06.08.2012 Views

Getting Started with CUDA

Getting Started with CUDA

Getting Started with CUDA

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Thread Batching<br />

Kernel launches a grid of thread blocks<br />

Threads <strong>with</strong>in a block cooperate via shared memory<br />

Threads <strong>with</strong>in a block can synchronize<br />

Threads in different blocks cannot cooperate<br />

Allows programs to transparently scale to<br />

different GPUs<br />

Grid<br />

Thread Block 0<br />

Shared Memory<br />

Thread Block 1<br />

Shared Memory<br />

…<br />

Thread Block N-1<br />

Shared Memory<br />

© 2008 NVIDIA Corporation.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!