[AMDGPU] Extend buffer intrinsics with swizzling

Summary:
Extend cachepolicy operand in the new VMEM buffer intrinsics
to supply information whether the buffer data is swizzled.
Also, propagate this information to MIR.

Intrinsics updated:
int_amdgcn_raw_buffer_load
int_amdgcn_raw_buffer_load_format
int_amdgcn_raw_buffer_store
int_amdgcn_raw_buffer_store_format
int_amdgcn_raw_tbuffer_load
int_amdgcn_raw_tbuffer_store
int_amdgcn_struct_buffer_load
int_amdgcn_struct_buffer_load_format
int_amdgcn_struct_buffer_store
int_amdgcn_struct_buffer_store_format
int_amdgcn_struct_tbuffer_load
int_amdgcn_struct_tbuffer_store

Furthermore, disable merging of VMEM buffer instructions
in SI Load/Store optimizer, if the "swizzled" bit on the instruction
is on.

The default value of the bit is 0, meaning that data in buffer
is linear and buffer instructions can be merged.

There is no difference in the generated code with this commit.
However, in the future it will be expected that front-ends
use buffer intrinsics with correct "swizzled" bit set.

Reviewers: arsenm, nhaehnle, tpr

Reviewed By: nhaehnle

Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, arphaman, jfb, Petar.Avramovic, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D68200

llvm-svn: 373491

1 files changed, 40 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.buffer.load.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.buffer.load.ll
index 958a72566b5..9f18f4df40b 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.buffer.load.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.buffer.load.ll
@@ -400,6 +400,46 @@ main_body:
   ret void
 }
 
+;CHECK-LABEL: {{^}}raw_buffer_load_x1_offset_merged:
+;CHECK-NEXT: %bb.
+;CHECK-NEXT: buffer_load_dwordx4 v[{{[0-9]}}:{{[0-9]}}], off, s[0:3], 0 offset:4
+;CHECK-NEXT: buffer_load_dwordx2 v[{{[0-9]}}:{{[0-9]}}], off, s[0:3], 0 offset:28
+;CHECK: s_waitcnt
+define amdgpu_ps void @raw_buffer_load_x1_offset_merged(<4 x i32> inreg %rsrc) {
+main_body:
+  %r1 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 4, i32 0, i32 0)
+  %r2 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 8, i32 0, i32 0)
+  %r3 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 12, i32 0, i32 0)
+  %r4 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 16, i32 0, i32 0)
+  %r5 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 28, i32 0, i32 0)
+  %r6 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 32, i32 0, i32 0)
+  call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %r1, float %r2, float %r3, float %r4, i1 true, i1 true)
+  call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %r5, float %r6, float undef, float undef, i1 true, i1 true)
+  ret void
+}
+
+;CHECK-LABEL: {{^}}raw_buffer_load_x1_offset_swizzled_not_merged:
+;CHECK-NEXT: %bb.
+;CHECK-NEXT: buffer_load_dword v{{[0-9]}}, off, s[0:3], 0 offset:4
+;CHECK-NEXT: buffer_load_dword v{{[0-9]}}, off, s[0:3], 0 offset:8
+;CHECK-NEXT: buffer_load_dword v{{[0-9]}}, off, s[0:3], 0 offset:12
+;CHECK-NEXT: buffer_load_dword v{{[0-9]}}, off, s[0:3], 0 offset:16
+;CHECK-NEXT: buffer_load_dword v{{[0-9]}}, off, s[0:3], 0 offset:28
+;CHECK-NEXT: buffer_load_dword v{{[0-9]}}, off, s[0:3], 0 offset:32
+;CHECK: s_waitcnt
+define amdgpu_ps void @raw_buffer_load_x1_offset_swizzled_not_merged(<4 x i32> inreg %rsrc) {
+main_body:
+  %r1 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 4, i32 0, i32 8)
+  %r2 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 8, i32 0, i32 8)
+  %r3 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 12, i32 0, i32 8)
+  %r4 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 16, i32 0, i32 8)
+  %r5 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 28, i32 0, i32 8)
+  %r6 = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 32, i32 0, i32 8)
+  call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %r1, float %r2, float %r3, float %r4, i1 true, i1 true)
+  call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %r5, float %r6, float undef, float undef, i1 true, i1 true)
+  ret void
+}
+
 declare float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32>, i32, i32, i32) #0
 declare <2 x float> @llvm.amdgcn.raw.buffer.load.v2f32(<4 x i32>, i32, i32, i32) #0
 declare <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32>, i32, i32, i32) #0