diff options
| author | Ilia Mirkin <imirkin@alum.mit.edu> | 2015-03-25 18:00:00 -0400 |
|---|---|---|
| committer | Emil Velikov <emil.l.velikov@gmail.com> | 2015-04-08 18:45:32 +0100 |
| commit | c8d962c205bc5d6f81f3d083d90ecaf66f78b1e8 (patch) | |
| tree | fec54c69a3af6d8f3e3655fff8f2b1c1fc90a47b | |
| parent | 407365e375ca7ba78d95c2e3f189a837201bcc22 (diff) | |
| download | mesa-c8d962c205bc5d6f81f3d083d90ecaf66f78b1e8.tar.gz | |
nv50/ir: take postFactor into account when doing peephole optimizations
Multiply operations can have a post-factor on them, which other ops
don't support. Only perform the peephole optimizations when there is no
post-factor involved.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89758
Cc: "10.4 10.5" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
(cherry picked from commit 49b86007aa2bb599ada6cdbed7ff56246917f12e)
| -rw-r--r-- | src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp | 12 |
1 files changed, 8 insertions, 4 deletions
diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp index 21d20caffdf..9e2a933daa4 100644 --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp @@ -422,7 +422,9 @@ ConstantFolding::expr(Instruction *i, b->data.f32 = 0.0f; } switch (i->dType) { - case TYPE_F32: res.data.f32 = a->data.f32 * b->data.f32; break; + case TYPE_F32: + res.data.f32 = a->data.f32 * b->data.f32 * exp2f(i->postFactor); + break; case TYPE_F64: res.data.f64 = a->data.f64 * b->data.f64; break; case TYPE_S32: if (i->subOp == NV50_IR_SUBOP_MUL_HIGH) { @@ -550,6 +552,7 @@ ConstantFolding::expr(Instruction *i, i->src(0).mod = Modifier(0); i->src(1).mod = Modifier(0); + i->postFactor = 0; i->setSrc(0, new_ImmediateValue(i->bb->getProgram(), res.data.u32)); i->setSrc(1, NULL); @@ -653,7 +656,7 @@ ConstantFolding::tryCollapseChainedMULs(Instruction *mul2, Instruction *insn; Instruction *mul1 = NULL; // mul1 before mul2 int e = 0; - float f = imm2.reg.data.f32; + float f = imm2.reg.data.f32 * exp2f(mul2->postFactor); ImmediateValue imm1; assert(mul2->op == OP_MUL && mul2->dType == TYPE_F32); @@ -753,9 +756,10 @@ ConstantFolding::opnd(Instruction *i, ImmediateValue &imm0, int s) i->op = OP_MOV; i->setSrc(0, new_ImmediateValue(prog, 0u)); i->src(0).mod = Modifier(0); + i->postFactor = 0; i->setSrc(1, NULL); } else - if (imm0.isInteger(1) || imm0.isInteger(-1)) { + if (!i->postFactor && (imm0.isInteger(1) || imm0.isInteger(-1))) { if (imm0.isNegative()) i->src(t).mod = i->src(t).mod ^ Modifier(NV50_IR_MOD_NEG); i->op = i->src(t).mod.getOp(); @@ -768,7 +772,7 @@ ConstantFolding::opnd(Instruction *i, ImmediateValue &imm0, int s) i->src(0).mod = 0; i->setSrc(1, NULL); } else - if (imm0.isInteger(2) || imm0.isInteger(-2)) { + if (!i->postFactor && (imm0.isInteger(2) || imm0.isInteger(-2))) { if (imm0.isNegative()) i->src(t).mod = i->src(t).mod ^ Modifier(NV50_IR_MOD_NEG); i->op = OP_ADD; |