
- d194311 4x16c4-aarch64-neondot-ld32 use LD1R instead of lanes by Frank Barchard · 3 years, 4 months ago
- 89cd59b Remove legacy QU8 GEMM/IGEMM/DWCONV microkernels by Marat Dukhan · 3 years, 4 months ago
- 43b46ee Use generated QU8 GEMM/IGEMM/DWCONV microkernels on ARM by Marat Dukhan · 3 years, 4 months ago
- 3d76e55 Reoptimize microkernel selection for WAsm MVP by Marat Dukhan · 3 years, 4 months ago
- 8172135 Use generated QU8 GEMM/IGEMM/DWCONV microkernels on ARM64 by Marat Dukhan · 3 years, 4 months ago
- 605696a NEON implementations of QU8 DWCONV microkernels by Marat Dukhan · 3 years, 4 months ago
- a97e975 Initialize QU8 microkernels for WebAssembly SIMD by Marat Dukhan · 3 years, 4 months ago
- 1f71428 Scalar implementations of QU8 DWCONV microkernels by Marat Dukhan · 3 years, 4 months ago
- f601135 WAsm SIMD implementations of QU8 DWCONV microkernels by Marat Dukhan · 3 years, 4 months ago
- 927d474 Scalar implementations of QU8 GEMM/IGEMM microkernels by Marat Dukhan · 3 years, 4 months ago
- 43bee05 WAsm SIMD implementation of QU8 GEMM/IGEMM microkernels by Marat Dukhan · 3 years, 4 months ago
- 69c8a29 NEON-MLAL implementations of QU8 GEMM/IGEMM microkernels by Marat Dukhan · 3 years, 4 months ago
- f6f6209 Refactoring in QS8/QC8 GEMM/IGEMM NEON-MLAL microkernels by Marat Dukhan · 3 years, 4 months ago
- a98a109 Fix missing braces around initializer warning by Nikita Shulga · 3 years, 4 months ago
- 8c8c159 Expose QU8 [Depthwise] Convolution 2D operators in Subgraph API by Marat Dukhan · 3 years, 4 months ago
- 8c8ce5d Include vcvtnq_s32_f32 polyfill in NEON requantization stubs by Marat Dukhan · 3 years, 4 months ago