You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
1992 lines
44 KiB
1992 lines
44 KiB
// Copyright (c) Facebook, Inc. and its affiliates.
|
|
// All rights reserved.
|
|
//
|
|
// 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.
|
|
//
|
|
// Auto-generated file. Do not edit!
|
|
// Specification: test/qu8-igemm-minmax.yaml
|
|
// Generator: tools/generate-gemm-test.py
|
|
|
|
|
|
#include <gtest/gtest.h>
|
|
|
|
#include <xnnpack/common.h>
|
|
#include <xnnpack/isa-checks.h>
|
|
|
|
#include <xnnpack/gemm.h>
|
|
#include <xnnpack/igemm.h>
|
|
#include <xnnpack/ppmm.h>
|
|
#include "gemm-microkernel-tester.h"
|
|
|
|
|
|
#if XNN_ARCH_ARM || XNN_ARCH_ARM64
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_eq_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(8)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, strided_cn) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(8)
|
|
.cn_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_eq_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_eq_8_subtile_m) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(8)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_eq_8_subtile_n) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(n)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_lt_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k < 8; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_lt_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k < 8; k++) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_gt_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 9; k < 16; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_gt_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 9; k < 16; k++) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_div_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 16; k <= 80; k += 8) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, k_div_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 16; k <= 80; k += 8) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_gt_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_gt_8_strided_cn) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.cn_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_gt_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_div_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_div_8_strided_cn) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(n)
|
|
.k(k)
|
|
.cn_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_div_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, small_kernel) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, small_kernel_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.ks(3)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_gt_8_small_kernel) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, n_div_8_small_kernel) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, strided_cm_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.cm_stride(11)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, a_offset) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(163)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, zero) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t mz = 0; mz < 4; mz++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(163)
|
|
.zero_index(mz)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, qmin) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(8)
|
|
.qmin(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, qmax) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(8)
|
|
.qmax(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, strided_cm) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(8)
|
|
.cm_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, no_a_zero_point) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, no_b_zero_point) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X8__NEON, no_zero_point) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(8)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x8__neon);
|
|
}
|
|
}
|
|
#endif // XNN_ARCH_ARM || XNN_ARCH_ARM64
|
|
|
|
|
|
#if XNN_ARCH_ARM || XNN_ARCH_ARM64
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_eq_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(8)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, strided_cn) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(8)
|
|
.cn_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_eq_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_eq_8_subtile_m) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(8)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_eq_8_subtile_n) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(n)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_lt_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k < 8; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_lt_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k < 8; k++) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_gt_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 9; k < 16; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_gt_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 9; k < 16; k++) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_div_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 16; k <= 80; k += 8) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, k_div_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 16; k <= 80; k += 8) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_gt_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_gt_8_strided_cn) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.cn_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_gt_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_div_8) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_div_8_strided_cn) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(n)
|
|
.k(k)
|
|
.cn_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_div_8_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, small_kernel) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, small_kernel_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.ks(3)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_gt_8_small_kernel) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 9; n < 16; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, n_div_8_small_kernel) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t n = 16; n <= 24; n += 8) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, strided_cm_subtile) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 8; m++) {
|
|
for (uint32_t n = 1; n <= 8; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.cm_stride(11)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, a_offset) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(331)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, zero) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (uint32_t mz = 0; mz < 8; mz++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(331)
|
|
.zero_index(mz)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, qmin) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(8)
|
|
.qmin(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, qmax) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(8)
|
|
.qmax(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, strided_cm) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(8)
|
|
.cm_stride(11)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, no_a_zero_point) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, no_b_zero_point) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_8X8__NEON, no_zero_point) {
|
|
TEST_REQUIRES_ARM_NEON;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(8)
|
|
.nr(8)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(8)
|
|
.n(8)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_8x8__neon);
|
|
}
|
|
}
|
|
#endif // XNN_ARCH_ARM || XNN_ARCH_ARM64
|
|
|
|
|
|
#if XNN_ARCH_X86 || XNN_ARCH_X86_64
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_eq_8) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(8)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, strided_cn) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(8)
