Matt Turner <[email protected]> writes: > On Fri, Mar 20, 2015 at 7:50 AM, Francisco Jerez <[email protected]> > wrote: >> --- >> src/mesa/drivers/dri/i965/brw_vec4.cpp | 15 ++++++--------- >> 1 file changed, 6 insertions(+), 9 deletions(-) >> >> diff --git a/src/mesa/drivers/dri/i965/brw_vec4.cpp >> b/src/mesa/drivers/dri/i965/brw_vec4.cpp >> index ed9944d..ce02fe3 100644 >> --- a/src/mesa/drivers/dri/i965/brw_vec4.cpp >> +++ b/src/mesa/drivers/dri/i965/brw_vec4.cpp >> @@ -1203,15 +1203,12 @@ vec4_visitor::split_virtual_grfs() >> * to split. >> */ >> foreach_block_and_inst(block, vec4_instruction, inst, cfg) { >> - /* If there's a SEND message loading from a GRF on gen7+, it needs to >> be >> - * contiguous. >> - */ >> - if (inst->is_send_from_grf()) { >> - for (int i = 0; i < 3; i++) { >> - if (inst->src[i].file == GRF) { >> - split_grf[inst->src[i].reg] = false; >> - } >> - } > > Wow, am I reading this right that we weren't setting split_grf[...] = > false for any destinations? > Yeah, you were reading it right...
>> + if (inst->dst.file == GRF && inst->regs_written > 1)
>> + split_grf[inst->dst.reg] = false;
>> +
>> + for (int i = 0; i < 3; i++) {
>> + if (inst->src[i].file == GRF && inst->regs_read(i) > 1)
>> + split_grf[inst->src[i].reg] = false;
>> }
>> }
>>
>> --
>> 2.1.3
signature.asc
Description: PGP signature
_______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
