fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-27 17:00:09 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

i965: Improve compute-to-mrf.

Browse source 
We were skipping it if the instruction producing the value we were
going to compute-to-mrf used its result reg as a source reg.  This
meant that the typical "write interpolated color to fragment color" or
"texture from interpolated texcoord" shader didn't compute-to-MRF.
Just don't check for the interference cases until after we've checked
if this is the instruction we wanted to compute-to-MRF.

Improves nexuiz high-settings performance on my laptop 0.48% +- 0.08%
(n=3).
This commit is contained in:
Eric Anholt 2010-11-18 15:03:50 +08:00
parent ac89a90401
commit 47b1aac1cf
1 changed files with 50 additions and 54 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

104
src/mesa/drivers/dri/i965/brw_fs.cpp
View file

@ -2963,11 +2963,60 @@ fs_visitor::compute_to_mrf()
/* Found a move of a GRF to a MRF. Let's see if we can go
* rewrite the thing that made this GRF to write into the MRF.
*/
bool found = false;
fs_inst *scan_inst;
for (scan_inst = (fs_inst *)inst->prev;
scan_inst->prev != NULL;
scan_inst = (fs_inst *)scan_inst->prev) {
if (scan_inst->dst.file == GRF &&
scan_inst->dst.reg == inst->src[0].reg) {
/* Found the last thing to write our reg we want to turn
* into a compute-to-MRF.
*/
if (scan_inst->opcode == FS_OPCODE_TEX) {
/* texturing writes several continuous regs, so we can't
* compute-to-mrf that.
*/
break;
}
/* If it's predicated, it (probably) didn't populate all
* the channels.
*/
if (scan_inst->predicated)
break;
/* SEND instructions can't have MRF as a destination. */
if (scan_inst->mlen)
break;
if (intel->gen >= 6) {
/* gen6 math instructions must have the destination be
* GRF, so no compute-to-MRF for them.
*/
if (scan_inst->opcode == FS_OPCODE_RCP ||
scan_inst->opcode == FS_OPCODE_RSQ ||
scan_inst->opcode == FS_OPCODE_SQRT ||
scan_inst->opcode == FS_OPCODE_EXP2 ||
scan_inst->opcode == FS_OPCODE_LOG2 ||
scan_inst->opcode == FS_OPCODE_SIN ||
scan_inst->opcode == FS_OPCODE_COS ||
scan_inst->opcode == FS_OPCODE_POW) {
break;
}
}
if (scan_inst->dst.reg_offset == inst->src[0].reg_offset) {
/* Found the creator of our MRF's source value. */
scan_inst->dst.file = MRF;
scan_inst->dst.hw_reg = inst->dst.hw_reg;
scan_inst->saturate |= inst->saturate;
inst->remove();
progress = true;
}
break;
}
/* We don't handle flow control here. Most computation of
* values that end up in MRFs are shortly before the MRF
* write anyway.
@ -3011,59 +3060,6 @@ fs_visitor::compute_to_mrf()
break;
}
}
if (scan_inst->dst.file == GRF &&
scan_inst->dst.reg == inst->src[0].reg) {
/* Found the last thing to write our reg we want to turn
* into a compute-to-MRF.
*/
if (scan_inst->opcode == FS_OPCODE_TEX) {
/* texturing writes several continuous regs, so we can't
* compute-to-mrf that.
*/
break;
}
/* If it's predicated, it (probably) didn't populate all
* the channels.
*/
if (scan_inst->predicated)
break;
/* SEND instructions can't have MRF as a destination. */
if (scan_inst->mlen)
break;
if (intel->gen >= 6) {
/* gen6 math instructions must have the destination be
* GRF, so no compute-to-MRF for them.
*/
if (scan_inst->opcode == FS_OPCODE_RCP ||
scan_inst->opcode == FS_OPCODE_RSQ ||
scan_inst->opcode == FS_OPCODE_SQRT ||
scan_inst->opcode == FS_OPCODE_EXP2 ||
scan_inst->opcode == FS_OPCODE_LOG2 ||
scan_inst->opcode == FS_OPCODE_SIN ||
scan_inst->opcode == FS_OPCODE_COS ||
scan_inst->opcode == FS_OPCODE_POW) {
break;
}
}
if (scan_inst->dst.reg_offset == inst->src[0].reg_offset) {
/* Found the creator of our MRF's source value. */
found = true;
break;
}
}
}
if (found) {
scan_inst->dst.file = MRF;
scan_inst->dst.hw_reg = inst->dst.hw_reg;
scan_inst->saturate |= inst->saturate;
inst->remove();
progress = true;
}
}