blob: e16abb30767fb2e396fdbd92028fb70c48f7c6bc (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
|
// No atomic support on CPU
//DISABLE_TEST(compute):COMPARE_COMPUTE_EX:-cpu -compute -shaderobj
// No support for int64_t on DX11
//DISABLE_TEST(compute):COMPARE_COMPUTE_EX:-slang -compute -shaderobj
// No support for int64_t on fxc - we need SM6.0 and dxil
// https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/hlsl-shader-model-6-0-features-for-direct3d-12
//DISABLE_TEST(compute):COMPARE_COMPUTE_EX:-slang -compute -dx12 -nvapi-slot u0 -shaderobj
//TEST(compute):COMPARE_COMPUTE_EX:-slang -compute -dx12 -profile cs_6_0 -render-features atomic-int64 -compile-arg -O2 -shaderobj
//TEST(compute, vulkan):COMPARE_COMPUTE_EX:-vk -compute -render-features atomic-int64 -shaderobj
//TEST(compute):COMPARE_COMPUTE_EX:-cuda -compute -shaderobj
// The test doesn't directly use this, but having this defined makes the 0 slot available if NVAPI is going to be used
// Only strictly necessary on the D3D12 path
//TEST_INPUT:ubuffer(data=[0 0 0 0 ], stride=4):name=nvapiBuffer
RWStructuredBuffer<int> nvapiBuffer;
//TEST_INPUT:ubuffer(data=[0 1 2 3 4 5 6 7]):out,name=outputBuffer
RWByteAddressBuffer outputBuffer;
[numthreads(16, 1, 1)]
void computeMain(int3 dispatchThreadID : SV_DispatchThreadID)
{
int tid = dispatchThreadID.x;
int idx = (tid & 3) ^ (tid >> 2);
int64_t previousValue = 0;
outputBuffer.InterlockedAddI64((idx << 3), 1, previousValue);
int anotherIdx = tid >> 2;
outputBuffer.InterlockedAddI64(anotherIdx << 3, 3);
// Bit logical
outputBuffer.InterlockedOrU64((idx << 3), (uint64_t(2) << 32) | (tid << 4));
outputBuffer.InterlockedXorU64((idx << 3), tid << 8);
outputBuffer.InterlockedAndU64((idx << 3), (uint64_t(tid | 2) << 32) | 0xffffffffffffffffULL);
}
|
