there are so many cores

Just another site

Finished adding built-ins

Finished adding 84 OpenCL built-in functions. Now retesting to verify I didn’t break anything. Looks good so far.

Next up is index array support. PeakStream had a clever construction with arrays of loop indexes. This allowed something like a stream closure in the array programming language. The closure could be mapped to either the enqueued global index space or as a loop done in the kernel for each work ID. They didn’t say this in their whitepapers, but having now written a similar JIT compiler, it’s obvious.

I won’t get to random number support this year or before the conference next year. There’s not enough time. I would rather have a stable release and reliability than more features. The RNG API support will still be there. But it won’t be implemented on the GPU. Random numbers will be generated on the CPU host and transferred over the bus to the GPU – inefficient, but at least the functionality is there.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: