6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-11-04 05:17:07 +00:00
Code Issues Projects Releases Wiki Activity

R600/SI: Don't promote f32 select to i32

Browse Source 
This is nice for the instruction patterns, but it complicates
min / max matching. The select doesn't have the correct type and would
require looking through the bitcasts for the real float operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224092 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Matt Arsenault
2014-12-12 02:30:29 +00:00
parent d948b98cba
commit 3d1ca355c4
6 changed files with 9 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
lib/Target/R600/SIISelLowering.cpp
View File

@@ -93,8 +93,6 @@ SITargetLowering::SITargetLowering(TargetMachine &TM) :
setOperationAction(ISD::STORE, MVT::v2i32, Custom); setOperationAction(ISD::STORE, MVT::v2i32, Custom);
setOperationAction(ISD::STORE, MVT::v4i32, Custom); setOperationAction(ISD::STORE, MVT::v4i32, Custom);
setOperationAction(ISD::SELECT, MVT::f32, Promote);
AddPromotedToType(ISD::SELECT, MVT::f32, MVT::i32);
setOperationAction(ISD::SELECT, MVT::i64, Custom); setOperationAction(ISD::SELECT, MVT::i64, Custom);
setOperationAction(ISD::SELECT, MVT::f64, Promote); setOperationAction(ISD::SELECT, MVT::f64, Promote);
AddPromotedToType(ISD::SELECT, MVT::f64, MVT::i64); AddPromotedToType(ISD::SELECT, MVT::f64, MVT::i64);

5
lib/Target/R600/SIInstructions.td
View File

@@ -3184,6 +3184,11 @@ def : Pat <
(V_ALIGNBIT_B32 $a, $a, 8)) (V_ALIGNBIT_B32 $a, $a, 8))
>; >;
def : Pat <
(f32 (select i1:$src2, f32:$src1, f32:$src0)),
(V_CNDMASK_B32_e64 $src0, $src1, $src2)
>;
//============================================================================// //============================================================================//
// Miscellaneous Optimization Patterns // Miscellaneous Optimization Patterns
//============================================================================// //============================================================================//

2
test/CodeGen/R600/seto.ll
View File

@@ -2,7 +2,7 @@
; CHECK-LABEL: {{^}}main: ; CHECK-LABEL: {{^}}main:
; CHECK: v_cmp_o_f32_e64 [[CMP:s\[[0-9]+:[0-9]+\]]], [[SREG:s[0-9]+]], [[SREG]] ; CHECK: v_cmp_o_f32_e64 [[CMP:s\[[0-9]+:[0-9]+\]]], [[SREG:s[0-9]+]], [[SREG]]
; CHECK-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, 1.0, [[CMP]] ; CHECK-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0.0, 1.0, [[CMP]]
define void @main(float %p) { define void @main(float %p) {
main_body: main_body:
%c = fcmp oeq float %p, %p %c = fcmp oeq float %p, %p

2
test/CodeGen/R600/setuo.ll
View File

@@ -2,7 +2,7 @@
; CHECK-LABEL: {{^}}main: ; CHECK-LABEL: {{^}}main:
; CHECK: v_cmp_u_f32_e64 [[CMP:s\[[0-9]+:[0-9]+\]]], [[SREG:s[0-9]+]], [[SREG]] ; CHECK: v_cmp_u_f32_e64 [[CMP:s\[[0-9]+:[0-9]+\]]], [[SREG:s[0-9]+]], [[SREG]]
; CHECK-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0, 1.0, [[CMP]] ; CHECK-NEXT: v_cndmask_b32_e64 {{v[0-9]+}}, 0.0, 1.0, [[CMP]]
define void @main(float %p) { define void @main(float %p) {
main_body: main_body:
%c = fcmp une float %p, %p %c = fcmp une float %p, %p

2
test/CodeGen/R600/sint_to_fp.ll
View File

@@ -42,7 +42,7 @@ define void @sint_to_fp_v4i32(<4 x float> addrspace(1)* %out, <4 x i32> addrspac
; FUNC-LABEL: {{^}}sint_to_fp_i1_f32: ; FUNC-LABEL: {{^}}sint_to_fp_i1_f32:
; SI: v_cmp_eq_i32_e64 [[CMP:s\[[0-9]+:[0-9]\]]], ; SI: v_cmp_eq_i32_e64 [[CMP:s\[[0-9]+:[0-9]\]]],
; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0, 1.0, [[CMP]] ; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0.0, 1.0, [[CMP]]
; SI: buffer_store_dword [[RESULT]], ; SI: buffer_store_dword [[RESULT]],
; SI: s_endpgm ; SI: s_endpgm
define void @sint_to_fp_i1_f32(float addrspace(1)* %out, i32 %in) { define void @sint_to_fp_i1_f32(float addrspace(1)* %out, i32 %in) {

2
test/CodeGen/R600/uint_to_fp.ll
View File

@@ -60,7 +60,7 @@ entry:
; FUNC-LABEL: {{^}}uint_to_fp_i1_to_f32: ; FUNC-LABEL: {{^}}uint_to_fp_i1_to_f32:
; SI: v_cmp_eq_i32_e64 [[CMP:s\[[0-9]+:[0-9]\]]], ; SI: v_cmp_eq_i32_e64 [[CMP:s\[[0-9]+:[0-9]\]]],
; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0, 1.0, [[CMP]] ; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0.0, 1.0, [[CMP]]
; SI: buffer_store_dword [[RESULT]], ; SI: buffer_store_dword [[RESULT]],
; SI: s_endpgm ; SI: s_endpgm
define void @uint_to_fp_i1_to_f32(float addrspace(1)* %out, i32 %in) { define void @uint_to_fp_i1_to_f32(float addrspace(1)* %out, i32 %in) {