Jump to content

How does GPU manufacturers come up with specific CUDA/Stream processor Numbers for each variant?

Just wondering how it is done, is it hardware limited or the company designs it around a specific core count & manufacture it around it? meaning are these preset long before even the sample GPU is manufactured ? or these are the most efficient specific amount meant to be filled to the brim into the architecture(to the max cores stuffed in there)?

Eg: current 2 flagships

GTX1080 > 2560 cores

RX 480    > 2304 Stream processors

Details separate people.

Link to comment
Share on other sites

Link to post
Share on other sites

I think Raja mentionned on pcper's stream this week that they started with the size of the gpu, probably based around the expected yields they were gonna get from the manufacturing process. Smaller die means better yields, and a mature process also means better yields. 14nm and 16nm are new so it doesn't make economic sense to make the big chips now.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×