-
Notifications
You must be signed in to change notification settings - Fork 145
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
SLOW Render kernel time #163
Comments
Yeah these numbers are unreasonably slow, about two orders of magnitude slower than expected. On fine rasterization some of it may be #155 but even aside from that these numbers don't look good. I will want to take a closer look at some point (I have a laptop 1060) but am underwater just at the moment, so make no promise when. |
Ok, trying this on the 1060. It's definitely #155, as reducing the blend stack size speeds that up considerably. I'm not sure the results are as good as I'd want them to be, but I haven't dug deeply yet. I'm pretty sure the driver has changed, as this is not consistent with results I was seeing before on the same hardware, but it's possible my memory is hazy. In any case, I know I'll want to fix blend memory (#156) and also will do some tinkering to see whether any of the other recent changes (gradients, blending, etc) might have caused a regression in fine rasterization performance on this device. Other than that, your numbers suggest that the GPU clock is really low. Ideally you can set that to a medium-high but stable value. In any case, it should be possible to use the Performance tab in Task Manager (assuming Windows here?) to see the value. |
Closing this, as the blend stack changes have long since landed, and none of the recent performance measurement work is suggesting that there is an issue with rendering performance. |
I originally thought #161 was the cause of the slowness, but apparently not.
Even after #162 PR the render kernel stage is very slow, unsure why.
GhostScript Tiger, GeForce 1060 3GB:
The text was updated successfully, but these errors were encountered: