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!]
kaspter has quit [Ping timeout: 256 seconds]
kaspter has joined #panfrost
stikonas_ has joined #panfrost
stikonas has quit [Read error: Connection reset by peer]
stikonas_ is now known as stikonas
Ashleee has quit [Ping timeout: 240 seconds]
Prf_Jakob has quit [Ping timeout: 240 seconds]
Prf_Jakob has joined #panfrost
<alyssa> icecream95: Passed: 1736/1740 (99.8%)
<alyssa> That's basically conformant right?
<icecream95> Oh, *of course* the fix had to be this simple: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8773
<icecream95> macc24: ^^
<icecream95> All of that debugging, for a one-line (not including comments) fix…
<alyssa> Been there :(
* alyssa in a MSAA rabbit hole
<alyssa> I have too many branches going I am easily distractable >_>
cowsay has quit [*.net *.split]
afaerber has quit [*.net *.split]
bnieuwenhuizen has quit [*.net *.split]
cowsay has joined #panfrost
afaerber has joined #panfrost
bnieuwenhuizen has joined #panfrost
clementp[m] has quit [Ping timeout: 240 seconds]
wiizzard has quit [Ping timeout: 258 seconds]
icecream95 has quit [Ping timeout: 258 seconds]
l-as has quit [Ping timeout: 268 seconds]
yawniek has quit [Ping timeout: 246 seconds]
yawniek has joined #panfrost
Ke has quit [Ping timeout: 258 seconds]
<alyssa> Our sysval indexing scheme is going to cause serious trouble when SSBOs/textures/images become dynamically indexed in ES32..
<alyssa> Actually, maybe not. Just push everything if there's ambiguity, and we do have dynamic UBO indexing so it's fine
<alyssa> The bigger problem is we prioritize pushing sysvals to FAU... really we need something like brw_nir_analyze_ubo_ranges
icecream95 has joined #panfrost
clementp[m] has joined #panfrost
wiizzard has joined #panfrost
stikonas has quit [Remote host closed the connection]
Ke has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 260 seconds]
camus1 is now known as kaspter
atler has quit [Killed (verne.freenode.net (Nickname regained by services))]
atler has joined #panfrost
kaspter has quit [Read error: Connection reset by peer]
kaspter has joined #panfrost
kaspter has quit [Excess Flood]
kaspter has joined #panfrost
icecream95 has quit [Ping timeout: 244 seconds]
clementp[m] has quit [Ping timeout: 260 seconds]
vstehle has quit [Ping timeout: 260 seconds]
Ke has quit [Ping timeout: 246 seconds]
wiizzard has quit [Ping timeout: 260 seconds]
* alyssa just got a Spanish-language ad for k-pop
<alyssa> I have questions
l-as has joined #panfrost
Ke has joined #panfrost
icecream95 has joined #panfrost
wiizzard has joined #panfrost
clementp[m] has joined #panfrost
urjaman has quit [Read error: Connection reset by peer]
stepri01_ has joined #panfrost
urjaman has joined #panfrost
camus1 has joined #panfrost
kinkinkijkin has quit [*.net *.split]
taowa has quit [*.net *.split]
stepri01 has quit [*.net *.split]
kaspter has quit [Ping timeout: 240 seconds]
camus1 is now known as kaspter
Ke has quit [Ping timeout: 246 seconds]
wiizzard has quit [Ping timeout: 240 seconds]
kinkinkijkin has joined #panfrost
Ke has joined #panfrost
wiizzard has joined #panfrost
taowa has joined #panfrost
davidlt has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 265 seconds]
camus1 is now known as kaspter
vstehle has joined #panfrost
chewitt_ has joined #panfrost
chewitt has quit [Ping timeout: 264 seconds]
macc24 has joined #panfrost
<macc24> oh boy i sure hope that hardkernel didn't absolutely kill G31 with a screen of higher resolution thant 480x320
alpernebbi has joined #panfrost
<HdkR> "5inch 854×480 TFT"
<HdkR> Seems fine
guillaume_g has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 246 seconds]
camus1 is now known as kaspter
<macc24> how costly is multisampling on bifrost?
