5af2ae34 · scalar qs8 rsum accumulating microkernels · May 13, 2024

909d30c7 · neon mlal qs8 rsum accumulating microkernels · May 13, 2024

edc2946b · Enable -mavx512fp16 needed for avx512fp16 microkernels · May 14, 2024

7c53581e · F16-GEMM-MINMAX-TEST use x16_packw microkernels · May 14, 2024

79a50691 · Enable F16C microkernels on Ivybridge · May 14, 2024

6e6dd409 · Add partial support for building/testing/benchmarking XNNPACK on Hexagon.... · May 15, 2024

30080c2c · Enable fp16 for risc-v · May 15, 2024

be6c51bf · Mean tests both keep dims and not keep dims. · May 16, 2024

a1a5016c · Use a better error bound for `fp16` tests of the `rsum` microkernel. · May 16, 2024

9ec06e13 · AVX2 qs8 rsum use vpmovsxbd to read bytes as ints · May 20, 2024

b4bab69f · avx512 qs8 rsum accumulating microkernels · May 21, 2024

fe05ed25 · Run templates to regenerate tests and benchmarks · May 21, 2024

94c42a89 · Disable F32-RSUM on RISCV to work around multithreads mean test failure · May 21, 2024

fc0787a9 · Introduce `SIMD` headers which provide arch-specific wrappers for some common operations. · May 24, 2024

7b61b39a · Clean up the `xnn_f32_abs_params` and `xnn_f32_neg_params` that are no longer needed. · May 24, 2024

dc13720d · Add a `xnn_datatype_qpint8` datatype for packed per-row dynamically quantized... · May 27, 2024

f5e33fb5 · Extend the `convert` operator for the new `qp8` packed per-row dynamic quantization. · May 27, 2024

9d5e952a · Add a new `x8-packq` microkernel that packs and per-row dynamically quantizes `fp32` to `qp8`. · May 27, 2024

b7dde43e · QS8-RSUM AVX2 use vpmaddubsw · May 27, 2024

2b977e36 · Remove XNN_TEST_MODE, it appears to be unnecessary, and we should test the... · May 27, 2024