Intel's Gallium3D Driver Gains Optimization For Helping With Java OpenGL Performance

[phoronix]

Phoronix: Intel's Gallium3D Driver Gains Optimization For Helping With Java OpenGL Performance

In our recent Mesa 19.2 benchmarks of Intel's old and new (Gallium3D) OpenGL Linux drivers one of the rare areas where the new "Iris" driver performed behind the classic driver was with "j2dbench" that stresses the Java OpenGL pipeline. At the time it was unknown why the new driver was performing subpar for this Java graphics test, but now at least there's one optimization so far in addressing that shortcoming...

http://www.phoronix.com/scan.php?page=news_item&px=Iris-Optimize-J2dbench-Perf

[FireBurn]

If it's the state tracker, then that'll surely improve things for all the gallium drivers no?

[oibaf]

If it's the state tracker, then that'll surely improve things for all the gallium drivers no?

The fix was for iris only, however: https://cgit.freedesktop.org/mesa/me...b1a58fdfb7f5a0

[Veto]

The fix was for iris only, however: https://cgit.freedesktop.org/mesa/me...b1a58fdfb7f5a0

Argh! This has all the ugly smell of an Intel-only band-aid If instead the root cause was fixed in the state tracker, it would likely benefit all Gallium users.

[OneTimeShot]

Argh! This has all the ugly smell of an Intel-only band-aid If instead the root cause was fixed in the state tracker, it would likely benefit all Gallium users.

Eh? Why would you think that a bug in a file named src/gallium/drivers/iris/iris_state.c affects anything other than Intel Iris?

[Veto]

Eh? Why would you think that a bug in a file named src/gallium/drivers/iris/iris_state.c affects anything other than Intel Iris?

Who says the bug is in that file?? We only know the fix is in that file. Do you see the difference?

Based on the description of the bug, it seems like the state tracker emits a large number of redundant function calls. The code in the fix looks like a simple detection and filtering of these redundant calls. If the real bug is in the shared state tracker emitting these calls in the first place, it could likely affect all drivers.

... However, I am not a Mesa developer, so hopefully they know what they are doing and not just making vendor specific kludges.