head 1.1; branch 1.1.1; access; symbols netbsd-11-0-RC4:1.1.1.3 netbsd-11-0-RC3:1.1.1.3 netbsd-11-0-RC2:1.1.1.3 netbsd-11-0-RC1:1.1.1.3 netbsd-11:1.1.1.3.0.4 netbsd-11-base:1.1.1.3 netbsd-10-1-RELEASE:1.1.1.3 netbsd-9-4-RELEASE:1.1.1.1 netbsd-10-0-RELEASE:1.1.1.3 netbsd-10-0-RC6:1.1.1.3 netbsd-10-0-RC5:1.1.1.3 netbsd-10-0-RC4:1.1.1.3 netbsd-10-0-RC3:1.1.1.3 netbsd-10-0-RC2:1.1.1.3 netbsd-10-0-RC1:1.1.1.3 netbsd-10:1.1.1.3.0.2 netbsd-10-base:1.1.1.3 netbsd-9-3-RELEASE:1.1.1.1 mesa-21-3-7:1.1.1.3 netbsd-9-2-RELEASE:1.1.1.1 netbsd-9-1-RELEASE:1.1.1.1 netbsd-9-0-RELEASE:1.1.1.1 netbsd-9-0-RC2:1.1.1.1 netbsd-9-0-RC1:1.1.1.1 mesalib-19-1-7:1.1.1.2 netbsd-9:1.1.1.1.0.2 netbsd-9-base:1.1.1.1 mesa-18-3-6:1.1.1.1 mesa-18-3-4:1.1.1.1 xorg:1.1.1; locks; strict; comment @// @; 1.1 date 2019.03.10.03.42.41; author mrg; state Exp; branches 1.1.1.1; next ; commitid r12jo1Nf3ebQKLeB; 1.1.1.1 date 2019.03.10.03.42.41; author mrg; state Exp; branches; next 1.1.1.2; commitid r12jo1Nf3ebQKLeB; 1.1.1.2 date 2019.09.24.17.41.20; author maya; state Exp; branches; next 1.1.1.3; commitid KJXusGl8fi9AAhEB; 1.1.1.3 date 2022.05.09.01.23.37; author mrg; state Exp; branches; next ; commitid UEBs6hNk81DdQjDD; desc @@ 1.1 log @Initial revision @ text @/* * Copyright © 2015 Intel Corporation * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS * IN THE SOFTWARE. */ #include #include "brw_fs.h" #include "brw_cfg.h" #include "program/program.h" using namespace brw; class saturate_propagation_test : public ::testing::Test { virtual void SetUp(); public: struct brw_compiler *compiler; struct gen_device_info *devinfo; struct gl_context *ctx; struct brw_wm_prog_data *prog_data; struct gl_shader_program *shader_prog; fs_visitor *v; }; class saturate_propagation_fs_visitor : public fs_visitor { public: saturate_propagation_fs_visitor(struct brw_compiler *compiler, struct brw_wm_prog_data *prog_data, nir_shader *shader) : fs_visitor(compiler, NULL, NULL, NULL, &prog_data->base, (struct gl_program *) NULL, shader, 8, -1) {} }; void saturate_propagation_test::SetUp() { ctx = (struct gl_context *)calloc(1, sizeof(*ctx)); compiler = (struct brw_compiler *)calloc(1, sizeof(*compiler)); devinfo = (struct gen_device_info *)calloc(1, sizeof(*devinfo)); compiler->devinfo = devinfo; prog_data = ralloc(NULL, struct brw_wm_prog_data); nir_shader *shader = nir_shader_create(NULL, MESA_SHADER_FRAGMENT, NULL, NULL); v = new saturate_propagation_fs_visitor(compiler, prog_data, shader); devinfo->gen = 6; } static fs_inst * instruction(bblock_t *block, int num) { fs_inst *inst = (fs_inst *)block->start(); for (int i = 0; i < num; i++) { inst = (fs_inst *)inst->next; } return inst; } static bool saturate_propagation(fs_visitor *v) { const bool print = false; if (print) { fprintf(stderr, "= Before =\n"); v->cfg->dump(v); } bool ret = v->opt_saturate_propagation(); if (print) { fprintf(stderr, "\n= After =\n"); v->cfg->dump(v); } return ret; } TEST_F(saturate_propagation_test, basic) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1); set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: add(8) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * * = After = * 0: add.sat(8) dst0 src0 src1 * 1: mov(8) dst1 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, other_non_saturated_use) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg dst2 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1); set_saturate(true, bld.MOV(dst1, dst0)); bld.ADD(dst2, dst0, src0); /* = Before = * * 0: add(8) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * 2: add(8) dst2 dst0 src0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 2)->opcode); } TEST_F(saturate_propagation_test, predicated_instruction) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1) ->predicate = BRW_PREDICATE_NORMAL; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: (+f0) add(8) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); bld.RNDU(dst0, src0); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: rndu(8) dst0 src0 * 1: mov.sat(8) dst1 -dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_RNDU, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, add_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: add(8) dst0 src0 src1 * 1: mov.sat(8) dst1 -dst0 * * = After = * 0: add.sat(8) dst0 -src0 -src1 * 1: mov(8) dst1 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_TRUE(instruction(block0, 0)->src[0].negate); EXPECT_TRUE(instruction(block0, 0)->src[1].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, add_imm_float_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = brw_imm_f(1.0f); bld.ADD(dst0, src0, src1); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: add(8) dst0 src0 1.0f * 1: mov.sat(8) dst1 -dst0 * * = After = * 0: add.sat(8) dst0 -src0 -1.0f * 1: mov(8) dst1 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_TRUE(instruction(block0, 0)->src[0].negate); EXPECT_EQ(instruction(block0, 0)->src[1].f, -1.0f); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, mul_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.MUL(dst0, src0, src1); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: mul(8) dst0 src0 src1 * 1: mov.sat(8) dst1 -dst0 * * = After = * 0: mul.sat(8) dst0 src0 -src1 * 1: mov(8) dst1 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_MUL, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_TRUE(instruction(block0, 0)->src[0].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); EXPECT_FALSE(instruction(block0, 1)->src[0].negate); } TEST_F(saturate_propagation_test, mad_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); fs_reg src2 = v->vgrf(glsl_type::float_type); bld.MAD(dst0, src0, src1, src2); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: mad(8) dst0 src0 src1 src2 * 1: mov.sat(8) dst1 -dst0 * * = After = * 0: mad.sat(8) dst0 -src0 -src1 src2 * 1: mov(8) dst1 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_MAD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_TRUE(instruction(block0, 0)->src[0].negate); EXPECT_TRUE(instruction(block0, 0)->src[1].negate); EXPECT_FALSE(instruction(block0, 0)->src[2].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); EXPECT_FALSE(instruction(block0, 1)->src[0].negate); } TEST_F(saturate_propagation_test, mad_imm_float_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = brw_imm_f(1.0f); fs_reg src1 = brw_imm_f(-2.0f); fs_reg src2 = v->vgrf(glsl_type::float_type); /* The builder for MAD tries to be helpful and not put immediates as direct * sources. We want to test specifically that case. */ fs_inst *mad = bld.MAD(dst0, src2, src2, src2); mad->src[0]= src0; mad->src[1] = src1; dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: mad(8) dst0 1.0f -2.0f src2 * 1: mov.sat(8) dst1 -dst0 * * = After = * 0: mad.sat(8) dst0 -1.0f 2.0f src2 * 1: mov(8) dst1 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_MAD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_EQ(instruction(block0, 0)->src[0].f, -1.0f); EXPECT_EQ(instruction(block0, 0)->src[1].f, 2.0f); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); EXPECT_FALSE(instruction(block0, 1)->src[0].negate); } TEST_F(saturate_propagation_test, mul_mov_sat_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg dst2 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.MUL(dst0, src0, src1); set_saturate(true, bld.MOV(dst1, dst0)); dst0.negate = true; set_saturate(true, bld.MOV(dst2, dst0)); /* = Before = * * 0: mul(8) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * 2: mov.sat(8) dst2 -dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_MUL, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_FALSE(instruction(block0, 0)->src[1].