Launch API to support the triple-chevron syntax#
Functions | |
hipError_t | hipConfigureCall (dim3 gridDim, dim3 blockDim, size_t sharedMem, hipStream_t stream) |
Configure a kernel launch. More... | |
hipError_t | hipSetupArgument (const void *arg, size_t size, size_t offset) |
Set a kernel argument. More... | |
hipError_t | hipLaunchByPtr (const void *func) |
Launch a kernel. More... | |
hipError_t | __hipPushCallConfiguration (dim3 gridDim, dim3 blockDim, size_t sharedMem, hipStream_t stream) |
Push configuration of a kernel launch. More... | |
hipError_t | __hipPopCallConfiguration (dim3 *gridDim, dim3 *blockDim, size_t *sharedMem, hipStream_t *stream) |
Pop configuration of a kernel launch. More... | |
hipError_t | hipLaunchKernel (const void *function_address, dim3 numBlocks, dim3 dimBlocks, void **args, size_t sharedMemBytes, hipStream_t stream) |
C compliant kernel launch API. More... | |
hipError_t | hipLaunchHostFunc (hipStream_t stream, hipHostFn_t fn, void *userData) |
Enqueues a host function call in a stream. More... | |
hipError_t | hipDrvMemcpy2DUnaligned (const hip_Memcpy2D *pCopy) |
hipError_t | hipExtLaunchKernel (const void *function_address, dim3 numBlocks, dim3 dimBlocks, void **args, size_t sharedMemBytes, hipStream_t stream, hipEvent_t startEvent, hipEvent_t stopEvent, int flags) |
Launches kernel from the pointer address, with arguments and shared memory on stream. More... | |
Detailed Description
This section describes the API to support the triple-chevron syntax.
Function Documentation
◆ __hipPopCallConfiguration()
hipError_t __hipPopCallConfiguration | ( | dim3 * | gridDim, |
dim3 * | blockDim, | ||
size_t * | sharedMem, | ||
hipStream_t * | stream | ||
) |
Pop configuration of a kernel launch.
- Parameters
-
[out] gridDim grid dimension specified as multiple of blockDim. [out] blockDim block dimensions specified in work-items [out] sharedMem Amount of dynamic shared memory to allocate for this kernel. The HIP-Clang compiler provides support for extern shared declarations. [out] stream Stream where the kernel should be dispatched. May be 0, in which case the default stream is used with associated synchronization rules.
Please note, HIP does not support kernel launch with total work items defined in dimension with size gridDim x blockDim >= 2^32.
Please note, HIP does not support kernel launch with total work items defined in dimension with size gridDim x blockDim >= 2^32.
◆ __hipPushCallConfiguration()
hipError_t __hipPushCallConfiguration | ( | dim3 | gridDim, |
dim3 | blockDim, | ||
size_t | sharedMem, | ||
hipStream_t | stream | ||
) |
Push configuration of a kernel launch.
- Parameters
-
[in] gridDim grid dimension specified as multiple of blockDim. [in] blockDim block dimensions specified in work-items [in] sharedMem Amount of dynamic shared memory to allocate for this kernel. The HIP-Clang compiler provides support for extern shared declarations. [in] stream Stream where the kernel should be dispatched. May be 0, in which case the default stream is used with associated synchronization rules.
Please note, HIP does not support kernel launch with total work items defined in dimension with size gridDim x blockDim >= 2^32.
◆ hipConfigureCall()
hipError_t hipConfigureCall | ( | dim3 | gridDim, |
dim3 | blockDim, | ||
size_t | sharedMem, | ||
hipStream_t | stream | ||
) |
Configure a kernel launch.
- Parameters
-
[in] gridDim grid dimension specified as multiple of blockDim. [in] blockDim block dimensions specified in work-items [in] sharedMem Amount of dynamic shared memory to allocate for this kernel. The HIP-Clang compiler provides support for extern shared declarations. [in] stream Stream where the kernel should be dispatched. May be 0, in which case the default stream is used with associated synchronization rules.
Please note, HIP does not support kernel launch with total work items defined in dimension with size gridDim x blockDim >= 2^32.
◆ hipDrvMemcpy2DUnaligned()
hipError_t hipDrvMemcpy2DUnaligned | ( | const hip_Memcpy2D * | pCopy | ) |
Copies memory for 2D arrays.
- Parameters
-
pCopy - Parameters for the memory copy
- Returns
- hipSuccess, hipErrorInvalidValue
◆ hipExtLaunchKernel()
hipError_t hipExtLaunchKernel | ( | const void * | function_address, |
dim3 | numBlocks, | ||
dim3 | dimBlocks, | ||
void ** | args, | ||
size_t | sharedMemBytes, | ||
hipStream_t | stream, | ||
hipEvent_t | startEvent, | ||
hipEvent_t | stopEvent, | ||
int | flags | ||
) |
Launches kernel from the pointer address, with arguments and shared memory on stream.
- Parameters
-
[in] function_address pointer to the Kernel to launch. [in] numBlocks number of blocks. [in] dimBlocks dimension of a block. [in] args pointer to kernel arguments. [in] sharedMemBytes Amount of dynamic shared memory to allocate for this kernel. HIP-Clang compiler provides support for extern shared declarations. [in] stream Stream where the kernel should be dispatched. May be 0, in which case the default stream is used with associated synchronization rules. [in] startEvent If non-null, specified event will be updated to track the start time of the kernel launch. The event must be created before calling this API. [in] stopEvent If non-null, specified event will be updated to track the stop time of the kernel launch. The event must be created before calling this API. [in] flags The value of hipExtAnyOrderLaunch, signifies if kernel can be launched in any order.
◆ hipLaunchByPtr()
hipError_t hipLaunchByPtr | ( | const void * | func | ) |
Launch a kernel.
- Parameters
-
[in] func Kernel to launch.
◆ hipLaunchHostFunc()
hipError_t hipLaunchHostFunc | ( | hipStream_t | stream, |
hipHostFn_t | fn, | ||
void * | userData | ||
) |
Enqueues a host function call in a stream.
- Parameters
-
[in] stream - The stream to enqueue work in. [in] fn - The function to call once enqueued preceeding operations are complete. [in] userData - User-specified data to be passed to the function.
The host function to call in this API will be executed after the preceding operations in the stream are complete. The function is a blocking operation that blocks operations in the stream that follow it, until the function is returned. Event synchronization and internal callback functions make sure enqueued operations will execute in order, in the stream.
The host function must not make any HIP API calls. The host function is non-reentrant. It must not perform sychronization with any operation that may depend on other processing execution but is not enqueued to run earlier in the stream.
Host functions that are enqueued respectively in different non-blocking streams can run concurrently.
- Warning
- This API is marked as beta, meaning, while this is feature complete, it is still open to changes and may have outstanding issues.
◆ hipLaunchKernel()
hipError_t hipLaunchKernel | ( | const void * | function_address, |
dim3 | numBlocks, | ||
dim3 | dimBlocks, | ||
void ** | args, | ||
size_t | sharedMemBytes, | ||
hipStream_t | stream | ||
) |
C compliant kernel launch API.
- Parameters
-
[in] function_address - kernel stub function pointer. [in] numBlocks - number of blocks [in] dimBlocks - dimension of a block [in] args - kernel arguments [in] sharedMemBytes - Amount of dynamic shared memory to allocate for this kernel. The HIP-Clang compiler provides support for extern shared declarations. [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, hipErrorInvalidValue
◆ hipSetupArgument()
hipError_t hipSetupArgument | ( | const void * | arg, |
size_t | size, | ||
size_t | offset | ||
) |
Set a kernel argument.
- Parameters
-
[in] arg Pointer the argument in host memory. [in] size Size of the argument. [in] offset Offset of the argument on the argument stack.