diff options
author | Kenneth Graunke <[email protected]> | 2014-06-07 02:21:47 -0700 |
---|---|---|
committer | Kenneth Graunke <[email protected]> | 2014-06-10 16:38:27 -0700 |
commit | 2bcd24c9f0ea70363a4604508981845273f6670a (patch) | |
tree | c52bf73e934694525264c7c559904418ff755605 /src/mesa | |
parent | 35e48bd6188a56097016eb1c05e16d1ba35345c4 (diff) |
i965/fs: Combine generate_math[12]_gen6 methods.
These used to call different math emitters (brw_math vs. brw_math2).
Now that they both call gen6_math, they're virtually identical.
When unrolling SIMD16 to multiple SIMD8 operations, we should take care
not to apply sechalf to brw_null_reg for src1. Otherwise, we'd end up
with BRW_ARF_NULL + 1 as the register number, and I'm not sure if that's
valid.
Signed-off-by: Kenneth Graunke <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Jordan Justen <[email protected]>
Diffstat (limited to 'src/mesa')
-rw-r--r-- | src/mesa/drivers/dri/i965/brw_fs.h | 11 | ||||
-rw-r--r-- | src/mesa/drivers/dri/i965/brw_fs_generator.cpp | 35 |
2 files changed, 13 insertions, 33 deletions
diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h index 9c4e116e3f6..f9e0daff175 100644 --- a/src/mesa/drivers/dri/i965/brw_fs.h +++ b/src/mesa/drivers/dri/i965/brw_fs.h @@ -627,13 +627,10 @@ private: void generate_linterp(fs_inst *inst, struct brw_reg dst, struct brw_reg *src); void generate_tex(fs_inst *inst, struct brw_reg dst, struct brw_reg src); - void generate_math1_gen6(fs_inst *inst, - struct brw_reg dst, - struct brw_reg src); - void generate_math2_gen6(fs_inst *inst, - struct brw_reg dst, - struct brw_reg src0, - struct brw_reg src1); + void generate_math_gen6(fs_inst *inst, + struct brw_reg dst, + struct brw_reg src0, + struct brw_reg src1); void generate_math_gen4(fs_inst *inst, struct brw_reg dst, struct brw_reg src); diff --git a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp index 9a592747458..3787d246801 100644 --- a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp +++ b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp @@ -293,38 +293,21 @@ fs_generator::generate_linterp(fs_inst *inst, } void -fs_generator::generate_math1_gen6(fs_inst *inst, - struct brw_reg dst, - struct brw_reg src0) -{ - int op = brw_math_function(inst->opcode); - - assert(inst->mlen == 0); - - brw_set_default_compression_control(p, BRW_COMPRESSION_NONE); - gen6_math(p, dst, op, src0, brw_null_reg()); - - if (dispatch_width == 16) { - brw_set_default_compression_control(p, BRW_COMPRESSION_2NDHALF); - gen6_math(p, sechalf(dst), op, sechalf(src0), brw_null_reg()); - brw_set_default_compression_control(p, BRW_COMPRESSION_COMPRESSED); - } -} - -void -fs_generator::generate_math2_gen6(fs_inst *inst, - struct brw_reg dst, - struct brw_reg src0, - struct brw_reg src1) +fs_generator::generate_math_gen6(fs_inst *inst, + struct brw_reg dst, + struct brw_reg src0, + struct brw_reg src1) { int op = brw_math_function(inst->opcode); + bool binop = src1.file == BRW_GENERAL_REGISTER_FILE; brw_set_default_compression_control(p, BRW_COMPRESSION_NONE); gen6_math(p, dst, op, src0, src1); if (dispatch_width == 16) { brw_set_default_compression_control(p, BRW_COMPRESSION_2NDHALF); - gen6_math(p, sechalf(dst), op, sechalf(src0), sechalf(src1)); + gen6_math(p, sechalf(dst), op, sechalf(src0), + binop ? sechalf(src1) : brw_null_reg()); brw_set_default_compression_control(p, BRW_COMPRESSION_COMPRESSED); } } @@ -1609,7 +1592,7 @@ fs_generator::generate_code(exec_list *instructions) gen6_math(p, dst, brw_math_function(inst->opcode), src[0], brw_null_reg()); } else if (brw->gen == 6) { - generate_math1_gen6(inst, dst, src[0]); + generate_math_gen6(inst, dst, src[0], brw_null_reg()); } else if (brw->gen == 5 || brw->is_g4x) { generate_math_g45(inst, dst, src[0]); } else { @@ -1623,7 +1606,7 @@ fs_generator::generate_code(exec_list *instructions) if (brw->gen >= 7) { gen6_math(p, dst, brw_math_function(inst->opcode), src[0], src[1]); } else if (brw->gen == 6) { - generate_math2_gen6(inst, dst, src[0], src[1]); + generate_math_gen6(inst, dst, src[0], src[1]); } else { generate_math_gen4(inst, dst, src[0]); } |