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 2)->opcode); EXPECT_TRUE(instruction(block0, 2)->src[0].negate); EXPECT_TRUE(instruction(block0, 2)->saturate); } TEST_F(saturate_propagation_test, mul_neg_mov_sat_neg_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg dst2 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.MUL(dst0, src0, src1); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); set_saturate(true, bld.MOV(dst2, dst0)); /* = Before = * * 0: mul(8) dst0 src0 src1 * 1: mov.sat(8) dst1 -dst0 * 2: mov.sat(8) dst2 -dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_MUL, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_FALSE(instruction(block0, 0)->src[1].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->src[0].negate); EXPECT_TRUE(instruction(block0, 1)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 2)->opcode); EXPECT_TRUE(instruction(block0, 2)->src[0].negate); EXPECT_TRUE(instruction(block0, 2)->saturate); } TEST_F(saturate_propagation_test, abs_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1); dst0.abs = true; set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: add(8) dst0 src0 src1 * 1: mov.sat(8) dst1 (abs)dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, producer_saturates) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg dst2 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); set_saturate(true, bld.ADD(dst0, src0, src1)); set_saturate(true, bld.MOV(dst1, dst0)); bld.MOV(dst2, dst0); /* = Before = * * 0: add.sat(8) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * 2: mov(8) dst2 dst0 * * = After = * 0: add.sat(8) dst0 src0 src1 * 1: mov(8) dst1 dst0 * 2: mov(8) dst2 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, intervening_saturating_copy) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg dst2 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1); set_saturate(true, bld.MOV(dst1, dst0)); set_saturate(true, bld.MOV(dst2, dst0)); /* = Before = * * 0: add(8) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * 2: mov.sat(8) dst2 dst0 * * = After = * 0: add.sat(8) dst0 src0 src1 * 1: mov(8) dst1 dst0 * 2: mov(8) dst2 dst0 */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_TRUE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_TRUE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 1)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 2)->opcode); EXPECT_FALSE(instruction(block0, 2)->saturate); } TEST_F(saturate_propagation_test, intervening_dest_write) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::vec4_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); fs_reg src2 = v->vgrf(glsl_type::vec2_type); bld.ADD(offset(dst0, bld, 2), src0, src1); bld.emit(SHADER_OPCODE_TEX, dst0, src2) ->size_written = 4 * REG_SIZE; set_saturate(true, bld.MOV(dst1, offset(dst0, bld, 2))); /* = Before = * * 0: add(8) dst0+2 src0 src1 * 1: tex(8) rlen 4 dst0+0 src2 * 2: mov.sat(8) dst1 dst0+2 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(SHADER_OPCODE_TEX, instruction(block0, 1)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 2)->opcode); EXPECT_TRUE(instruction(block0, 2)->saturate); } TEST_F(saturate_propagation_test, mul_neg_mov_sat_mov_sat) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg dst2 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.MUL(dst0, src0, src1); dst0.negate = true; set_saturate(true, bld.MOV(dst1, dst0)); dst0.negate = false; set_saturate(true, bld.MOV(dst2, dst0)); /* = Before = * * 0: mul(8) dst0 src0 src1 * 1: mov.sat(8) dst1 -dst0 * 2: mov.sat(8) dst2 dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(2, block0->end_ip); EXPECT_EQ(BRW_OPCODE_MUL, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_FALSE(instruction(block0, 0)->src[1].