fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-06 11:38:05 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

aco: Use v_fmaak/v_fmamk if two operands are the same literal.

Browse source 
Foz-DB Navi21:
Totals from 5744 (4.26% of 134913) affected shaders:
VGPRs: 237128 -> 237056 (-0.03%); split: -0.04%, +0.01%
CodeSize: 16654484 -> 16620668 (-0.20%); split: -0.23%, +0.03%
MaxWaves: 152838 -> 152846 (+0.01%)
Instrs: 3063214 -> 3058572 (-0.15%); split: -0.17%, +0.02%
Latency: 23935195 -> 23934827 (-0.00%); split: -0.03%, +0.03%
InvThroughput: 5478562 -> 5478160 (-0.01%); split: -0.01%, +0.01%
VClause: 60432 -> 60435 (+0.00%); split: -0.02%, +0.03%
SClause: 121032 -> 120896 (-0.11%); split: -0.20%, +0.09%
Copies: 147865 -> 143144 (-3.19%); split: -3.59%, +0.40%
PreSGPRs: 195722 -> 195661 (-0.03%); split: -0.06%, +0.03%
PreVGPRs: 182849 -> 182787 (-0.03%)

Foz-DB Vega10:
Totals from 5290 (3.92% of 135041) affected shaders:
SGPRs: 357952 -> 359616 (+0.46%); split: -0.11%, +0.57%
VGPRs: 204048 -> 203928 (-0.06%); split: -0.08%, +0.02%
CodeSize: 14043176 -> 14003100 (-0.29%); split: -0.29%, +0.00%
MaxWaves: 39401 -> 39398 (-0.01%); split: +0.01%, -0.02%
Instrs: 2636739 -> 2631246 (-0.21%); split: -0.21%, +0.00%
Latency: 25264088 -> 25256482 (-0.03%); split: -0.05%, +0.02%
InvThroughput: 12039643 -> 12039346 (-0.00%); split: -0.00%, +0.00%
VClause: 55603 -> 55584 (-0.03%); split: -0.04%, +0.00%
SClause: 101577 -> 101342 (-0.23%); split: -0.30%, +0.07%
Copies: 213344 -> 207929 (-2.54%); split: -2.58%, +0.05%
Branches: 34053 -> 34054 (+0.00%)
PreSGPRs: 172405 -> 172260 (-0.08%)

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18645>
This commit is contained in:
Georg Lehmann 2022-09-17 20:52:24 +02:00 • committed by Marge Bot
parent 075cd3ca94
commit 94b5225c9b
1 changed files with 55 additions and 57 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

