Generate vectorized version of byteaddress load/store methods (#4036)

Fixes #3533

- Add logic to perform aligned memory operations for loading from and storing to composite resources, like vectors within the ByteAddress legalize pass.

- Checks 
Added a new test for byte address with/without alignment.

---------

Co-authored-by: Yong He <yonghe@outlook.com>

1 files changed, 116 insertions, 0 deletions

diff --git a/tests/compute/byte-address-buffer-aligned.slang b/tests/compute/byte-address-buffer-aligned.slang
new file mode 100644
index 000000000..5024987aa
--- /dev/null
+++ b/tests/compute/byte-address-buffer-aligned.slang
@@ -0,0 +1,116 @@
+// byte-address-buffer-aligned.slang
+//TEST(compute, vulkan):COMPARE_COMPUTE_EX(filecheck-buffer=CHECK):-slang -compute -d3d12 -profile cs_6_0 -use-dxil -shaderobj -output-using-type
+//DISABLED_TEST(compute, vulkan):COMPARE_COMPUTE_EX(filecheck-buffer=CHECK):-slang -compute -vk -shaderobj -output-using-type
+
+//TEST:SIMPLE(filecheck=CHECK1):-target glsl -entry computeMain -stage compute
+//TEST:SIMPLE(filecheck=CHECK2):-target hlsl -entry computeMain -stage compute
+//TEST:SIMPLE(filecheck=CHECK3):-target spirv -entry computeMain -stage compute
+//TEST:SIMPLE(filecheck=CHECK3):-target spirv -emit-spirv-directly -entry computeMain -stage compute
+
+// Confirm compilation of `(RW)ByteAddressBuffer` with aligned load / stores to wider data types.
+
+[vk::binding(2, 3)] RWByteAddressBuffer buffer0;
+
+[shader("compute")]
+[numthreads(1,1,1)]
+void computeMain(uint3 threadId : SV_DispatchThreadID)
+{
+    // CHECK-NOT: warning
+
+    // CHECK1: vec4 {{.*}} = buffer0_{{.*}}.{{.*}} = buffer0_{{.*}}.{{.*}};
+    // CHECK1: vec4 {{.*}} = buffer0_{{.*}}.{{.*}} = vec4(buffer0_{{.*}}.{{.*}}, buffer0_{{.*}}.{{.*}}, buffer0_{{.*}}.{{.*}}, buffer0_{{.*}}.{{.*}});
+    // CHECK1: vec4 {{.*}} = buffer0_{{.*}}.{{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = _S3[{{.*}}];
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = _S3[{{.*}}];
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = _S3[{{.*}}];
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = _S3[{{.*}}];
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = buffer0_{{.*}}.{{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = {{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = {{.*}};
+    // CHECK1: float {{.*}} = buffer0_{{.*}}.{{.*}} = {{.*}};
+
+    // CHECK2: float4 {{.*}} = (buffer0_0).Load<float4 >(int(32));
+    // CHECK2: buffer0_0.Store(int(32),{{.*}});
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(8));
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(12));
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(16));
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(20));
+    // CHECK2: buffer0_0.Store(int(32),float4({{.*}}, {{.*}}, {{.*}}, {{.*}}));
+    // CHECK2: float4 {{.*}} = (buffer0_0).Load<float4 >(int(32));
+    // CHECK2: buffer0_0.Store(int(8),{{.*}}[int(0)]);
+    // CHECK2: buffer0_0.Store(int(12),{{.*}}[int(1)]);
+    // CHECK2: buffer0_0.Store(int(16),{{.*}}[int(2)]);
+    // CHECK2: buffer0_0.Store(int(20),{{.*}}[int(3)]);
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(8));
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(12));
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(16));
+    // CHECK2: float {{.*}} = (buffer0_0).Load<float >(int(20));
+    // CHECK2: buffer0_0.Store(int(8),{{.*}});
+    // CHECK2: buffer0_0.Store(int(12),{{.*}});
+    // CHECK2: buffer0_0.Store(int(16),{{.*}});
+    // CHECK2: buffer0_0.Store(int(20),{{.*}});
+
+    // CHECK3-DAG: %[[v4f:[a-zA-Z0-9_]+]] = OpTypeVector %float 4
+    // CHECK3-DAG: %[[SBv4f:[a-zA-Z0-9_]+]] = OpTypePointer StorageBuffer %[[v4f]]