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); EXPECT_TRUE(instruction(block0, 1)->src[0].negate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 2)->opcode); EXPECT_TRUE(instruction(block0, 2)->saturate); } @ 1.1.1.1 log @from maya: Import mesa 18.3.4. Mesa 18.3.4 implements the OpenGL 4.5 API. Some drivers don't support all the features required in OpenGL 4.5. @ text @@ 1.1.1.2 log @Import mesa 19.1.7 New features in mesa 19.1.0: GL_ARB_parallel_shader_compile on all drivers. GL_EXT_gpu_shader4 on all GL 3.1 drivers. GL_EXT_shader_image_load_formatted on radeonsi. GL_EXT_texture_buffer_object on all GL 3.1 drivers. GL_EXT_texture_compression_s3tc_srgb on Gallium drivers and i965 (ES extension). GL_NV_compute_shader_derivatives on iris and i965. GL_KHR_parallel_shader_compile on all drivers. VK_EXT_buffer_device_address on Intel and RADV. VK_EXT_depth_clip_enable on Intel and RADV. VK_KHR_ycbcr_image_arrays on Intel. VK_EXT_inline_uniform_block on Intel and RADV. VK_EXT_external_memory_host on Intel. VK_EXT_host_query_reset on Intel and RADV. VK_KHR_surface_protected_capabilities on Intel and RADV. VK_EXT_pipeline_creation_feedback on Intel and RADV. VK_KHR_8bit_storage on RADV. VK_AMD_gpu_shader_int16 on RADV. VK_AMD_gpu_shader_half_float on RADV. VK_NV_compute_shader_derivatives on Intel. VK_KHR_shader_float16_int8 on Intel and RADV (RADV only supports int8). VK_KHR_shader_atomic_int64 on Intel. VK_EXT_descriptor_indexing on Intel. VK_KHR_shader_float16_int8 on Intel and RADV. GL_INTEL_conservative_rasterization on iris. VK_EXT_memory_budget on Intel. New features in mesa 19.0.0: GL_AMD_texture_texture4 on all GL 4.0 drivers. GL_EXT_shader_implicit_conversions on all drivers (ES extension). GL_EXT_texture_compression_bptc on all GL 4.0 drivers (ES extension). GL_EXT_texture_compression_rgtc on all GL 3.0 drivers (ES extension). GL_EXT_render_snorm on gallium drivers (ES extension). GL_EXT_texture_view on drivers supporting texture views (ES extension). GL_OES_texture_view on drivers supporting texture views (ES extension). GL_NV_shader_atomic_float on nvc0 (Fermi/Kepler only). Shader-based software implementations of GL_ARB_gpu_shader_fp64, GL_ARB_gpu_shader_int64, GL_ARB_vertex_attrib_64bit, and GL_ARB_shader_ballot on i965. VK_ANDROID_external_memory_android_hardware_buffer on Intel Fixed and re-exposed VK_EXT_pci_bus_info on Intel and RADV VK_EXT_scalar_block_layout on Intel and RADV VK_KHR_depth_stencil_resolve on Intel VK_KHR_draw_indirect_count on Intel VK_EXT_conditional_rendering on Intel VK_EXT_memory_budget on RADV Also, bug fixes. @ text @d51 1 a51 1 shader, 16, -1) {} d113 2 a114 2 * 0: add(16) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 d117 2 a118 2 * 0: add.sat(16) dst0 src0 src1 * 1: mov(16) dst1 dst0 d150 3 a152 3 * 0: add(16) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 * 2: add(16) dst2 dst0 src0 d187 2 a188 2 * 0: (+f0) add(16) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 d221 2 a222 2 * 0: rndu(16) dst0 src0 * 1: mov.sat(16) dst1 -dst0 d256 2 a257 2 * 0: add(16) dst0 src0 src1 * 1: mov.sat(16) dst1 -dst0 d260 2 a261 2 * 0: add.sat(16) dst0 -src0 -src1 * 1: mov(16) dst1 dst0 d294 2 a295 2 * 0: add(16) dst0 src0 1.0f * 1: mov.sat(16) dst1 -dst0 d298 2 a299 2 * 0: add.sat(16) dst0 -src0 -1.0f * 1: mov(16) dst1 dst0 d332 2 a333 2 * 0: mul(16) dst0 src0 src1 * 1: mov.sat(16) dst1 -dst0 d336 2 a337 2 * 0: mul.sat(16) dst0 src0 -src1 * 1: mov(16) dst1 dst0 d371 2 a372 2 * 0: mad(16) dst0 src0 