diff options
author | Francisco Jerez <[email protected]> | 2016-05-19 00:10:03 -0700 |
---|---|---|
committer | Francisco Jerez <[email protected]> | 2016-05-27 23:29:04 -0700 |
commit | 81bc6de8c0f7faafd0f3b0aee944a14ba3ef0b64 (patch) | |
tree | 15be42f1cedc34eb105d7d8fb7d0838c5c9a005c /src/mesa/drivers | |
parent | 41562eb8f33558f02ff8f53b3094a0e6d54e4c49 (diff) |
i965/ir: Make BROADCAST emit an unmasked single-channel move.
Alternatively we could have extended the current semantics to 32-wide
mode by changing brw_broadcast() to emit multiple indexed MOV
instructions in the generator copying the selected value to all
destination registers, but it seemed rather silly to waste EU cycles
unnecessarily copying the exact same value 32 times in the GRF.
The vstride change in the Align16 path is required to avoid assertions
in validate_reg() since the change causes the execution size of the
MOV and SEL instructions to be equal to the source region width.
Reviewed-by: Jason Ekstrand <[email protected]>
Diffstat (limited to 'src/mesa/drivers')
-rw-r--r-- | src/mesa/drivers/dri/i965/brw_defines.h | 6 | ||||
-rw-r--r-- | src/mesa/drivers/dri/i965/brw_eu_emit.c | 12 | ||||
-rw-r--r-- | src/mesa/drivers/dri/i965/brw_fs_generator.cpp | 1 | ||||
-rw-r--r-- | src/mesa/drivers/dri/i965/brw_vec4_generator.cpp | 1 |
4 files changed, 17 insertions, 3 deletions
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h index a5f3b52fa35..51eb856d7b1 100644 --- a/src/mesa/drivers/dri/i965/brw_defines.h +++ b/src/mesa/drivers/dri/i965/brw_defines.h @@ -1082,6 +1082,12 @@ enum opcode { /** * Pick the channel from its first source register given by the index * specified as second source. Useful for variable indexing of surfaces. + * + * Note that because the result of this instruction is by definition + * uniform and it can always be splatted to multiple channels using a + * scalar regioning mode, only the first channel of the destination region + * is guaranteed to be updated, which implies that BROADCAST instructions + * should usually be marked force_writemask_all. */ SHADER_OPCODE_BROADCAST, diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c index 2a3b0b03d80..1c2ccb4e163 100644 --- a/src/mesa/drivers/dri/i965/brw_eu_emit.c +++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c @@ -3426,6 +3426,10 @@ brw_broadcast(struct brw_codegen *p, const bool align1 = brw_inst_access_mode(devinfo, p->current) == BRW_ALIGN_1; brw_inst *inst; + brw_push_insn_state(p); + brw_set_default_mask_control(p, BRW_MASK_DISABLE); + brw_set_default_exec_size(p, align1 ? BRW_EXECUTE_1 : BRW_EXECUTE_4); + assert(src.file == BRW_GENERAL_REGISTER_FILE && src.address_mode == BRW_ADDRESS_DIRECT); @@ -3476,19 +3480,21 @@ brw_broadcast(struct brw_codegen *p, */ inst = brw_MOV(p, brw_null_reg(), - stride(brw_swizzle(idx, BRW_SWIZZLE_XXXX), 0, 4, 1)); + stride(brw_swizzle(idx, BRW_SWIZZLE_XXXX), 4, 4, 1)); brw_inst_set_pred_control(devinfo, inst, BRW_PREDICATE_NONE); brw_inst_set_cond_modifier(devinfo, inst, BRW_CONDITIONAL_NZ); brw_inst_set_flag_reg_nr(devinfo, inst, 1); /* and use predicated SEL to pick the right channel. */ inst = brw_SEL(p, dst, - stride(suboffset(src, 4), 0, 4, 1), - stride(src, 0, 4, 1)); + stride(suboffset(src, 4), 4, 4, 1), + stride(src, 4, 4, 1)); brw_inst_set_pred_control(devinfo, inst, BRW_PREDICATE_NORMAL); brw_inst_set_flag_reg_nr(devinfo, inst, 1); } } + + brw_pop_insn_state(p); } /** diff --git a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp index 1d7fc6c9b40..3b0717e17d4 100644 --- a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp +++ b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp @@ -2023,6 +2023,7 @@ fs_generator::generate_code(const cfg_t *cfg, int dispatch_width) break; case SHADER_OPCODE_BROADCAST: + assert(inst->force_writemask_all); brw_broadcast(p, dst, src[0], src[1]); break; diff --git a/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp b/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp index baf442212cc..bb0254ee51e 100644 --- a/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp +++ b/src/mesa/drivers/dri/i965/brw_vec4_generator.cpp @@ -1886,6 +1886,7 @@ generate_code(struct brw_codegen *p, break; case SHADER_OPCODE_BROADCAST: + assert(inst->force_writemask_all); brw_broadcast(p, dst, src[0], src[1]); break; |