112
src/amd/compiler/aco_optimizer.cpp
View file

@ -75,11 +75,10 @@ perfwarn(Program* program, bool cond, const char* msg, Instruction* instr)
struct mad_info {
aco_ptr<Instruction> add_instr;
uint32_t mul_temp_id;
uint16_t literal_idx;
bool check_literal;
uint16_t literal_mask;
mad_info(aco_ptr<Instruction> instr, uint32_t id)
: add_instr(std::move(instr)), mul_temp_id(id), literal_idx(0), check_literal(false)
: add_instr(std::move(instr)), mul_temp_id(id), literal_mask(0)
{}
};
@ -4398,49 +4397,42 @@ select_instruction(opt_ctx& ctx, aco_ptr<Instruction>& instr)
if (instr->opcode == aco_opcode::v_fma_legacy_f16)
return;
uint32_t literal_idx = 0;
uint32_t literal_mask = 0;
uint32_t sgpr_mask = 0;
uint32_t vgpr_mask = 0;
uint32_t literal_uses = UINT32_MAX;
uint32_t literal_value = 0;
/* Try using v_madak/v_fmaak */
if (instr->operands[2].isTemp() &&
ctx.info[instr->operands[2].tempId()].is_literal(get_operand_size(instr, 2))) {
bool has_sgpr = false;
bool has_vgpr = false;
for (unsigned i = 0; i < 2; i++) {
if (!instr->operands[i].isTemp())
/* Iterate in reverse to prefer v_madak/v_fmaak. */
for (int i = 2; i >= 0; i--) {
Operand& op = instr->operands[i];
if (!op.isTemp())
continue;
if (ctx.info[op.tempId()].is_literal(get_operand_size(instr, i))) {
uint32_t new_literal = ctx.info[op.tempId()].val;
if (!literal_mask || literal_value == new_literal) {
literal_value = new_literal;
literal_uses = MIN2(literal_uses, ctx.uses[op.tempId()]);
literal_mask |= 1 << i;
continue;
has_sgpr |= instr->operands[i].getTemp().type() == RegType::sgpr;
has_vgpr |= instr->operands[i].getTemp().type() == RegType::vgpr;
}
/* Encoding limitations requires a VGPR operand. The constant bus limitations before
* GFX10 disallows SGPRs.
*/
if ((!has_sgpr || ctx.program->gfx_level >= GFX10) && has_vgpr) {
literal_idx = 2;
literal_uses = ctx.uses[instr->operands[2].tempId()];
}
}
/* Try using v_madmk/v_fmamk */
/* Encoding limitations requires a VGPR operand. */
if (instr->operands[2].isTemp() && instr->operands[2].getTemp().type() == RegType::vgpr) {
for (unsigned i = 0; i < 2; i++) {
if (!instr->operands[i].isTemp())
continue;
/* The constant bus limitations before GFX10 disallows SGPRs. */
if (ctx.program->gfx_level < GFX10 && instr->operands[!i].isTemp() &&
instr->operands[!i].getTemp().type() == RegType::sgpr)
continue;
if (ctx.info[instr->operands[i].tempId()].is_literal(get_operand_size(instr, i)) &&
ctx.uses[instr->operands[i].tempId()] < literal_uses) {
literal_idx = i;
literal_uses = ctx.uses[instr->operands[i].tempId()];
}
}
sgpr_mask |= op.isOfType(RegType::sgpr) << i;
vgpr_mask |= op.isOfType(RegType::vgpr) << i;
}
/* The constant bus limitations before GFX10 disallows SGPRs. */
if (sgpr_mask && ctx.program->gfx_level < GFX10)
literal_mask = 0;
/* Encoding needs a vgpr. */
if (!vgpr_mask)
literal_mask = 0;
/* v_madmk/v_fmamk needs a vgpr in the third source. */
if (!(literal_mask & 0b100) && !(vgpr_mask & 0b100))
literal_mask = 0;
/* Limit the number of literals to apply to not increase the code
* size too much, but always apply literals for v_mad->v_madak
* because both instructions are 64-bit and this doesn't increase
@ -4448,10 +4440,10 @@ select_instruction(opt_ctx& ctx, aco_ptr<Instruction>& instr)
* TODO: try to apply the literals earlier to lower the number of
* uses below threshold
*/
if (literal_uses < threshold || literal_idx == 2) {
ctx.uses[instr->operands[literal_idx].tempId()]--;
mad_info->check_literal = true;
mad_info->literal_idx = literal_idx;
if (literal_mask && (literal_uses < threshold || (literal_mask & 0b100))) {
u_foreach_bit (i, literal_mask)
ctx.uses[instr->operands[i].tempId()]--;
mad_info->literal_mask = literal_mask;
return;
}
}
@ -4733,33 +4725,39 @@ apply_literals(opt_ctx& ctx, aco_ptr<Instruction>& instr)
/* apply literals on MAD */
if (!instr->definitions.empty() && ctx.info[instr->definitions[0].tempId()].is_mad()) {
mad_info* info = &ctx.mad_infos[ctx.info[instr->definitions[0].tempId()].instr->pass_flags];
if (info->check_literal &&
(ctx.uses[instr->operands[info->literal_idx].tempId()] == 0 || info->literal_idx == 2)) {
const bool madak = (info->literal_mask & 0b100);
bool has_dead_literal = false;
u_foreach_bit (i, info->literal_mask)
has_dead_literal |= ctx.uses[instr->operands[i].tempId()] == 0;
if (has_dead_literal || madak) {
aco_ptr<Instruction> new_mad;
aco_opcode new_op =
info->literal_idx == 2 ? aco_opcode::v_madak_f32 : aco_opcode::v_madmk_f32;
aco_opcode new_op = madak ? aco_opcode::v_madak_f32 : aco_opcode::v_madmk_f32;
if (instr->opcode == aco_opcode::v_fma_f32)
new_op = info->literal_idx == 2 ? aco_opcode::v_fmaak_f32 : aco_opcode::v_fmamk_f32;
new_op = madak ? aco_opcode::v_fmaak_f32 : aco_opcode::v_fmamk_f32;
else if (instr->opcode == aco_opcode::v_mad_f16 ||
instr->opcode == aco_opcode::v_mad_legacy_f16)
new_op = info->literal_idx == 2 ? aco_opcode::v_madak_f16 : aco_opcode::v_madmk_f16;
new_op = madak ? aco_opcode::v_madak_f16 : aco_opcode::v_madmk_f16;
else if (instr->opcode == aco_opcode::v_fma_f16)
new_op = info->literal_idx == 2 ? aco_opcode::v_fmaak_f16 : aco_opcode::v_fmamk_f16;
new_op = madak ? aco_opcode::v_fmaak_f16 : aco_opcode::v_fmamk_f16;
uint32_t literal = ctx.info[instr->operands[ffs(info->literal_mask) - 1].tempId()].val;
new_mad.reset(create_instruction<VOP2_instruction>(new_op, Format::VOP2, 3, 1));
if (info->literal_idx == 2) { /* add literal -> madak */
new_mad->operands[0] = instr->operands[0];
new_mad->operands[1] = instr->operands[1];
for (unsigned i = 0; i < 3; i++) {
if (info->literal_mask & (1 << i))
new_mad->operands[i] = Operand::literal32(literal);
else
new_mad->operands[i] = instr->operands[i];
}
if (madak) { /* add literal -> madak */
if (!new_mad->operands[1].isTemp() ||
new_mad->operands[1].getTemp().type() == RegType::sgpr)
std::swap(new_mad->operands[0], new_mad->operands[1]);
} else { /* mul literal -> madmk */
new_mad->operands[0] = instr->operands[1 - info->literal_idx];
new_mad->operands[1] = instr->operands[2];
if (!(info->literal_mask & 0b10))
std::swap(new_mad->operands[0], new_mad->operands[1]);
std::swap(new_mad->operands[1], new_mad->operands[2]);
}
new_mad->operands[2] =
Operand::c32(ctx.info[instr->operands[info->literal_idx].tempId()].val);
new_mad->definitions[0] = instr->definitions[0];
ctx.instructions.emplace_back(std::move(new_mad));
return;