Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[question] how does ZCULL work? #11

Open
karolherbst opened this issue Feb 6, 2016 · 2 comments
Open

[question] how does ZCULL work? #11

karolherbst opened this issue Feb 6, 2016 · 2 comments
Labels

Comments

@karolherbst
Copy link
Contributor

I think I managed to work out what I have to push to the gpu to configure zcull itself, but I kind of get the feeling, that either something is missing on the kernel side or some strange interactions with other engines on the gpu.

Anyway I get a "80000001" TRAP on the ZCULL, which means something we don't know yet. It is on the 0x0900 register of the GPC

@sooda
Copy link

sooda commented Feb 15, 2016

I have some nvidia-internal work in progress for a specific detail in zcull that should be polished enough for public review soon. The error you're seeing may or may not be related, but in recent chips there is a mode for saving and restoring zcull data automatically to a buffer on context switch, and if that mode is set but the address for this buffer is not, then you'd see an error.

@karolherbst
Copy link
Contributor Author

thanks for your reply.

I don't even got zcull working at all. I think I figured out how I set the buffer and how to set the parameters for it, but that's pretty much it.

I don't know when to change the address and when to invalidate the zcull context and so on. Generally I am missing some kind of feedback from the zcull engine telling me how much it is doing and what is the issue it has with my input.

So maybe it would already help if I could read out the rejected fragments by zcull and by EarlyZ, so that I will know if it's partly working or not.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

3 participants