austriancoder changed the topic of #etnaviv to: #etnaviv - the home of the reverse-engineered Vivante GPU driver - Logs https://freenode.irclog.whitequark.org/etnaviv
pcercuei has quit [Quit: dodo]
JohnnyonFlame has quit [Ping timeout: 240 seconds]
tchebb_ has joined #etnaviv
serg-z has quit [*.net *.split]
tchebb has quit [*.net *.split]
tchebb_ is now known as tchebb
pcercuei has joined #etnaviv
JohnnyonFlame has joined #etnaviv
JohnnyonFlame has quit [Ping timeout: 240 seconds]
JohnnyonFlame has joined #etnaviv
paulk-leonov has quit [Ping timeout: 260 seconds]
paulk-leonov has joined #etnaviv
<mntmn> i wonder how i could go about figuring out what draw calls or shader stuff firefox does that causes etnaviv bugs (some initial draws not happening, textures scribbled with junk)
<austriancoder> mntmn: record the API calls via renderdoc or apitrace
<mntmn> problem goes away when running through renderdoc... wat
paulk-leonov has quit [Ping timeout: 276 seconds]
<mntmn> austriancoder: when running firefox from renderdoc, the glitches disappear
paulk-leonov has joined #etnaviv
<mntmn> ah i think it falls back to llvmpipe or so
paulk-leonov has quit [Excess Flood]
<mntmn> ok, the fix is just not forcing layer acceleration in firefox... it seems to be faster anyway now with FF88
<mntmn> the only downside is non-accelerated scaled surfaces, like video
paulk-leonov has joined #etnaviv
<austriancoder> mntmn: okay.. atm I am only looking at synthetic benchmarks and conformance apps and not at real apps
<mntmn> yeah
<mntmn> gonna try apitrace, as i can't get qrenderdoc to capture anything
<mntmn> austriancoder: i managed to capture with apitrace, and play back, which is super cool. but! the playback is clean. that's really weird
<mntmn> so maybe some cpu/gpu interaction that happens in the real software but not in the trace playback
<austriancoder> maybe
<mntmn> yeah it has to do with the multi process model of firefox
<mntmn> when i set dom.ipc.processCount to 1 (default 8), the glitches basically go away
<mntmn> could there be some issues sharing buffers or contexts between processes or cpu cores?
<austriancoder> maybe some sync problems..
* austriancoder tries to add a piano line to an almost finished song .. soo not programming for this night - sorry
<mntmn> austriancoder: haha, sorry! enjoy composing
<mntmn> at least now i have a workaround so i can surf the web in peace
robclark has quit [Read error: Connection reset by peer]
robclark has joined #etnaviv
daniels has quit [Ping timeout: 258 seconds]
daniels has joined #etnaviv
pcercuei has quit [Quit: dodo]