[Bug 204345] New: Unigine Superposition benchmark notoriously unstable since 5.1.18 (and possibly before)
https://bugzilla.kernel.org/show_bug.cgi?id=204345
Bug ID: 204345 Summary: Unigine Superposition benchmark notoriously unstable since 5.1.18 (and possibly before) Product: Drivers Version: 2.5 Kernel Version: 5.1.18 Hardware: x86-64 OS: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: Video(DRI - non Intel) Assignee: drivers_video-dri@kernel-bugs.osdl.org Reporter: edisonalvaringo@bol.com.br Regression: No
At first glance I though it was an Overclocking problem because I oc my hardware just slightly. But no even with no OCing, zero, nada... since 5.1.18 Unigine benchmark crashes. It is really unstable 90% of the times it will crash when it's about to display the final results, on 5.1.20 is even worse and it gets crazy because fps drop really hard. If you are really luck it will finish. I have tried 5.1.18/19 and 20. 5.0.20 doesn't not have this bug. I will try to investigate further to at least know which is the precise point release when this bug was introduced.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #1 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- This is with Linux Mint (Ubuntu 18.04.2 LTS really) I forgot to mention that this is with the liquorix kernel, maybe I should try with vanilla kernel first.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #2 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- With ubuntu ppa vanilla kernel this happens as well.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #3 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- I should be more precise it doesn't crash really it hard locks the pc completely only the reset button will fix the problem.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #4 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- interesting, kernel 5.2.3 doesn't hard lock the PC, but fps drop, abnormally (like with 5.1.20) but when looking at the final results, it's not that bad, about 5% performance hit.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
Joaquin Otsoa (edisonalvaringo@bol.com.br) changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Unigine Superposition |Unigine Superposition hard |benchmark notoriously |lock the PC since 5.1.18 |unstable since 5.1.18 (and |(and possibly before) |possibly before) |
https://bugzilla.kernel.org/show_bug.cgi?id=204345
Joaquin Otsoa (edisonalvaringo@bol.com.br) changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Unigine Superposition hard |Unigine Superposition hard |lock the PC since 5.1.18 |locks the PC since 5.1.18 |(and possibly before) |(and possibly before)
https://bugzilla.kernel.org/show_bug.cgi?id=204345
Joaquin Otsoa (edisonalvaringo@bol.com.br) changed:
What |Removed |Added ---------------------------------------------------------------------------- Kernel Version|5.1.18 |5.1.20
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #5 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- unfortunately the bug was introduced in 5.1.0 so this is going to take thousands of commits, I'll do my best.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #6 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- it doesn't happen with 4.19.61 lts. I think it will be easier to check all the 5.1 rc and check when this was introduced.
https://bugzilla.kernel.org/show_bug.cgi?id=204345
--- Comment #7 from Jani Nikula (jani.nikula@intel.com) --- What hardware?
https://bugzilla.kernel.org/show_bug.cgi?id=204345
Joaquin Otsoa (edisonalvaringo@bol.com.br) changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution|--- |INVALID
--- Comment #8 from Joaquin Otsoa (edisonalvaringo@bol.com.br) --- I'm gonna close this because the regression on the 5.2 are fixed now performs even better than the 5.1 branch. And fixing the 5.1 branch makes no sense because it has reached EOL.
dri-devel@lists.freedesktop.org
-
bugzilla-daemon＠bugzilla.kernel.org