[llvm] 3680b57 - AMDGPU: Add baseline tests for mbcnt.hi combine

[Matt Arsenault via llvm-commits]

Author: Matt Arsenault
Date: 2023-06-30T15:04:03-04:00
New Revision: 3680b57a88233cd87928aa6741c5e04522b3b4a5

URL: https://github.com/llvm/llvm-project/commit/3680b57a88233cd87928aa6741c5e04522b3b4a5
DIFF: https://github.com/llvm/llvm-project/commit/3680b57a88233cd87928aa6741c5e04522b3b4a5.diff

LOG: AMDGPU: Add baseline tests for mbcnt.hi combine

Added: 
    llvm/test/Transforms/InstCombine/AMDGPU/mbcnt.ll

Modified: 
    

Removed: 
    


################################################################################
diff  --git a/llvm/test/Transforms/InstCombine/AMDGPU/mbcnt.ll b/llvm/test/Transforms/InstCombine/AMDGPU/mbcnt.ll
new file mode 100644
index 00000000000000..24505881ea0904
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/AMDGPU/mbcnt.ll
@@ -0,0 +1,99 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --scrub-attributes --version 2
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=instcombine < %s | FileCheck --check-prefixes=DEFAULT,CHECK %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mattr=+wavefrontsize32 -passes=instcombine < %s | FileCheck --check-prefixes=WAVE32,CHECK %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mattr=+wavefrontsize64 -passes=instcombine < %s | FileCheck --check-prefixes=WAVE64,CHECK %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1030 -passes=instcombine < %s | FileCheck --check-prefixes=WAVE32,CHECK %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1030 -mattr=+wavefrontsize64 -passes=instcombine < %s | FileCheck --check-prefixes=WAVE64,CHECK %s
+
+declare i32 @llvm.amdgcn.mbcnt.lo(i32, i32)
+declare i32 @llvm.amdgcn.mbcnt.hi(i32, i32)
+
+define i32 @mbcnt_lo(i32 %x, i32 %y) {
+; DEFAULT-LABEL: define i32 @mbcnt_lo
+; DEFAULT-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) {
+; DEFAULT-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
+; DEFAULT-NEXT:    ret i32 [[LO]]
+;
+; WAVE32-LABEL: define i32 @mbcnt_lo
+; WAVE32-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR1:[0-9]+]] {
+; WAVE32-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
+; WAVE32-NEXT:    ret i32 [[LO]]
+;
+; WAVE64-LABEL: define i32 @mbcnt_lo
+; WAVE64-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR1:[0-9]+]] {
+; WAVE64-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
+; WAVE64-NEXT:    ret i32 [[LO]]
+;
+  %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 %x, i32 %y)
+  ret i32 %lo
+}
+
+define i32 @mbcnt_hi(i32 %x, i32 %y) {
+; DEFAULT-LABEL: define i32 @mbcnt_hi
+; DEFAULT-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) {
+; DEFAULT-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[X]], i32 [[Y]])
+; DEFAULT-NEXT:    ret i32 [[HI]]
+;
+; WAVE32-LABEL: define i32 @mbcnt_hi
+; WAVE32-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR1]] {
+; WAVE32-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[X]], i32 [[Y]])
+; WAVE32-NEXT:    ret i32 [[HI]]
+;
+; WAVE64-LABEL: define i32 @mbcnt_hi
+; WAVE64-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR1]] {
+; WAVE64-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[X]], i32 [[Y]])
+; WAVE64-NEXT:    ret i32 [[HI]]
+;
+  %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 %x, i32 %y)
+  ret i32 %hi
+}
+
+define i32 @mbcnt_lo_hi(i32 %x, i32 %y, i32 %z) {
+; DEFAULT-LABEL: define i32 @mbcnt_lo_hi
+; DEFAULT-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]], i32 [[Z:%.*]]) {
+; DEFAULT-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
+; DEFAULT-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[Z]], i32 [[LO]])
+; DEFAULT-NEXT:    ret i32 [[HI]]
+;
+; WAVE32-LABEL: define i32 @mbcnt_lo_hi
+; WAVE32-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]], i32 [[Z:%.*]]) #[[ATTR1]] {
+; WAVE32-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
+; WAVE32-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[Z]], i32 [[LO]])
+; WAVE32-NEXT:    ret i32 [[HI]]
+;
+; WAVE64-LABEL: define i32 @mbcnt_lo_hi
+; WAVE64-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]], i32 [[Z:%.*]]) #[[ATTR1]] {
+; WAVE64-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
+; WAVE64-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[Z]], i32 [[LO]])
+; WAVE64-NEXT:    ret i32 [[HI]]
+;
+  %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 %x, i32 %y)
+  %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 %z, i32 %lo)
+  ret i32 %hi
+}
+
+define i32 @ockl_lane_u32() {
+; DEFAULT-LABEL: define i32 @ockl_lane_u32() {
+; DEFAULT-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
+; DEFAULT-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 [[LO]])
+; DEFAULT-NEXT:    ret i32 [[HI]]
+;
+; WAVE32-LABEL: define i32 @ockl_lane_u32
+; WAVE32-SAME: () #[[ATTR1]] {
+; WAVE32-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
+; WAVE32-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 [[LO]])
+; WAVE32-NEXT:    ret i32 [[HI]]
+;
+; WAVE64-LABEL: define i32 @ockl_lane_u32
+; WAVE64-SAME: () #[[ATTR1]] {
+; WAVE64-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
+; WAVE64-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 [[LO]])
+; WAVE64-NEXT:    ret i32 [[HI]]
+;
+  %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
+  %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
+  ret i32 %hi
+}
+
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; CHECK: {{.*}}