[HIP][HIPIFY] Add some missing flags for cooperative launch and occupancy APIs

[ROCm/clr commit: 6db9b782be]

projects/clr/hipamd/include/hip/hcc_detail/hip_runtime_api.h

@@ -212,6 +212,11 @@ enum hipLimit_t {
 #define hipArrayCubemap 0x04
 #define hipArrayTextureGather 0x08
 
+#define hipOccupancyDefault 0x00
+
+#define hipCooperativeLaunchMultiDeviceNoPreSync 0x01
+#define hipCooperativeLaunchMultiDeviceNoPostSync 0x02
+
 /*
  * @brief hipJitOption
  * @enum
@@ -2903,7 +2908,7 @@ hipError_t hipModuleLaunchKernel(hipFunction_t f, unsigned int gridDimX, unsigne
  * @param [in] stream    Stream where the kernel should be dispatched.  May be 0, in which case th
  * default stream is used with associated synchronization rules.
  *
- * @returns hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue
+ * @returns hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue, hipErrorCooperativeLaunchTooLarge
  */
 hipError_t hipLaunchCooperativeKernel(const void* f, dim3 gridDim, dim3 blockDimX,
                                       void** kernelParams, unsigned int sharedMemBytes,
@@ -2917,7 +2922,7 @@ hipError_t hipLaunchCooperativeKernel(const void* f, dim3 gridDim, dim3 blockDim
  * @param [in] numDevices               Size of the launchParamsList array.
  * @param [in] flags                    Flags to control launch behavior.
  *
- * @returns hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue
+ * @returns hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue, hipErrorCooperativeLaunchTooLarge
  */
 hipError_t hipLaunchCooperativeKernelMultiDevice(hipLaunchParams* launchParamsList,
                                                  int  numDevices, unsigned int  flags);
@@ -2960,7 +2965,7 @@ hipError_t hipOccupancyMaxActiveBlocksPerMultiprocessor(
  * @param [in]  flags            Extra flags for occupancy calculation (currently ignored)
  */
 hipError_t hipOccupancyMaxActiveBlocksPerMultiprocessorWithFlags(
-   int* numBlocks, const void* f, int blockSize, size_t dynSharedMemPerBlk, unsigned int flags);
+   int* numBlocks, const void* f, int blockSize, size_t dynSharedMemPerBlk, unsigned int flags __dparm(hipOccupancyDefault));
 
 #if __HIP_VDI__ && !defined(__HCC__)
 /**

projects/clr/hipamd/include/hip/hip_runtime_api.h

@@ -255,6 +255,10 @@ typedef enum __HIP_NODISCARD hipError_t {
         713,  ///< Produced when trying to unlock a non-page-locked memory.
     hipErrorLaunchFailure =
         719,  ///< An exception occurred on the device while executing a kernel.
+    hipErrorCooperativeLaunchTooLarge =
+        720,  ///< This error indicates that the number of blocks launched per grid for a kernel
+              ///< that was launched via cooperative launch APIs exceeds the maximum number of
+              ///< allowed blocks for the current device
     hipErrorNotSupported = 801,  ///< Produced when the hip API is not supported/implemented
     hipErrorUnknown = 999,  //< Unknown error.
     // HSA Runtime Error Codes start here.

projects/clr/hipamd/include/hip/nvcc_detail/hip_runtime_api.h

@@ -140,6 +140,12 @@ typedef enum cudaChannelFormatKind hipChannelFormatKind;
 #define hipLimitMallocHeapSize cudaLimitMallocHeapSize
 #define hipIpcMemLazyEnablePeerAccess cudaIpcMemLazyEnablePeerAccess
 
+#define hipOccupancyDefault cudaOccupancyDefault
+
+#define hipCooperativeLaunchMultiDeviceNoPreSync cudaCooperativeLaunchMultiDeviceNoPreSync
+#define hipCooperativeLaunchMultiDeviceNoPostSync cudaCooperativeLaunchMultiDeviceNoPostSync
+
+
 // enum CUjit_option redefines
 #define hipJitOptionMaxRegisters CU_JIT_MAX_REGISTERS
 #define hipJitOptionThreadsPerBlock CU_JIT_THREADS_PER_BLOCK
@@ -267,6 +273,8 @@ inline static hipError_t hipCUDAErrorTohipError(cudaError_t cuError) {
             return hipErrorNotInitialized;
         case cudaErrorLaunchFailure:
             return hipErrorLaunchFailure;
+        case cudaErrorCooperativeLaunchTooLarge:
+            return hipErrorCooperativeLaunchTooLarge;
         case cudaErrorPriorLaunchFailure:
             return hipErrorPriorLaunchFailure;
         case cudaErrorLaunchOutOfResources:
@@ -445,6 +453,8 @@ inline static hipError_t hipCUResultTohipError(CUresult cuError) {
             return hipErrorHostMemoryNotRegistered;
         case CUDA_ERROR_LAUNCH_FAILED:
             return hipErrorLaunchFailure;
+        case CUDA_ERROR_COOPERATIVE_LAUNCH_TOO_LARGE:
+            return hipErrorCooperativeLaunchTooLarge;
         case CUDA_ERROR_NOT_SUPPORTED:
             return hipErrorNotSupported;
         case CUDA_ERROR_UNKNOWN:
@@ -601,6 +611,8 @@ inline static cudaError_t hipErrorToCudaError(hipError_t hError) {
             return cudaErrorSetOnActiveProcess;
         case hipErrorLaunchFailure:
             return cudaErrorLaunchFailure;
+        case hipErrorCooperativeLaunchTooLarge:
+            return cudaErrorCooperativeLaunchTooLarge;
         case hipErrorNotSupported:
             return cudaErrorNotSupported;
         // HSA: does not exist in CUDA