Run DCE after a LoopFlatten test to reduce spurious output [nfc]
[llvm-project.git] / clang / test / CodeGenCUDA / Inputs / cuda.h
blob8df425b77ac823f08950421a0b8712e06c68149d
1 /* Minimal declarations for CUDA support. Testing purposes only. */
3 #include <stddef.h>
5 #if __HIP__ || __CUDA__
6 #define __constant__ __attribute__((constant))
7 #define __device__ __attribute__((device))
8 #define __global__ __attribute__((global))
9 #define __host__ __attribute__((host))
10 #define __shared__ __attribute__((shared))
11 #if __HIP__
12 #define __managed__ __attribute__((managed))
13 #endif
14 #define __launch_bounds__(...) __attribute__((launch_bounds(__VA_ARGS__)))
15 #else
16 #define __constant__
17 #define __device__
18 #define __global__
19 #define __host__
20 #define __shared__
21 #define __managed__
22 #define __launch_bounds__(...)
23 #endif
25 struct dim3 {
26 unsigned x, y, z;
27 __host__ __device__ dim3(unsigned x, unsigned y = 1, unsigned z = 1) : x(x), y(y), z(z) {}
30 #if __HIP__ || HIP_PLATFORM
31 typedef struct hipStream *hipStream_t;
32 typedef enum hipError {} hipError_t;
33 int hipConfigureCall(dim3 gridSize, dim3 blockSize, size_t sharedSize = 0,
34 hipStream_t stream = 0);
35 extern "C" hipError_t __hipPushCallConfiguration(dim3 gridSize, dim3 blockSize,
36 size_t sharedSize = 0,
37 hipStream_t stream = 0);
38 #ifndef __HIP_API_PER_THREAD_DEFAULT_STREAM__
39 extern "C" hipError_t hipLaunchKernel(const void *func, dim3 gridDim,
40 dim3 blockDim, void **args,
41 size_t sharedMem,
42 hipStream_t stream);
43 #else
44 extern "C" hipError_t hipLaunchKernel_spt(const void *func, dim3 gridDim,
45 dim3 blockDim, void **args,
46 size_t sharedMem,
47 hipStream_t stream);
48 #endif // __HIP_API_PER_THREAD_DEFAULT_STREAM__
49 #else
50 typedef struct cudaStream *cudaStream_t;
51 typedef enum cudaError {} cudaError_t;
52 extern "C" int cudaConfigureCall(dim3 gridSize, dim3 blockSize,
53 size_t sharedSize = 0,
54 cudaStream_t stream = 0);
55 extern "C" int __cudaPushCallConfiguration(dim3 gridSize, dim3 blockSize,
56 size_t sharedSize = 0,
57 cudaStream_t stream = 0);
58 extern "C" cudaError_t cudaLaunchKernel(const void *func, dim3 gridDim,
59 dim3 blockDim, void **args,
60 size_t sharedMem, cudaStream_t stream);
61 extern "C" cudaError_t cudaLaunchKernel_ptsz(const void *func, dim3 gridDim,
62 dim3 blockDim, void **args,
63 size_t sharedMem, cudaStream_t stream);
65 #endif
67 extern "C" __device__ int printf(const char*, ...);