diff options
author | Samuel Iglesias Gonsálvez <[email protected]> | 2016-08-29 10:10:30 +0200 |
---|---|---|
committer | Francisco Jerez <[email protected]> | 2017-04-14 14:56:08 -0700 |
commit | a21dc2b500cff6e0aaf31867c5b42651306ddaf1 (patch) | |
tree | 38ba3688df51336582c62cf7ad9470de7ec7eaf0 /src/intel/compiler/brw_ir_vec4.h | |
parent | a5399e8b1cc3e2e12b8aa067e8380d1b088c35ca (diff) |
i965/vec4: split DF instructions and later double its execsize in IVB/BYT
We need to split DF instructions in two on IVB/BYT as it needs an
execsize 8 to process 4 DF values (one GRF in total).
v2:
- Rename helper and make it static inline function (Matt).
- Fix indention and add braces (Matt).
v3:
- Don't edit IR instruction when doubling exec_size (Curro)
- Add comment into the code (Curro).
- Manage ARF registers like the others (Curro)
v4:
- Add get_exec_type() function and use it to calculate the execution
size.
Signed-off-by: Samuel Iglesias Gonsálvez <[email protected]>
[ Francisco Jerez: Fix bogus 'type != BAD_FILE' check. Take
destination type as execution type where there is no valid source.
Assert-fail if the deduced execution type is byte. Clarify comment
in get_lowered_simd_width(). Move SIMD width workaround outside of
'if (...inst->size_written > REG_SIZE)' conditional block, since the
problem should be independent of whether the amount of data written
by the instruction is greater or lower than a GRF. Drop redundant
is_ivb_df definition. Drop bogus inst->exec_size < 8 check.
Simplify channel group assertion. ]
Reviewed-by: Francisco Jerez <[email protected]>
Diffstat (limited to 'src/intel/compiler/brw_ir_vec4.h')
-rw-r--r-- | src/intel/compiler/brw_ir_vec4.h | 33 |
1 files changed, 33 insertions, 0 deletions
diff --git a/src/intel/compiler/brw_ir_vec4.h b/src/intel/compiler/brw_ir_vec4.h index bd026eb2aeb..56548c38830 100644 --- a/src/intel/compiler/brw_ir_vec4.h +++ b/src/intel/compiler/brw_ir_vec4.h @@ -404,6 +404,39 @@ regs_read(const vec4_instruction *inst, unsigned i) reg_size); } +static inline enum brw_reg_type +get_exec_type(const vec4_instruction *inst) +{ + enum brw_reg_type exec_type = BRW_REGISTER_TYPE_B; + + for (int i = 0; i < 3; i++) { + if (inst->src[i].file != BAD_FILE) { + const brw_reg_type t = get_exec_type(brw_reg_type(inst->src[i].type)); + if (type_sz(t) > type_sz(exec_type)) + exec_type = t; + else if (type_sz(t) == type_sz(exec_type) && + brw_reg_type_is_floating_point(t)) + exec_type = t; + } + } + + if (exec_type == BRW_REGISTER_TYPE_B) + exec_type = inst->dst.type; + + /* TODO: We need to handle half-float conversions. */ + assert(exec_type != BRW_REGISTER_TYPE_HF || + inst->dst.type == BRW_REGISTER_TYPE_HF); + assert(exec_type != BRW_REGISTER_TYPE_B); + + return exec_type; +} + +static inline unsigned +get_exec_type_size(const vec4_instruction *inst) +{ + return type_sz(get_exec_type(inst)); +} + } /* namespace brw */ #endif |