[PATCH] nv50/ir: change the way float face is returned

[PATCH] nv50/ir: change the way float face is returned

[Ilia Mirkin]

The old way made it impossible for the optimizer to reason about what
was going on. The new way is the same number of instructions (the neg
gets folded into the cvt) but enables the optimizer to be cleverer if
comparing to a constant (most common case). [The optimizer is presently
not sufficiently clever to work this out, but it could relatively easily
be made to be. The old way would have required significant complexity to
work out.]

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
---

This is all a bit annoying, since gl_FrontFacing is a bool in GL, but
SEMANTIC_FACE is supposed to return a float -1/1. As-is the shader input load
returns an int 0/-1, hence these shenanigans.

 src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp | 5 +++--
 src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp | 5 +++--
 2 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp
index 0d7612e..1ad0860 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp
@@ -1094,8 +1094,9 @@ NV50LoweringPreSSA::handleRDSV(Instruction *i)
    case SV_FACE:
       bld.mkInterp(NV50_IR_INTERP_FLAT, def, addr, NULL);
       if (i->dType == TYPE_F32) {
-         bld.mkOp2(OP_AND, TYPE_U32, def, def, bld.mkImm(0x80000000));
-         bld.mkOp2(OP_XOR, TYPE_U32, def, def, bld.mkImm(0xbf800000));
+         bld.mkOp2(OP_OR, TYPE_U32, def, def, bld.mkImm(0x00000001));
+         bld.mkOp1(OP_NEG, TYPE_S32, def, def);
+         bld.mkCvt(OP_CVT, TYPE_F32, def, TYPE_S32, def);
       }
       break;
    case SV_NCTAID:
diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp
index ff48e94..04cd19e 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp
@@ -1455,8 +1455,9 @@ NVC0LoweringPass::handleRDSV(Instruction *i)
       Value *face = i->getDef(0);
       bld.mkInterp(NV50_IR_INTERP_FLAT, face, addr, NULL);
       if (i->dType == TYPE_F32) {
-         bld.mkOp2(OP_AND, TYPE_U32, face, face, bld.mkImm(0x80000000));
-         bld.mkOp2(OP_XOR, TYPE_U32, face, face, bld.mkImm(0xbf800000));
+         bld.mkOp2(OP_OR, TYPE_U32, face, face, bld.mkImm(0x00000001));
+         bld.mkOp1(OP_NEG, TYPE_S32, face, face);
+         bld.mkCvt(OP_CVT, TYPE_F32, face, TYPE_S32, face);
       }
    }
       break;
-- 
2.0.5

_______________________________________________
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau