diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/intel/compiler/brw_vec4_cmod_propagation.cpp | 20 | ||||
| -rw-r--r-- | src/intel/compiler/test_vec4_cmod_propagation.cpp | 72 |
2 files changed, 87 insertions, 5 deletions
diff --git a/src/intel/compiler/brw_vec4_cmod_propagation.cpp b/src/intel/compiler/brw_vec4_cmod_propagation.cpp index 5205da4983c..a1d46dc8dca 100644 --- a/src/intel/compiler/brw_vec4_cmod_propagation.cpp +++ b/src/intel/compiler/brw_vec4_cmod_propagation.cpp @@ -36,6 +36,17 @@ namespace brw { static bool +writemasks_incompatible(const vec4_instruction *earlier, + const vec4_instruction *later) +{ + return (earlier->dst.writemask != WRITEMASK_X && + earlier->dst.writemask != WRITEMASK_XYZW) || + (earlier->dst.writemask == WRITEMASK_XYZW && + later->src[0].swizzle != BRW_SWIZZLE_XYZW) || + (later->dst.writemask & ~earlier->dst.writemask) != 0; +} + +static bool opt_cmod_propagation_local(bblock_t *block) { bool progress = false; @@ -82,6 +93,9 @@ opt_cmod_propagation_local(bblock_t *block) if (scan_inst->opcode != BRW_OPCODE_ADD) goto not_match; + if (writemasks_incompatible(scan_inst, inst)) + goto not_match; + /* A CMP is basically a subtraction. The result of the * subtraction must be the same as the result of the addition. * This means that one of the operands must be negated. So (a + @@ -132,11 +146,7 @@ opt_cmod_propagation_local(bblock_t *block) scan_inst->dst, scan_inst->size_written)) { if ((scan_inst->predicate && scan_inst->opcode != BRW_OPCODE_SEL) || scan_inst->dst.offset != inst->src[0].offset || - (scan_inst->dst.writemask != WRITEMASK_X && - scan_inst->dst.writemask != WRITEMASK_XYZW) || - (scan_inst->dst.writemask == WRITEMASK_XYZW && - inst->src[0].swizzle != BRW_SWIZZLE_XYZW) || - (inst->dst.writemask & ~scan_inst->dst.writemask) != 0 || + writemasks_incompatible(scan_inst, inst) || scan_inst->exec_size != inst->exec_size || scan_inst->group != inst->group) { break; diff --git a/src/intel/compiler/test_vec4_cmod_propagation.cpp b/src/intel/compiler/test_vec4_cmod_propagation.cpp index 7d9792b4a55..8430924de63 100644 --- a/src/intel/compiler/test_vec4_cmod_propagation.cpp +++ b/src/intel/compiler/test_vec4_cmod_propagation.cpp @@ -821,3 +821,75 @@ TEST_F(cmod_propagation_test, mul_cmp_different_channels_vec4) EXPECT_EQ(BRW_OPCODE_CMP, instruction(block0, 1)->opcode); EXPECT_EQ(BRW_CONDITIONAL_NZ, instruction(block0, 1)->conditional_mod); } + +TEST_F(cmod_propagation_test, add_cmp_same_dst_writemask) +{ + const vec4_builder bld = vec4_builder(v).at_end(); + dst_reg dest = dst_reg(v, glsl_type::vec4_type); + src_reg src0 = src_reg(v, glsl_type::vec4_type); + src_reg src1 = src_reg(v, glsl_type::vec4_type); + dst_reg dest_null = bld.null_reg_f(); + + bld.ADD(dest, src0, src1); + vec4_instruction *inst = bld.CMP(dest_null, src0, src1, BRW_CONDITIONAL_GE); + inst->src[1].negate = true; + + /* = Before = + * + * 0: add dest.xyzw src0 src1 + * 1: cmp.ge.f0 null.xyzw src0 -src1 + * + * = After = + * 0: add.ge.f0 dest.xyzw src0 src1 + */ + + v->calculate_cfg(); + bblock_t *block0 = v->cfg->blocks[0]; + + EXPECT_EQ(0, block0->start_ip); + EXPECT_EQ(1, block0->end_ip); + + EXPECT_TRUE(cmod_propagation(v)); + + ASSERT_EQ(0, block0->start_ip); + ASSERT_EQ(0, block0->end_ip); + EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); + EXPECT_EQ(BRW_CONDITIONAL_GE, instruction(block0, 0)->conditional_mod); +} + +TEST_F(cmod_propagation_test, add_cmp_different_dst_writemask) +{ + const vec4_builder bld = vec4_builder(v).at_end(); + dst_reg dest = dst_reg(v, glsl_type::float_type); + src_reg src0 = src_reg(v, glsl_type::vec4_type); + src_reg src1 = src_reg(v, glsl_type::vec4_type); + dst_reg dest_null = bld.null_reg_f(); + + bld.ADD(dest, src0, src1); + vec4_instruction *inst = bld.CMP(dest_null, src0, src1, BRW_CONDITIONAL_GE); + inst->src[1].negate = true; + + /* = Before = + * + * 0: add dest.x src0 src1 + * 1: cmp.ge.f0 null.xyzw src0 -src1 + * + * = 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(cmod_propagation(v)); + + ASSERT_EQ(0, block0->start_ip); + ASSERT_EQ(1, block0->end_ip); + EXPECT_EQ(BRW_OPCODE_ADD, instruction(block0, 0)->opcode); + EXPECT_EQ(BRW_CONDITIONAL_NONE, instruction(block0, 0)->conditional_mod); + EXPECT_EQ(BRW_OPCODE_CMP, instruction(block0, 1)->opcode); + EXPECT_EQ(BRW_CONDITIONAL_GE, instruction(block0, 1)->conditional_mod); +} |