alyssa changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - Logs https://freenode.irclog.whitequark.org/panfrost - <daniels> avoiding X is a huge feature
raster has quit [Quit: Gettin' stinky!]
stikonas has quit [Remote host closed the connection]
rokquarry has joined #panfrost
buzzmarshall has quit [Ping timeout: 240 seconds]
nerdboy has quit [Ping timeout: 265 seconds]
<anarsoul> alyssa: can you look into making https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3880 generic so lima can also benefic from it?
<anarsoul> btw, one more idea for index draw optimization: https://gitlab.freedesktop.org/lima/mesa/issues/129
<anarsoul> Utgard blob tries to save vertex shader invocations by splitting it into several draws
<anarsoul> I have strong suspicion that Midgard does the same
vstehle has quit [Ping timeout: 272 seconds]
thecycoone has joined #panfrost
<thecycoone> AS_ACTIVE bit stuck
<thecycoone> the mesa fire demo breaks pretty badly :(
rokquarry has quit [Quit: Leaving]
icecream95 has joined #panfrost
icecrea105 has joined #panfrost
robert_ancell has joined #panfrost
icecream95 has quit [Ping timeout: 255 seconds]
icecrea105 is now known as icecream95
rcf has quit [Quit: WeeChat 2.7]
rcf has joined #panfrost
buzzmarshall has joined #panfrost
robert_ancell has quit [Ping timeout: 255 seconds]
davidlt has joined #panfrost
davidlt has quit [Remote host closed the connection]
davidlt has joined #panfrost
buzzmarshall has quit [Remote host closed the connection]
davidlt has quit [Remote host closed the connection]
davidlt has joined #panfrost
vstehle has joined #panfrost
megi has quit [Ping timeout: 268 seconds]
kaspter has quit [Remote host closed the connection]
kaspter has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo has joined #panfrost
robert_ancell has joined #panfrost
robert_ancell has quit [Ping timeout: 260 seconds]
icecream95 has quit [Ping timeout: 240 seconds]
pH5 has joined #panfrost
cowsay_ has joined #panfrost
cowsay has quit [Ping timeout: 268 seconds]
nerdboy has joined #panfrost
Xalius has joined #panfrost
Xalius has quit [Remote host closed the connection]
<tomeu> robmur01: the opp table looks quite different from downstream, indeed: http://paste.debian.net/1131215/
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 255 seconds]
camus1 is now known as kaspter
raster has joined #panfrost
megi has joined #panfrost
<alyssa> bbrezillon: I'm not sure how MS wallpapering is supposed to work.
<alyssa> anarsoul: I have been told that apparently there's a generic one for that in mesa/st already
<alyssa> Right now mostly i don't want to be poking more common code than neccessary..
<alyssa> thecycoone: seems okay to me?
<alyssa> anarsoul: I haven't checked if the midgard blob does that opt too but it seems fairly low prio, all considered
<alyssa> ATM we've been largely CPU bound or fragment bound so
* alyssa isn't excited for opts which trade CPU time for lower VS usage
<thecycoone> interesting alyssa
<thecycoone> supposedly we're on the same hardware. But for me (at least launched from sway) it breaks rendering until the computer is power cycled.
<alyssa> Uh oh.
<thecycoone> anything I should provide to help diagnose, besides those dmesg logs?
<alyssa> Not sure...
<alyssa> which kernel again?
* alyssa has noticed regressions upgrading 5.2 to 5.4 unfortunately
<bbrezillon> alyssa: which version of 5.4?
<bbrezillon> most of them should be gone now
<thecycoone> I upgraded since last time - 5.5.4 now
<thecycoone> (was 5.5.3)
<bbrezillon> thecycoone: can you try with http://code.bulix.org/z46swf-1132902 ?
<bbrezillon> thecycoone: can you also provide the dmesg output?
<alyssa> bbrezillon: 5.4.0-3-arm64
<alyssa> from Debian bullseye
<bbrezillon> alyssa: yes, 5.4.0 is buggy
<alyssa> There's 5.5 in experimental but I'd rather stick to known-buggy than unknown-buggy
<thecycoone> bbrezillon: https://sealbaker.com/dmesg.log
<thecycoone> I'll try the kernel patch at lunch. Using distro supplied kernels so far
<thecycoone> (archlinuxarm's linux-aarch64)
tgall_foo has quit [Ping timeout: 265 seconds]
<robmur01> alyssa: FWIW a "{key,value}[]" layout would probably be mildly more efficient than "{key[], value[]}", since the former should be able to compile to LDP/STP (LDRD/STRD) to minimise load/store overhead
tgall_foo has joined #panfrost
vstehle has quit [Ping timeout: 258 seconds]
vstehle has joined #panfrost
<alyssa> robmur01: I thought it'd be the other way, since you can fit 2x the keys in a given cache line/etc, so for searching this should be better?
Xalius has joined #panfrost
<robmur01> alyssa: I guess it depends mostly on your expected hit rate - if searching (and missing) is overwhelmingly more common than insertion/invalidation, then it may be more beneficial to optimise for
<alyssa> robmur01: Probably this is all micro-optimizing at this point ;)
<alyssa> and all of the above is probably still cheaper than a full blown hash table
<robmur01> but if you get the values for free when reading the keys, hit latency is zero (and the other operations can be effectively twice as efficient)
<alyssa> (It's certainly easier to manage the memory footprint of)
<robmur01> micro-optimising is the best kind of optimising :D
gcl has joined #panfrost
yann has joined #panfrost
<robmur01> maybe I'll give both versions a try and see if there's any observable difference - is there any particular workload that's good to exercise it?
