HIP: Heterogenous-computing Interface for Portability
Functions
features

Functions

HIP_PUBLIC_API hipError_t hipHccGetAccelerator (int deviceId, hc::accelerator *acc)
 Return hc::accelerator associated with the specified deviceId. More...
 
HIP_PUBLIC_API hipError_t hipHccGetAcceleratorView (hipStream_t stream, hc::accelerator_view **av)
 Return hc::accelerator_view associated with the specified stream. More...
 
HIP_PUBLIC_API hipError_t hipExtModuleLaunchKernel (hipFunction_t f, uint32_t globalWorkSizeX, uint32_t globalWorkSizeY, uint32_t globalWorkSizeZ, uint32_t localWorkSizeX, uint32_t localWorkSizeY, uint32_t localWorkSizeZ, size_t sharedMemBytes, hipStream_t hStream, void **kernelParams, void **extra, hipEvent_t startEvent=nullptr, hipEvent_t stopEvent=nullptr, uint32_t flags=0)
 launches kernel f with launch parameters and shared memory on stream with arguments passed to kernelparams or extra More...
 
HIP_PUBLIC_API hipError_t hipHccModuleLaunchKernel (hipFunction_t f, uint32_t globalWorkSizeX, uint32_t globalWorkSizeY, uint32_t globalWorkSizeZ, uint32_t localWorkSizeX, uint32_t localWorkSizeY, uint32_t localWorkSizeZ, size_t sharedMemBytes, hipStream_t hStream, void **kernelParams, void **extra, hipEvent_t startEvent=nullptr, hipEvent_t stopEvent=nullptr) __attribute__((deprecated("use hipExtModuleLaunchKernel instead")))
 

Detailed Description



Warning
These APIs provide access to special features of HCC compiler and are not available through the CUDA path.

Function Documentation

◆ hipExtModuleLaunchKernel()

HIP_PUBLIC_API hipError_t hipExtModuleLaunchKernel ( hipFunction_t  f,
uint32_t  globalWorkSizeX,
uint32_t  globalWorkSizeY,
uint32_t  globalWorkSizeZ,
uint32_t  localWorkSizeX,
uint32_t  localWorkSizeY,
uint32_t  localWorkSizeZ,
size_t  sharedMemBytes,
hipStream_t  hStream,
void **  kernelParams,
void **  extra,
hipEvent_t  startEvent = nullptr,
hipEvent_t  stopEvent = nullptr,
uint32_t  flags = 0 
)

launches kernel f with launch parameters and shared memory on stream with arguments passed to kernelparams or extra

Parameters
[in[f Kernel to launch.
[in]gridDimXX grid dimension specified in work-items
[in]gridDimYY grid dimension specified in work-items
[in]gridDimZZ grid dimension specified in work-items
[in]blockDimXX block dimensions specified in work-items
[in]blockDimYY grid dimension specified in work-items
[in]blockDimZZ grid dimension specified in work-items
[in]sharedMemBytesAmount of dynamic shared memory to allocate for this kernel. The kernel can access this with HIP_DYNAMIC_SHARED.
[in]streamStream where the kernel should be dispatched. May be 0, in which case th default stream is used with associated synchronization rules.
[in]kernelParams
[in]extraPointer to kernel arguments. These are passed directly to the kernel and must be in the memory layout and alignment expected by the kernel.
[in]startEventIf 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]stopEventIf 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.
Returns
hipSuccess, hipInvalidDevice, hipErrorNotInitialized, hipErrorInvalidValue
Warning
kernellParams argument is not yet implemented in HIP. Please use extra instead. Please refer to hip_porting_driver_api.md for sample usage. HIP/ROCm actually updates the start event when the associated kernel completes.

◆ hipHccGetAccelerator()

HIP_PUBLIC_API hipError_t hipHccGetAccelerator ( int  deviceId,
hc::accelerator *  acc 
)

Return hc::accelerator associated with the specified deviceId.

Returns
hipSuccess, hipErrorInvalidDevice

◆ hipHccGetAcceleratorView()

HIP_PUBLIC_API hipError_t hipHccGetAcceleratorView ( hipStream_t  stream,
hc::accelerator_view **  av 
)

Return hc::accelerator_view associated with the specified stream.

If stream is 0, the accelerator_view for the default stream is returned.

Returns
hipSuccess