<HdkR> MSAA up to 4x is usually "cheapish"
<HdkR> on tilers
<macc24> hmmmmmmmmm
<HdkR> You end up paying a cost whenever you need to detile though
<macc24> what if i rendered the scene at half the resolution and multisampled it
raster has joined #panfrost
kaspter has quit [Quit: kaspter]
<bbrezillon> icecream95: do the fault happen when the pan_device object is destroyed?
<bbrezillon> if that's the case, it's certainly a bad a thing, the the user shouldn't see the effect (except for the faults being printed in the kernel logs)
<bbrezillon> *but the user
<macc24> yo i think i got some cursor corruption on master
<macc24> on G72
<bbrezillon> I mean, the HEAP BO stays around for the device/screen lifetime, so adding a BO to the job won't have any effect if you're keeping the GL context around
davidlt has quit [Ping timeout: 240 seconds]
stikonas has joined #panfrost
<icecream95> Is the 0.5 second hang before the job times out not a user-visible effect?
karolherbst has joined #panfrost
<bbrezillon> icecream95: ok, so that's the case you're trying to solve => context being destroyed while the job is still in flight
<bbrezillon> 0.5s hang? I'd expect a mem access fault to cause the job to fail immediately
<bbrezillon> icecream95: did you have "AS_ACTIVE bit stuck" next to these faults, or just the faults and a hang?
davidlt has joined #panfrost
atler has quit [Quit: atler]
atler has joined #panfrost
yann has quit [Ping timeout: 256 seconds]
yann has joined #panfrost
icecream95 has quit [*.net *.split]
clementp[m] has quit [*.net *.split]
l-as has quit [*.net *.split]
dstzd has quit [*.net *.split]
karolherbst has quit [*.net *.split]
stikonas has quit [*.net *.split]
wiizzard has quit [*.net *.split]
yawniek has quit [*.net *.split]
Stenzek has quit [*.net *.split]
icecream95 has joined #panfrost
dstzd has joined #panfrost
clementp[m] has joined #panfrost
l-as has joined #panfrost
karolherbst has joined #panfrost
Stenzek has joined #panfrost
stikonas has joined #panfrost
wiizzard has joined #panfrost
yawniek has joined #panfrost
leper` has quit [Ping timeout: 246 seconds]
Ke has quit [*.net *.split]
clementp[m] has quit [Ping timeout: 244 seconds]
leper` has joined #panfrost
Ke has joined #panfrost
Ke has quit [Max SendQ exceeded]
icecream95 has quit [Ping timeout: 244 seconds]
wiizzard has quit [Ping timeout: 265 seconds]
l-as has quit [Ping timeout: 244 seconds]
Net147 has quit [Quit: Quit]
Net147 has joined #panfrost
wiizzard has joined #panfrost
wiizzard has quit [Ping timeout: 240 seconds]
wiizzard has joined #panfrost
icecream95 has joined #panfrost
clementp[m] has joined #panfrost
Ke has joined #panfrost
l-as has joined #panfrost
uis has joined #panfrost
<uis> Any news about datasheets?
<HdkR> Are we getting datasheets?
archetech has joined #panfrost
<robmur01> sure, here you go: https://www.ti.com/lit/ds/symlink/lm317.pdf
yann has quit [Read error: Connection reset by peer]
yann has joined #panfrost
<robmur01> TI datasheets are always good
<HdkR> Until this: `I just learned that @TXInstruments sent a DMCA takedown to a site archiving datasheets last year.`
<alyssa> Yep
<robmur01> but only because they care and want to make sure you don't get old out-of-date copies, apparently
<robmur01> maintained and updated datasheets is good datasheets
<amonakov> there are more graceful way to achieve that, like leaving a URL for the maintained datasheet repository in released docs
jernej has quit [Remote host closed the connection]
jernej_ has joined #panfrost
jernej_ is now known as jernej
archetech has quit [Quit: Textual IRC Client: www.textualapp.com]
<robmur01> I think you'd have to direct that argument at a vendor who doesn't already litter their PDFs with links back to the relevant product pages :/
nlhowell has joined #panfrost
guillaume_g has quit [Ping timeout: 246 seconds]
nlhowell has quit [Ping timeout: 260 seconds]
macc24 has quit [Ping timeout: 272 seconds]
warpme_ has joined #panfrost
macc24 has joined #panfrost
karolherbst has quit [Quit: duh 🐧]
<macc24> ping?