|
|
.cn_stride(7)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_eq_8_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_eq_8_subtile_m) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(4)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_eq_8_subtile_n) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(n)
|
|
.k(8)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_lt_8) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k < 8; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_lt_8_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k < 8; k++) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_gt_8) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 9; k < 16; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_gt_8_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 9; k < 16; k++) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_div_8) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 16; k <= 80; k += 8) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, k_div_8_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 16; k <= 80; k += 8) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_gt_4) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 5; n < 8; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_gt_4_strided_cn) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 5; n < 8; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.cn_stride(7)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_gt_4_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 5; n < 8; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_div_4) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 8; n <= 12; n += 4) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_div_4_strided_cn) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 8; n <= 12; n += 4) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(n)
|
|
.k(k)
|
|
.cn_stride(7)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_div_4_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 8; n <= 12; n += 4) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, small_kernel) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, small_kernel_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.ks(3)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_gt_4_small_kernel) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 5; n < 8; n++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, n_div_4_small_kernel) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t n = 8; n <= 12; n += 4) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, strided_cm_subtile) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
for (uint32_t m = 1; m <= 4; m++) {
|
|
for (uint32_t n = 1; n <= 4; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.cm_stride(7)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, a_offset) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(163)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, zero) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (uint32_t mz = 0; mz < 4; mz++) {
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(163)
|
|
.zero_index(mz)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, qmin) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(8)
|
|
.qmin(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, qmax) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(8)
|
|
.qmax(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, strided_cm) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(8)
|
|
.cm_stride(7)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, no_a_zero_point) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, no_b_zero_point) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_4X4C2__SSE2, no_zero_point) {
|
|
TEST_REQUIRES_X86_SSE2;
|
|
for (size_t k = 1; k <= 40; k += 9) {
|
|
GemmMicrokernelTester()
|
|
.mr(4)
|
|
.nr(4)
|
|
.kr(2)
|
|
.sr(1)
|
|
.m(4)
|
|
.n(4)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_4x4c2__sse2);
|
|
}
|
|
}
|
|
#endif // XNN_ARCH_X86 || XNN_ARCH_X86_64
|
|
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, k_eq_1) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, strided_cn) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(1)
|
|
.cn_stride(5)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, k_eq_1_subtile) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
for (uint32_t n = 1; n <= 2; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(1)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, k_eq_1_subtile_m) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(2)
|
|
.k(1)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, k_eq_1_subtile_n) {
|
|
for (uint32_t n = 1; n <= 2; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(n)
|
|
.k(1)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, k_gt_1) {
|
|
for (size_t k = 2; k < 10; k++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, k_gt_1_subtile) {
|
|
for (size_t k = 2; k < 10; k++) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
for (uint32_t n = 1; n <= 2; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_gt_2) {
|
|
for (uint32_t n = 3; n < 4; n++) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_gt_2_strided_cn) {
|
|
for (uint32_t n = 3; n < 4; n++) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.cn_stride(5)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_gt_2_subtile) {
|
|
for (uint32_t n = 3; n < 4; n++) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_div_2) {
|
|
for (uint32_t n = 4; n <= 6; n += 2) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_div_2_strided_cn) {
|
|
for (uint32_t n = 4; n <= 6; n += 2) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(n)
|
|
.k(k)
|
|
.cn_stride(5)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_div_2_subtile) {
|
|
for (uint32_t n = 4; n <= 6; n += 2) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, small_kernel) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, small_kernel_subtile) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
for (uint32_t n = 1; n <= 2; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.ks(3)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_gt_2_small_kernel) {
|
|
for (uint32_t n = 3; n < 4; n++) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, n_div_2_small_kernel) {
|
|
for (uint32_t n = 4; n <= 6; n += 2) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.ks(3)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, strided_cm_subtile) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
for (uint32_t m = 1; m <= 2; m++) {
|
|
for (uint32_t n = 1; n <= 2; n++) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(m)
|
|
.n(n)
|
|
.k(k)
|
|
.cm_stride(5)
|
|
.iterations(1)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, a_offset) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(13)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, zero) {
|
|
for (uint32_t mz = 0; mz < 2; mz++) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.ks(3)
|
|
.a_offset(13)
|
|
.zero_index(mz)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, qmin) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(1)
|
|
.qmin(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, qmax) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(1)
|
|
.qmax(128)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, strided_cm) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(1)
|
|
.cm_stride(5)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, no_a_zero_point) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, no_b_zero_point) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
}
|
|
|
|
TEST(QU8_IGEMM_MINMAX_2X2__SCALAR, no_zero_point) {
|
|
for (size_t k = 1; k <= 5; k += 2) {
|
|
GemmMicrokernelTester()
|
|
.mr(2)
|
|
.nr(2)
|
|
.kr(1)
|
|
.sr(1)
|
|
.m(2)
|
|
.n(2)
|
|
.k(k)
|
|
.a_zero_point(0)
|
|
.b_zero_point(0)
|
|
.Test(xnn_qu8_igemm_minmax_ukernel_2x2__scalar, GemmMicrokernelTester::Variant::Scalar);
|
|
}
|
|
} |