+    // CHECK3-DAG: %[[RTv4f:[a-zA-Z0-9_]+]] = OpTypeRuntimeArray %[[v4f]]
+    // CHECK3-DAG: %[[RWBv4f:[a-zA-Z0-9_]+]] = OpTypeStruct %[[RTv4f]]
+    // CHECK3-DAG: %[[SBRW:[a-zA-Z0-9_]+]] = OpTypePointer StorageBuffer %[[RWBv4f]]
+    // CHECK3-DAG: %[[RTf:[a-zA-Z0-9_]+]] = OpTypeRuntimeArray %float
+    // CHECK3-DAG: %[[RWBf:[a-zA-Z0-9_]+]] = OpTypeStruct %[[RTf]]
+    // CHECK3-DAG: %[[SBRW0:[a-zA-Z0-9_]+]] = OpTypePointer StorageBuffer %[[RWBf]]
+    // CHECK3-DAG: %[[SBf:[a-zA-Z0-9_]+]] = OpTypePointer StorageBuffer %float
+    // CHECK3-DAG: %[[BUF0:[a-zA-Z0-9_]+]] =  OpVariable %[[SBRW]] StorageBuffer
+    // CHECK3-DAG: %[[BUF00:[a-zA-Z0-9_]+]] =  OpVariable %[[SBRW0]] StorageBuffer
+    // CHECK3-DAG: %[[V0:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBv4f]] %[[BUF0]] %{{.*}}
+    // CHECK3-DAG: %[[V1:[a-zA-Z0-9_]+]] = OpLoad %[[v4f]]  %[[V0]]
+    // CHECK3-DAG: %[[V2:[a-zA-Z0-9_]+]] = OpAccessChain  %[[SBv4f]] %[[BUF0]] %{{.*}}
+    // CHECK3-DAG: OpStore %[[V2]] %[[V1]]
+    // CHECK3-DAG: %[[V3:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V4:[a-zA-Z0-9_]+]] = OpLoad %float %[[V3]]
+    // CHECK3-DAG: %[[V5:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V6:[a-zA-Z0-9_]+]] = OpLoad %float %[[V5]]
+    // CHECK3-DAG: %[[V7:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V8:[a-zA-Z0-9_]+]] = OpLoad %float %[[V7]]
+    // CHECK3-DAG: %[[V9:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V10:[a-zA-Z0-9_]+]] = OpLoad %float %[[V9]]
+    // CHECK3-DAG: %[[V11:[a-zA-Z0-9_]+]] = OpCompositeConstruct %[[v4f]] %[[V4]] %[[V6]] %[[V8]] %[[V10]]
+    // CHECK3-DAG: %[[V12:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBv4f]] %[[BUF0]]
+    // CHECK3-DAG: OpStore %[[V12]] %[[V11]]
+    // CHECK3-DAG: %[[V13:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBv4f]] %[[BUF0]]
+    // CHECK3-DAG: %[[V14:[a-zA-Z0-9_]+]] = OpLoad %[[v4f]] %[[V13]]
+    // CHECK3-DAG: %[[V15:[a-zA-Z0-9_]+]] = OpCompositeExtract %float %[[V14]] 0
+    // CHECK3-DAG: %[[V16:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V16]] %[[V15]]
+    // CHECK3-DAG: %[[V17:[a-zA-Z0-9_]+]] = OpCompositeExtract %float %[[V14]] 1
+    // CHECK3-DAG: %[[V18:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V18]] %[[V17]]
+    // CHECK3-DAG: %[[V19:[a-zA-Z0-9_]+]] = OpCompositeExtract %float %[[V14]] 2
+    // CHECK3-DAG: %[[V20:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V20]] %[[V19]]
+    // CHECK3-DAG: %[[V21:[a-zA-Z0-9_]+]] = OpCompositeExtract %float %[[V14]] 3
+    // CHECK3-DAG: %[[V22:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V22]] %[[V21]]
+    // CHECK3-DAG: %[[V23:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V24:[a-zA-Z0-9_]+]] = OpLoad %float %[[V23]]
+    // CHECK3-DAG: %[[V25:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V26:[a-zA-Z0-9_]+]] = OpLoad %float %[[V25]]
+    // CHECK3-DAG: %[[V27:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V28:[a-zA-Z0-9_]+]] = OpLoad %float %[[V27]]
+    // CHECK3-DAG: %[[V29:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: %[[V30:[a-zA-Z0-9_]+]] = OpLoad %float %[[V29]]
+    // CHECK3-DAG: %[[V31:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V31]] %[[V24]]
+    // CHECK3-DAG: %[[V32:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V32]] %[[V26]]
+    // CHECK3-DAG: %[[V33:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V33]] %[[V28]]
+    // CHECK3-DAG: %[[V34:[a-zA-Z0-9_]+]] = OpAccessChain %[[SBf]] %[[BUF00]]
+    // CHECK3-DAG: OpStore %[[V34]] %[[V30]]
+    buffer0.Store(32, buffer0.Load<float4>(32));
+	buffer0.Store(32, buffer0.Load<float4>(8));
+	buffer0.Store(8, buffer0.Load<float4>(32));
+	buffer0.Store(8, buffer0.Load<float4>(8));
+}