<thecycoone> png
Ashleee has joined #panfrost
archetech has joined #panfrost
archetech has quit [Client Quit]
archetech has joined #panfrost
atler has quit [Quit: atler]
atler has joined #panfrost
hexdump0815 has joined #panfrost
archetech has quit [Quit: Konversation terminated!]
rando25892 has joined #panfrost
archetech has joined #panfrost
<uis> Mali datasheets
<macc24> uis: define "datasheet"
<uis> Some manuals
<macc24> what exactly are you looking for?
<uis> E.g. with ISA. Registers addresses
<macc24> well then i don't know
<amonakov> why are such inquiries met with mocking responses at all?
<macc24> amonakov: this wasn't mocking
<amonakov> and Robin's "sure, here you go" above?
<macc24> ¯\_(ツ)_/¯
hexdump0815 has quit [Quit: Ping timeout (120 seconds)]
<alyssa> Is everything ok?
rando25892 has quit [Read error: Connection reset by peer]
davidlt has quit [Ping timeout: 246 seconds]
<alyssa> bbrezillon: https://rosenzweig.io/0001-WIP-pan_samples.c.patch Down a rabbit hole 🐰
<macc24> odroid go super has a rotated screen .______.
karolherbst has joined #panfrost
<alyssa> bbrezillon: So what's the story with batch serialization anyhow?
archetech has quit [Quit: Konversation terminated!]
<icecream95> bbrezillon: I don't think I've ever gotten "AS_ACTIVE bit stuck" except due to the GPU voltage being too low
alpernebbi has quit [Quit: alpernebbi]
warpme_ has quit [Quit: Connection closed for inactivity]
<icecream95> bbrezillon: Is this supposed to be possible? https://gitlab.freedesktop.org/-/snippets/1525
raster has quit [Quit: Gettin' stinky!]
dhewg has quit [Read error: Connection reset by peer]
dhewg has joined #panfrost
dhewg has quit [Client Quit]
dhewg has joined #panfrost
dhewg has quit [Read error: Connection reset by peer]
dhewg has joined #panfrost
dhewg has quit [Remote host closed the connection]
<uis> macc24: ... This is... Not what I mean.
<uis> Ok. Reference manual. Like Intel's GPU PRMs.
<macc24> uis: that's why i said that "arm calls this a datsheet" and not "this is a datasheet"
<uis> Like this
<alyssa> uis: I'm not aware of any such release for Mali at this time, sorry.
<alyssa> In practice, almost everything should be in machine-readable formats in Mesa under a permissive license.
<anarsoul> uis: get hired by Collabora and you'll likely get access to the docs :)
<macc24> anarsoul: does that apply to mt8183 too? ;)
<alyssa> anarsoul: s/Collabora/Arm/ would be a lot more effective I assume
<anarsoul> no idea
<anarsoul> alyssa: yeah, maybe
<daniels> uis: what specifically would you like to know?
<uis> I don't know what I want to know, but I want to know what I don't know
<uis> And something about lima/utgard
<macc24> uis: utgard knowledge would be very important, because mali 400 is going to be still alive even after midgard and bifrost are obsolete xD
<alyssa> macc24: :trollestia:
<uis> Is is sarcasm?
<macc24> uis: yes this is a joke, because mali 400 is still being used in new soc-s despite being over 10 years old
<alyssa> Yes.
dhewg has joined #panfrost
<uis> 450 used in Allwinner A64
dhewg has quit [Remote host closed the connection]
<alyssa> bbrezillon: I have a branch passing all of dEQP-GLES31.functional.shaders.sample_variables.* on Bifrost =)
<alyssa> there are... a lot of commits in it labeled WIP or HACK though..
<chrisf> nice
<alyssa> chrisf: Biggest issue is it disables an unrelated optimization on Midgard because it was in the way :p
<macc24> lol
karolherbst has quit [Ping timeout: 265 seconds]