* robmur01 isn't sure of the context of "Aquarium"
<tomeu> robmur01: the webgl fishtank demo
<robmur01> tomeu: thanks!
<alyssa> bbrezillon: I recall you fixed a bug a while back (end of 2019) that manifested as vertices going to the origin instead of their current location (so random triangles stretching to the corner of the screen) in ExtremeTuxRacer and others
<alyssa> Do you remember which patch fixed that / what the issue was?
<bbrezillon> alyssa: not at all, actually I don't remember fixing something like that :)
<tomeu> alyssa dreaming again of bug fixes...
<alyssa> bbrezillon: it was definitely you! :D
<bbrezillon> alyssa: then I fixed something without knowing :)
<robmur01> yeah, I remember seeing that, and it clearing up at some point
<robmur01> but not for any obvious reason
<alyssa> robmur01: It was definitely bbrezillon :p
<robmur01> IIRC I was momentarily tempted to bisect it out of pure curiosity
<robmur01> but... real work :(
shadeslayer has joined #panfrost
<alyssa> Alas.
<tomeu> alyssa: has anything changed in how panwrap works? just preloading it doesn't generate any traces
<alyssa> tomeu: uhhhhh
<alyssa> rprobably lemme see
<tomeu> I see the dvalin knob has gone away
<alyssa> Shoudl still be there?
<alyssa> LD_PRELOAD=/home/rock64/libpanwrap.so DISPLAY=:0 [test]
<alyssa> and that writes to stdou
<tomeu> hrm
<tomeu> wonder if you have the kbase with a totally different ioctl abi
<alyssa> It's possible
<tomeu> I used to switch to it with the dvalin define
<alyssa> That'd be why then
<alyssa> dvalin knob is still there?
* alyssa hadn't realized that there were midgard devices that needed that abi but the knob is there anyway
<alyssa> If your name is Ryan, uncomment "#define dvalin" in include/mali-ioctl.h and panwrap will be built to trace a Dvalin kernel instead.
<alyssa> ^ from the README
<alyssa> Oh but your name isn't Ryan which is why you didn't read that ;)
<alyssa> I don't understand this code but I know I'm responsible for its existence :-(
* alyssa isn't sure where to go about debugging this
<alyssa> tomeu: stderr, not stdout, actually
<alyssa> BREEEEEEEEEHHHHHHHHHHHHHHH
* alyssa breathes fire
<alyssa> tomeu: Well, I fixed my bug.... and a number of other things...
guillaume_g has quit [Remote host closed the connection]
<alyssa> Or not, uh
<alyssa> but we're getting closer
guillaume_g has joined #panfrost
yann has quit [Ping timeout: 255 seconds]
buzzmarshall has joined #panfrost
buzzmarshall has quit [Client Quit]
buzzmarshall has joined #panfrost
<alyssa> Wait. That's not correct.
<alyssa> what is going on
<alyssa> This does seem to be related to varyings / instancing / etc
<alyssa> And I'm having major deja vu..
yann has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
* alyssa gonna call it a break
guillaume_g has quit [Quit: Konversation terminated!]
pH5 has quit [Quit: bye]
pH5 has joined #panfrost
gcl_ has joined #panfrost
gcl has quit [Ping timeout: 240 seconds]
davidlt has quit [Remote host closed the connection]
davidlt has joined #panfrost
yann has quit [Ping timeout: 260 seconds]
<alyssa> Oof, so there's some pretty massive spilling... I sort of wonder if a bug there is the issue
* alyssa is pretty sure there's a spilling related issue... *somewhere*
<alyssa> and possibly there are multiple bugs at once
xdarklight_ is now known as xdarklight
jernej has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
jernej has joined #panfrost
yann has joined #panfrost
buzzmarshall has quit [Quit: Leaving]
nerdboy has quit [Ping timeout: 265 seconds]
stikonas has joined #panfrost
nerdboy has joined #panfrost
buzzmarshall has joined #panfrost
mixfix41 has joined #panfrost
<thecycoone> bbrezillon: do you have that kernel patch as a raw patch instead of a pastebin?
<thecycoone> (or a link to pull it out raw)
<thecycoone> nm, got it
yann has quit [Ping timeout: 258 seconds]
davidlt has quit [Remote host closed the connection]
<thecycoone> Compiling the kernel on kevin takes awhile eh....
<anarsoul> thecycoone: try distcc?
<thecycoone> But it's complicated:p I probably don't need nouveau and everything else the distro includes in their config either
<thecycoone> Also, my desktop is a clarksdale i3 from 2010. Not going to win any races there either. Need on of those 22 second threadrippers.
<thecycoone> * Also, my desktop is a clarksdale i3 from 2010. Not going to win any races there either. Need on of those 22 second threadrippers.
<alyssa> HdkR: ^
<alyssa> Note to self: the analogue to dirty tracking on Mali would be uploading the postfix descriptors at bind time rather than draw time.
pH5 has quit [Quit: -_-]
<HdkR> Hm? Turns out that distcc from a ARM device to a threadripper is actually not great. Threadripper is just waiting for the ARM device to preprocess everything :P
<HdkR> If you can get away with it, cross-compile instead
<HdkR> or ccache also works great :)
<anarsoul> HdkR: ccache + distcc works well for me
<HdkR> I may just be spoiled by how fast 64 hardware threads can compile something
<HdkR> Although cross-compiling the kernel is pretty straightforward compared to some projects