sglang_v0.5.2/pytorch_2.8.0/third_party/XNNPACK/test/f32-gemm-minmax.yaml

1123 lines
40 KiB
YAML

# Copyright 2019 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.
# AArch32 assembly
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch32_neon_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch32_neon_cortex_a53_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x4__asm_aarch32_vfp_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_cortex_a7
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_cortex_a53_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_cortex_a55
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_cortex_a75
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_cortex_a75_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch32_neon_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
# AArch64 assembly
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neon_ld128_acc2
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neon_ld128_acc2_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_cortex_a53_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_cortex_a75
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_cortex_a75_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld64_acc2
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld64_acc2_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld64_acc4
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld64_acc4_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld64_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld128_acc2
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld128_acc2_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld128_acc4
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld128_acc4_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch64_neonfma_ld128_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x12__asm_aarch64_neonfma_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x1__asm_aarch64_neonfma_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x1__asm_aarch64_neonfma_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2__asm_aarch64_neonfma_cortex_a75
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
- name: xnn_f32_gemm_minmax_ukernel_4x2__asm_aarch64_neonfma_cortex_a75_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
- name: xnn_f32_gemm_minmax_ukernel_4x2__asm_aarch64_neonfma_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x2__asm_aarch64_neonfma_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_cortex_a53_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_cortex_a55
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_cortex_a75
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_cortex_a75_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__asm_aarch64_neonfma_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x12__asm_aarch64_neonfma_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_5x8__asm_aarch64_neonfma_cortex_a75
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_5x8__asm_aarch64_neonfma_cortex_a75_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_cortex_a53
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_cortex_a53_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_cortex_a55
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_cortex_a73
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_cortex_a75
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_cortex_a75_prfm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 8
pipelined: true
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x8__asm_aarch64_neonfma_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
# ARM NEON
- name: xnn_f32_gemm_minmax_ukernel_1x16__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_2x16__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x16__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x16__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x16__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x16__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x16__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_2x16__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x16__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x16__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x16__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x16__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__neon_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_lane_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__aarch64_neonfma_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__neon_dup_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__neon_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8__neonfma_dup_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__neon
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__neonfma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2__aarch64_neonfma_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x2__neon_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__aarch64_neonfma_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__neon_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__neon
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__neonfma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__aarch64_neonfma_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_5x8__neon_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x2__aarch64_neonfma_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x2__neon_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x8__aarch64_neonfma_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__neon_lane_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 2
- name: xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__neon
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__neonfma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_8x8s4__neon
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_8x8s4__neonfma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
# x86 SSE
- name: xnn_f32_gemm_minmax_ukernel_1x8__sse_dup
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__sse_load1
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8__sse_dup
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8__sse_load1
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x8s4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2c4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__sse_dup
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__sse_load1
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__sse_dup
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__sse_load1
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x8s4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x2c4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__sse_dup
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__sse_load1
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__sse
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
# x86 AVX
- name: xnn_f32_gemm_minmax_ukernel_1x8__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x16__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x16__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x8__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x16__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x8__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x16__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x8__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x16__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_7x8__avx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
# x86 FMA3
- name: xnn_f32_gemm_minmax_ukernel_1x8__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x16__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x16s4__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x16__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x16s4__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x16__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x16s4__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x16__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x16s4__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x16__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x16s4__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_7x8__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_8x8__fma3_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
# x86 AVX512
- name: xnn_f32_gemm_minmax_ukernel_1x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_7x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_8x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_9x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_10x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_11x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_12x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_13x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_14x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_15x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_16x16__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_7x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_8x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_9x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_10x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_11x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_12x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_13x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_14x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_15x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_16x32__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_7x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_8x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_9x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_10x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_11x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_12x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_13x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_14x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_15x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_16x64__avx512f_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
# WAsm SIMD
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__wasmsimd_arm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__wasmsimd_x86
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8s4__wasmsimd_arm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8s4__wasmsimd_x86
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2c4__wasmsimd_arm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2c4__wasmsimd_x86
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__wasmsimd_arm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__wasmsimd_x86
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8s4__wasmsimd_arm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8s4__wasmsimd_x86
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__wasmsimd_arm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__wasmsimd_x86
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
# WAsm Relaxed SIMD
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmrelaxedsimd_fma_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmrelaxedsimd_fma_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmrelaxedsimd_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x8__wasmrelaxedsimd_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__wasmrelaxedsimd
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_1x8s4__wasmrelaxedsimd_fma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_fma_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_fma_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_3x8__wasmrelaxedsimd_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8s4__wasmrelaxedsimd
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_3x8s4__wasmrelaxedsimd_fma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2c4__wasmrelaxedsimd
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x2c4__wasmrelaxedsimd_fma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__wasmrelaxedsimd
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_4x8s4__wasmrelaxedsimd_fma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_fma_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_fma_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_5x8__wasmrelaxedsimd_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8s4__wasmrelaxedsimd
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_5x8s4__wasmrelaxedsimd_fma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_fma_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_fma_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_loadsplat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_6x8__wasmrelaxedsimd_splat
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__wasmrelaxedsimd
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
- name: xnn_f32_gemm_minmax_ukernel_6x8s4__wasmrelaxedsimd_fma
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 4
# WAsm
- name: xnn_f32_gemm_minmax_ukernel_1x4__wasm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_2x4__wasm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x2__wasm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x4__wasm
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
# Scalar
- name: xnn_f32_gemm_minmax_ukernel_1x4__scalar
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_2x4__scalar
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x2__scalar
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x4__scalar
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
# RISC-V
- name: xnn_f32_gemm_minmax_ukernel_1x4v__rvv
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_7x4v__rvv
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
# HEXAGON
- name: xnn_f32_gemm_minmax_ukernel_1x32__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x64__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_1x128__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_2x128__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_4x64__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_7x64__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_8x32__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1
- name: xnn_f32_gemm_minmax_ukernel_16x32__hvx_broadcast
init: xnn_init_f32_minmax_scalar_params
pack: xnn_pack_f32_gemm_goi_w
k-block: 1