AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

nir: Drop imov/fmov in favor of one mov instruction

Browse Source 
The difference between imov and fmov has been a constant source of
confusion in NIR for years.  No one really knows why we have two or when
to use one vs. the other.  The real reason is that they do different
things in the presence of source and destination modifiers.  However,
without modifiers (which many back-ends don't have), they are identical.
Now that we've reworked nir_lower_to_source_mods to leave one abs/neg
instruction in place rather than replacing them with imov or fmov
instructions, we don't need two different instructions at all anymore.

Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Acked-by: Rob Clark <robdclark@chromium.org>
This commit is contained in:
Jason Ekstrand
2019-05-06 11:45:46 -05:00
parent 22421ca7be
commit f2dc0f2872
40 changed files with 80 additions and 106 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/compiler/nir/nir_lower_bool_to_float.c
+2 -3
View File
@@ -58,8 +58,8 @@ lower_alu_instr(nir_builder *b, nir_alu_instr *alu)
/* These we expect to have booleans but the opcode doesn't change */
break;
case nir_op_b2f32: alu->op = nir_op_fmov; break;
case nir_op_b2i32: alu->op = nir_op_fmov; break;
case nir_op_b2f32: alu->op = nir_op_mov; break;
case nir_op_b2i32: alu->op = nir_op_mov; break;
case nir_op_f2b1:
case nir_op_i2b1:
rep = nir_sne(b, nir_ssa_for_alu_src(b, alu, 0),
@@ -92,7 +92,6 @@ lower_alu_instr(nir_builder *b, nir_alu_instr *alu)
case nir_op_bcsel: alu->op = nir_op_fcsel; break;
case nir_op_imov: alu->op = nir_op_fmov; break;
case nir_op_iand: alu->op = nir_op_fmul; break;
case nir_op_ixor: alu->op = nir_op_sne; break;
case nir_op_ior: alu->op = nir_op_fmax; break;