sglang_v0.5.2/pytorch_2.8.0/third_party/XNNPACK/test/qs8-qc8w-gemm-minmax-fp32.yaml

1716 lines
70 KiB
YAML

# Copyright 2021 Google LLC
#
# This source code is licensed under the BSD-style license found in the
# LICENSE file in the root directory of this source tree.
# x86 AVX512AMX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_16x16c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_16x16c4__avx512amx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x32c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x32c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_16x32c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_16x32c4__avx512amx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x64c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x64c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_16x64c4__avx512amx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_16x64c4__avx512amx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 64
# AArch32 assembly
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__asm_aarch32_neon_mlal_lane_cortex_a7
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__asm_aarch32_neon_mlal_lane_cortex_a7_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__asm_aarch32_neonv8_mlal_lane_cortex_a35
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__asm_aarch32_neonv8_mlal_lane_cortex_a35_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neon_mlal_lane_cortex_a7
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neon_mlal_lane_cortex_a7_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neon_mlal_lane_cortex_a53
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neon_mlal_lane_cortex_a53_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neon_mlal_lane_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neon_mlal_lane_ld64_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neonv8_mlal_lane_cortex_a35
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neonv8_mlal_lane_cortex_a35_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neonv8_mlal_lane_cortex_a53
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neonv8_mlal_lane_cortex_a53_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neonv8_mlal_lane_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__asm_aarch32_neonv8_mlal_lane_ld64_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c4__asm_aarch32_neondot_cortex_a55
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c4__asm_aarch32_neondot_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# AArch64 assembly
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__asm_aarch64_neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__asm_aarch64_neon_mlal_cortex_a53
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__asm_aarch64_neon_mlal_cortex_a53_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__asm_aarch64_neon_mlal_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__asm_aarch64_neondot_ld32
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 4
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__asm_aarch64_neondot_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__asm_aarch64_neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__asm_aarch64_neon_mlal_cortex_a53
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__asm_aarch64_neon_mlal_cortex_a53_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__asm_aarch64_neon_mlal_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__asm_aarch64_neon_mull
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c16__asm_aarch64_neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__asm_aarch64_neon_mlal_lane_cortex_a53
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__asm_aarch64_neon_mlal_lane_cortex_a53_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__asm_aarch64_neon_mlal_lane_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__asm_aarch64_neon_mlal_lane_ld64_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__asm_aarch64_neondot_cortex_a55
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__asm_aarch64_neondot_ld32
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 4
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__asm_aarch64_neondot_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__asm_aarch64_neondot_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
# ARMv6 SIMD
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x1c4__armsimd32
init: xnn_init_qs8_qc8w_conv_minmax_fp32_armsimd32_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 4
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x2c4__armsimd32
init: xnn_init_qs8_qc8w_conv_minmax_fp32_armsimd32_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 4
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x1c4__armsimd32
init: xnn_init_qs8_qc8w_conv_minmax_fp32_armsimd32_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 4
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x2c4__armsimd32
init: xnn_init_qs8_qc8w_conv_minmax_fp32_armsimd32_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 4
# ARM NEONI8MM
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x16c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x16c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x8c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c8__neoni8mm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
# ARM NEON
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neon_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neon_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neon_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neon_mlal_ld4r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neonv8_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neonv8_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neonv8_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2__neonv8_mlal_ld4r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2s4__neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c2s4__neonv8_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neon_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neon_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neon_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neonv8_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neonv8_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4__neonv8_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4s2__neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c4s2__neonv8_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__aarch64_neondot_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__neondot_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__neonv8_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__aarch64_neondot_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__neondot_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neon_mlal_ld4r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2__neonv8_mlal_ld4r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2s4__neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c2s4__neonv8_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4__neon_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_dup
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld1r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4__neonv8_mlal_ld2r
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4s2__neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c4s2__neonv8_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__neon_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__neonv8_mlal
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x16__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x16__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x16__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x16__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x16__neon_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x16__neon_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neon_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x16__neonv8_mlal_lane
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x16__neonv8_mlal_lane_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x16c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x8c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c4__neondot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_neonv8_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 SSE
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 SSE
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 SSE
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 SSE
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 SSE
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__sse2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__sse41_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__avx_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 SSE
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__sse2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__sse41_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__avx_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX2
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avx2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__avx2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__avx2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__avx2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX256SKX
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avx256skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__avx256skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__avx256skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__avx256skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX512
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__avx512skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x16c8__avx512skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c8__avx512skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c8__avx512skx
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__avx512skx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x16c8__avx512skx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c8__avx512skx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c8__avx512skx_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# x86 AVX512VNNI
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_9x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_10x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_12x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_14x16c4__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_9x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_10x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_12x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_14x16c4__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_9x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_10x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_12x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_14x16c8__avx512vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_9x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_10x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_12x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_14x16c8__avx512vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_9x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_10x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_12x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_14x8c8__avx256vnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_9x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_10x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_12x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_14x8c8__avx256vnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
# AVXVNNI
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x8c8__avxvnni
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_6x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_7x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_8x8c8__avxvnni_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
# AVXVNNIINT8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__avxvnniint8_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_5x8c8__avxvnniint8_prfm
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
# WAsm SIMD
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c2s4__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c8__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c2s4__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c8__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c2s4__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c8__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c2s4__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c8__wasmsimd_dot16x2_ld64
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c8__wasmsimd_dot16x2_ld128
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
# WAsm Relaxed SIMD
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c16__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c16__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 16
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x8c8__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x8c8__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x8c8__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x8c8__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__wasmusdot_u2_acc2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c4__wasmusdot_u2_acc2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c4__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c4__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__wasmusdot_u2_acc2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__wasmusdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__wasmusdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x16c4__wasmsdot_u2_acc2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c4__wasmsdot_u2_acc2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c4__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x16c4__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__wasmsdot_u2_acc2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__wasmsdot
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x16c4__wasmsdot_u2
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_to_qu8_gemm_goi_w
k-block: 8
# WAsm
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x2__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x2__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x2__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x2__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4__wasm_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
# Scalar
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x2__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x2__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x2__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_1x4__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x2__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x2__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x2__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_2x4__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x2__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x2__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x2__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_3x4__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x2__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x2__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x2__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4__scalar_fmagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4__scalar_imagic
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1
- name: xnn_qs8_qc8w_gemm_minmax_fp32_ukernel_4x4__scalar_lrintf
init: xnn_init_qs8_qc8w_conv_minmax_fp32_scalar_params
pack: xnn_pack_qs8_gemm_goi_w
k-block: 1