Shader debugging on a Maxwell GPU
Wow, it is still not implemented..... "Good" job.
Wow, it is still not implemented..... "Good" job.

#16
Posted 10/12/2016 02:28 AM   
Is this still not implemented? Can someone name a card currently available for purchase that does support Shader Debugging?
Is this still not implemented? Can someone name a card currently available for purchase that does support Shader Debugging?

#17
Posted 04/04/2017 11:57 PM   
Any kepler GPU will be able to do some debugging (actually very slow, even using minimal experience debugging). Fermi has been removed from the supported hardware, which is weird, because we want to do profilling mostly on low-spec minimum requirement for our games... Don't forget that time scrubber is in empty pipeline cost so it won't show the real time of the frame. Frame-Timing has never worked on our end, since ever. It shows huge gaps between draws and the resulting timing is totally over scope (over 60~100 ms odd) Hopefully, the next version will not tackle only new feature but will revised the way that this nice tool can serves the need of pure performance profiling + addition off nice drill-down GPU such as neat representation of L2 cache miss, overdraws, parallelism and contention, bandwidth, filtering and so forth.
Any kepler GPU will be able to do some debugging (actually very slow, even using minimal experience debugging). Fermi has been removed from the supported hardware, which is weird, because we want to do profilling mostly on low-spec minimum requirement for our games...

Don't forget that time scrubber is in empty pipeline cost so it won't show the real time of the frame. Frame-Timing has never worked on our end, since ever. It shows huge gaps between draws and the resulting timing is totally over scope (over 60~100 ms odd)

Hopefully, the next version will not tackle only new feature but will revised the way that this nice tool can serves the need of pure performance profiling + addition off nice drill-down GPU such as neat representation of L2 cache miss, overdraws, parallelism and contention, bandwidth, filtering and so forth.

#18
Posted 04/10/2017 11:51 AM   
Scroll To Top

Add Reply