yum-mirror/slang

Making it easier to work with shaders

git clone https://git.yummers.dev/yum-mirror/slang

Harsh Aggarwal (NVIDIA)Fix#8084: Batch-8: Enable cuda tests (#8268)1562f98c0

master
branchesmasterall refs →
2.5 KiB57 linesraw
1//IGNORE_TEST:CPU_REFLECTION: -profile cs_5_0 -entry computeMain -target cpp
2//DISABLED_TEST(compute):COMPARE_COMPUTE_EX:-cpu -compute
3//TEST:SIMPLE(filecheck=DXIL):-target dxbc-assembly -entry computeMain -profile cs_5_1
4//TEST:SIMPLE(filecheck=CUDA):-target cuda -entry computeMain -profile cs_5_1
5//TEST:SIMPLE(filecheck=CHECK):-target spirv-assembly -entry computeMain -profile cs_5_1 -emit-spirv-via-glsl
6//DISABLED_TEST(compute):COMPARE_COMPUTE_EX:-cuda -compute
7
8//TEST_INPUT:ubuffer(data=[0 0 0 0 0 0 0 0], stride=4):out,name outputBuffer
9RWStructuredBuffer<int> outputBuffer;
10
11//TEST_INPUT:array(size=2):name g_aoa
12RWStructuredBuffer<int> g_aoa[];
13
14//TEST_INPUT:ubuffer(data=[1 2 3 4], stride=4):name=g_aoa[0]
15//TEST_INPUT:ubuffer(data=[8 17 34], stride=4):name=g_aoa[1]
16
17[numthreads(8, 1, 1)]
18void computeMain(uint3 dispatchThreadID : SV_DispatchThreadID)
19{
20    // DXIL: cs_5_1
21    // CHECK: OpCapability {{(ShaderNonUniform|StorageBufferArrayNonUniformIndexing)}}
22    // CHECK: OpCapability {{(ShaderNonUniform|StorageBufferArrayNonUniformIndexing)}}
23    // CHECK-DAG: OpDecorate %[[N1:[a-zA-Z0-9_]+]] NonUniform
24    // CHECK-DAG: OpDecorate %[[N2:[a-zA-Z0-9_]+]] NonUniform
25    // CHECK-DAG: OpDecorate %[[N3:[a-zA-Z0-9_]+]] NonUniform
26    // CHECK-DAG: OpDecorate %[[N4:[a-zA-Z0-9_]+]] NonUniform
27    // CHECK-DAG: OpDecorate %[[B0:[a-zA-Z0-9_]+]] Binding 0
28    // CHECK-DAG: OpDecorate %[[B1:[a-zA-Z0-9_]+]] Binding 1
29    // CHECK-DAG: %[[P1:[a-zA-Z0-9_]+]] = OpTypePointer Uniform %int
30    // CHECK-DAG: %[[P2:[a-zA-Z0-9_]+]] = OpTypePointer Uniform %StructuredBuffer_int_t_0
31    // CHECK: %[[N1]] = OpCopyObject %int
32    // CHECK: %{{.*}} = OpAccessChain %[[P2]] %[[B1]] %[[N1]]
33    // CHECK: %[[N2]] = OpCopyObject %int
34    // CHECK: %[[N3]] = OpAccessChain %[[P1]] %[[B1]] %[[N2]]
35    // CHECK: %[[N4]] = OpLoad %int %[[N3]]
36    // CHECK: %{{.*}} = OpAccessChain %[[P1]] %[[B0]] %int_{{.*}}
37    // CHECK: OpStore %{{.*}} %[[N4]]
38    // CUDA: Array<RWStructuredBuffer<int>>  g_aoa_0;
39    // CUDA: globalParams_0->g_aoa_0[index_0 >> int(2)].GetDimensions
40    // CUDA: _S1[index_0 >> int(2)]
41    int index = int(dispatchThreadID.x);
42    
43    int baseIndex = index >> 2;
44    int innerIndex = index & 3;
45
46    RWStructuredBuffer<int> buffer = g_aoa[NonUniformResourceIndex(baseIndex)];
47
48    // Get the size 
49    uint bufferCount, bufferStride;
50    buffer.GetDimensions(bufferCount, bufferStride);
51
52    if (innerIndex >= bufferCount)
53    {
54        innerIndex = int(bufferCount - 1);
55    }
56	outputBuffer[index] = buffer[innerIndex]; 
57}