Nouveau - Jun 2014 - [PATCH v3 0/4] Constant folding of new Instructions

Yet another try for constant folding of Instructions for nvc0.

Please Review this again! (Hopefully the last time ;-) )


Tobias Klausmann (4):
  nvc0/ir: clear subop when folding constant expressions
  nvc0/ir: Handle reverse subop for OP_EXTBF when folding constant
    expressions
  nvc0/ir: Handle OP_BFIND when folding constant expressions
  nvc0/ir: Handle OP_POPCNT when folding constant expressions

 .../drivers/nouveau/codegen/nv50_ir_peephole.cpp   | 41 ++++++++++++++++++++--
 1 file changed, 39 insertions(+), 2 deletions(-)

-- 
1.8.4.5

Some operations (e.g. OP_MUL/OP_MAD/OP_EXTBF might have a subop set.
After folding, make sure that it is cleared

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at mni.thm.de>
Reviewed-by: Ilia Mirkin <imirkin at alum.mit.edu>
---
 src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
index 1a2c2e6..58092f4 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
@@ -563,6 +563,7 @@ ConstantFolding::expr(Instruction *i,
    } else {
       i->op = i->saturate ? OP_SAT : OP_MOV; /* SAT handled by unary() */
    }
+   i->subOp = 0;
 }
 
 void
-- 
1.8.4.5

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at mni.thm.de>
---
V2: Handle the instruction right (shift after reverse)
V3: Reverse once, not independently for every TYPE

 src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
index 58092f4..538e745 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
@@ -528,9 +528,17 @@ ConstantFolding::expr(Instruction *i,
          rshift = 32 - width;
          lshift = 32 - width - offset;
       }
+      if (i->subOp == NV50_IR_SUBOP_EXTBF_REV)
+         res.data.u32 = util_bitreverse(a->data.u32);
+      else
+         res.data.u32 = a->data.u32;
       switch (i->dType) {
-      case TYPE_S32: res.data.s32 = (a->data.s32 << lshift) >>
rshift; break;
-      case TYPE_U32: res.data.u32 = (a->data.u32 << lshift) >>
rshift; break;
+      case TYPE_S32:
+         res.data.s32 = (res.data.s32 << lshift) >> rshift;
+         break;
+      case TYPE_U32:
+         res.data.u32 = (res.data.u32 << lshift) >> rshift;
+         break;
       default:
          return;
       }
-- 
1.8.4.5

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at mni.thm.de>
---
V3: Use BuildUtil for the Immediate instead of a type conversion

 .../drivers/nouveau/codegen/nv50_ir_peephole.cpp        | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
index 538e745..e4d91d7 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
@@ -944,6 +944,23 @@ ConstantFolding::opnd(Instruction *i, ImmediateValue
&imm0, int s)
    case OP_EX2:
       unary(i, imm0);
       break;
+   case OP_BFIND: {
+      int32_t res;
+      switch (i->dType) {
+      case TYPE_S32: res = util_last_bit_signed(imm0.reg.data.s32) - 1; break;
+      case TYPE_U32: res = util_last_bit(imm0.reg.data.u32) - 1; break;
+      default:
+         return;
+      }
+      if (i->subOp ==  NV50_IR_SUBOP_BFIND_SAMT && res >= 0)
+         res = 31 - res;
+      bld.setPosition(i, false); /* make sure bld is init'ed */
+      i->setSrc(0, bld.mkImm(res));
+      i->setSrc(1, NULL);
+      i->op = OP_MOV;
+      i->subOp = 0;
+      break;
+   }
    default:
       return;
    }
-- 
1.8.4.5

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at mni.thm.de>
---
V2: Add support for a single-argument version of POPCNT for Maxwell (SM5)
V3: Clean up a bit more

 src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
index e4d91d7..8dde7c1 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp
@@ -544,6 +544,8 @@ ConstantFolding::expr(Instruction *i,
       }
       break;
    }
+   case OP_POPCNT: res.data.u32 = util_bitcount(a->data.u32 &
b->data.u32);
+      break;
    default:
       return;
    }
@@ -961,6 +963,15 @@ ConstantFolding::opnd(Instruction *i, ImmediateValue
&imm0, int s)
       i->subOp = 0;
       break;
    }
+   case OP_POPCNT: {
+      uint32_t res;
+      res = util_bitcount(imm0.reg.data.u32);
+      i->setSrc(0, new_ImmediateValue(i->bb->getProgram(), res));
+      i->setSrc(1, NULL);
+      i->op = OP_MOV;
+      i->subOp = 0;
+      break;
+   }
    default:
       return;
    }
-- 
1.8.4.5

Series is Reviewed-by: Ilia Mirkin <imirkin at alum.mit.edu>

There are a few minor formatting items that I'll take care of before
pushing this out later tonight... take a look at the changes I made
for the future :)

On Tue, Jun 3, 2014 at 6:35 PM, Tobias Klausmann
<tobias.johannes.klausmann at mni.thm.de> wrote:
> Yet another try for constant folding of Instructions for nvc0.
>
> Please Review this again! (Hopefully the last time ;-) )
>
>
> Tobias Klausmann (4):
>   nvc0/ir: clear subop when folding constant expressions
>   nvc0/ir: Handle reverse subop for OP_EXTBF when folding constant
>     expressions
>   nvc0/ir: Handle OP_BFIND when folding constant expressions
>   nvc0/ir: Handle OP_POPCNT when folding constant expressions
>
>  .../drivers/nouveau/codegen/nv50_ir_peephole.cpp   | 41
++++++++++++++++++++--
>  1 file changed, 39 insertions(+), 2 deletions(-)
>
> --
> 1.8.4.5
>

Tobias Klausmann <tobias.johannes.klausmann at mni.thm.de>
writes:
> Some operations (e.g. OP_MUL/OP_MAD/OP_EXTBF might have a subop set.
> After folding, make sure that it is cleared
>
> Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at
mni.thm.de>
> Reviewed-by: Ilia Mirkin <imirkin at alum.mit.edu>
Cc: "10.1 10.2" <mesa-stable at lists.freedesktop.org>

Hi Tobias and Ilia,

This patch isn't picking cleanly over to the 10.1 branch.

Can you give me some guidance here? Either of the following replies
would be great:

	Don't worry about this for 10.1 because...

	Here's a backported patch for 10.1...

Thanks!

-Carl

-- 
carl.d.worth at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL:
<http://lists.freedesktop.org/archives/nouveau/attachments/20140623/060f9933/attachment.sig>