nvbio::cuda::syncblocks Struct Reference

Detailed description

implements an inter-CTA synchronization primitive which can be called multiple times from the same grid, or even across multiple kernel launches, as long as all kernel launches have the same size.

ACHTUNG! this primitive is NOT SAFE, and only works if all CTAs are resident!

Definition at line 44 of file syncblocks.h.

#include <syncblocks.h>

Public Methods
NVBIO_FORCEINLINE NVBIO_HOST_DEVICE	syncblocks (int32 *counter=NULL)
NVBIO_FORCEINLINE NVBIO_DEVICE bool	enact (const uint32 max_iter=uint32(-1))

Public Members
int32 *	m_counter

Constructor & Destructor Documentation

NVBIO_FORCEINLINE NVBIO_HOST_DEVICE nvbio::cuda::syncblocks::syncblocks

(

int32 *

counter = NULL)

constructor

Definition at line 41 of file syncblocks_inl.h.

Member Function Documentation

NVBIO_FORCEINLINE NVBIO_DEVICE bool nvbio::cuda::syncblocks::enact

(

const uint32

max_iter = uint32(-1))

enact the syncblocks

Returns

true on successful completion, false otherwise

Definition at line 46 of file syncblocks_inl.h.

Member Data Documentation

int32* nvbio::cuda::syncblocks::m_counter

Definition at line 58 of file syncblocks.h.

---

The documentation for this struct was generated from the following files:

• E:/git/nvbio/nvbio/basic/cuda/syncblocks.h
• E:/git/nvbio/nvbio/basic/cuda/syncblocks_inl.h