src1 src2 * 1: mov.sat(16) dst1 -dst0 d375 2 a376 2 * 0: mad.sat(16) dst0 -src0 -src1 src2 * 1: mov(16) dst1 dst0 d417 2 a418 2 * 0: mad(16) dst0 1.0f -2.0f src2 * 1: mov.sat(16) dst1 -dst0 d421 2 a422 2 * 0: mad.sat(16) dst0 -1.0f 2.0f src2 * 1: mov(16) dst1 dst0 d458 3 a460 3 * 0: mul(16) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 * 2: mov.sat(16) dst2 -dst0 d500 3 a502 3 * 0: mul(16) dst0 src0 src1 * 1: mov.sat(16) dst1 -dst0 * 2: mov.sat(16) dst2 -dst0 d541 2 a542 2 * 0: add(16) dst0 src0 src1 * 1: mov.sat(16) dst1 (abs)dst0 d577 3 a579 3 * 0: add.sat(16) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 * 2: mov(16) dst2 dst0 d582 3 a584 3 * 0: add.sat(16) dst0 src0 src1 * 1: mov(16) dst1 dst0 * 2: mov(16) dst2 dst0 d616 3 a618 3 * 0: add(16) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 * 2: mov.sat(16) dst2 dst0 d621 3 a623 3 * 0: add.sat(16) dst0 src0 src1 * 1: mov(16) dst1 dst0 * 2: mov(16) dst2 dst0 d653 1 a653 1 ->size_written = 8 * REG_SIZE; d658 3 a660 3 * 0: add(16) dst0+2 src0 src1 * 1: tex(16) rlen 4 dst0+0 src2 * 2: mov.sat(16) dst1 dst0+2 d699 3 a701 3 * 0: mul(16) dst0 src0 src1 * 1: mov.sat(16) dst1 -dst0 * 2: mov.sat(16) dst2 dst0 a724 68 TEST_F(saturate_propagation_test, smaller_exec_size_consumer) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.ADD(dst0, src0, src1); set_saturate(true, bld.group(8, 0).MOV(dst1, dst0)); /* = Before = * * 0: add(16) dst0 src0 src1 * 1: mov.sat(8) dst1 dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); } TEST_F(saturate_propagation_test, larger_exec_size_consumer) { const fs_builder &bld = v->bld; fs_reg dst0 = v->vgrf(glsl_type::float_type); fs_reg dst1 = v->vgrf(glsl_type::float_type); fs_reg src0 = v->vgrf(glsl_type::float_type); fs_reg src1 = v->vgrf(glsl_type::float_type); bld.group(8, 0).ADD(dst0, src0, src1); set_saturate(true, bld.MOV(dst1, dst0)); /* = Before = * * 0: add(8) dst0 src0 src1 * 1: mov.sat(16) dst1 dst0 * * = After = * (no changes) */ v->calculate_cfg(); bblock_t *block0 = v->cfg->blocks[0]; EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_FALSE(saturate_propagation(v)); EXPECT_EQ(0, block0->start_ip); EXPECT_EQ(1, block0->end_ip); EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); EXPECT_FALSE(instruction(block0, 0)->saturate); EXPECT_EQ(BRW_OPCODE_MOV, instruction(block0, 1)->opcode); EXPECT_TRUE(instruction(block0, 1)->saturate); } @ 1.1.1.3 log @initial import of mesa 21.3.7 main changes since 19.1.7 include: - more support for Vulkan functions - better supported for newer radeonsi (both amdgpu and radeon backends) - various bug fixes in many drivers - many fixes and enhancements for intel drivers - some fixes for nvidia - OpenGL 4.6 for some drivers (intel, radeonsi) - intel Tigerlake and Rocketlake support - Vulkan 1.2 for some drivers - OpenGL 4.5, GLES 3.2, and more on llvmpipe - working Panfrost and Midgard drivers - fix warnings in radeonsi vs newer llvm @ text @a32 1 virtual void TearDown(); d36 2 a37 2 struct intel_device_info *devinfo; void *ctx; a46 1 void *mem_ctx, d49 3 a51 2 : fs_visitor(compiler, NULL, mem_ctx, NULL, &prog_data->base, shader, 16, -1, false) {} d57 3 a59 3 ctx = ralloc_context(NULL); compiler = rzalloc(ctx, struct brw_compiler); devinfo = rzalloc(ctx, struct intel_device_info); d62 1 a62 1 prog_data = ralloc(ctx, struct brw_wm_prog_data); d64 1 a64 1 nir_shader_create(ctx, MESA_SHADER_FRAGMENT, NULL, NULL); d66 1 a66 1 v = new saturate_propagation_fs_visitor(compiler, ctx, prog_data, shader); d68 1 a68 2 devinfo->ver = 6; devinfo->verx10 = devinfo->ver * 10; a70 10 void saturate_propagation_test::TearDown() { delete v; v = NULL; ralloc_free(ctx); ctx = NULL; } d88 1 a88 1 v->cfg->dump(); d95 1 a95 1 v->cfg->dump(); @