ChanServ changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - Logs https://oftc.irclog.whitequark.org/panfrost - <macc24> i have been here before it was popular
<alyssa> I wonder if the DDK's Valhall scheduler tries to "balance" the pipes.
<alyssa> Certain instructions have equivalents on other pipes. Rewriting to balance the utilization of each functional unit improves throughput *in theory*.
<alyssa> I have no idea if that works in practice.
<alyssa> Or if it matters in practice ... there's not many nonorthogonal instructions on Valhall to make that useful.
<alyssa> (Compare Bifrost which has tons, and schedule quality depends on doing this trick)
<icecream95> alyssa: bi_clper_xor is broken on G31...
<icecream95> So I guess you don't have to do so much G31 bugfixing yourself now
<alyssa> icecream95: It's not broken, it doesn't exist.
<alyssa> Found that one out the hard way.
<icecream95> No, I mean the C function bi_clper_xor
<alyssa> Oh, ffs
<alyssa> fricking
<alyssa> fuck
<alyssa> patch incoming
<icecream95> Okay sure I'll let you increase your patch count then
<icecream95> Possibly Closes: #6550
<alyssa> look zmike is ahead of me :V
<icecream95> alyssa: But mind your language. There could be minors on the channel, they might not know what those words mean
<alyssa> "fricking"?
<alyssa> It's a typo, you know how many typos i make
<alyssa> Meant to say "fracking"
<alyssa> It's a controversial way of getting oil
<alyssa> :p
<icecream95> No, find first set. *shudders*
<alyssa> Uh
<alyssa> clz
<alyssa> pardon my gcc
<icecream95> *gasp*
amazingfate_ has joined #panfrost
<alyssa> icecream95: By the way, "CLPER_V6" is named like that because, at the time, we saw it with G72 but not G52 and assumed it was an arch issue
<alyssa> Then I tried running deqp on g31...
<alyssa> should maybe rename to CLPER_OLD?
amazingfate has joined #panfrost
<icecream95> CLPER_TOENAILS?
<alyssa> :thinking_face:
camus1 has quit [Ping timeout: 480 seconds]
amazingfate_ has quit [Ping timeout: 480 seconds]
* icecream95 gets back to reviewing patches which have landed in main weeks ago
<icecream95> I wonder... is there any reason that we can't use AFBC for at least DXTn texture formats?
<icecream95> It's possible that doing that would be able to get at least *some* compression, but probably the GPU wouldn't be able to handle it, so...
<icecream95> (That is, AFBC-compressing the actual block-compressed data)
<icecream95> But panfrost_afbc_format should probably be expanded to handle more formats, like R8_UNORM
<icecream95> I think sRGB should also be added to workaround crashes with Gallium HUD
camus has joined #panfrost
amazingfate_ has joined #panfrost
amazingfate has quit [Ping timeout: 480 seconds]
<cphealy> icecream95: with AFBC, the RAM requirements are the worst case which is no compression + metadata while texture compression formats such as DXTn are lossy and come with the benefit of smaller RAM requirement. What would the benefit be of using AFBC for DXTn textures?
<icecream95> cphealy: I was thinking of doing AFBC on the already-compressed DXTn data, but the GPU probably doesn't support that, and it wouldn't work that well anyway
<icecream95> And my patches to reduce RAM requirements for AFBC have been working pretty well for a while now, I just need to find the time to upstream them...
cbueno has joined #panfrost
camus has quit []
camus has joined #panfrost
<cphealy> icecream95: I'm not aware of your patches to reduce RAM requirements for AFBC, can you point to the patches?
<anarsoul> alyssa: ping
<anarsoul> alyssa: so I tried dEQP-GLES2.functional.multisample.* on panfrost and it passes just fine on surfaceless
<anarsoul> alyssa: does it do any tricks with msaa like always doing in-place resolve?
<anarsoul> trace says "Writeback MSAA: Layered"
Daanct12 has joined #panfrost
guillaume_g has joined #panfrost
jekstrand has quit [Ping timeout: 480 seconds]
jekstrand has joined #panfrost
Daanct12 has quit [Remote host closed the connection]
Daanct12 has joined #panfrost
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
<icecream95> Hmm... so either pandecode_dump_mappings is completely broken, or AFBC has a mode for compressing gradients which I've never heard of
<icecream95> My guess is that it is the former
<icecream95> sigh, we really need better tools for exploring dumps
<icecream95> Heavily using SQLite, of course
<HdkR> Gotta eat those 3ms/commit times on every sqlite entry addition :P
rasterman has joined #panfrost
Daanct12 has quit [Quit: Leaving]
cbueno has quit []
nlhowell has quit [Ping timeout: 480 seconds]
nlhowell has joined #panfrost
indy has quit [Max SendQ exceeded]
icecream95 has quit [Ping timeout: 480 seconds]
indy has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
indy has quit [Remote host closed the connection]
indy has joined #panfrost
indy has quit [Max SendQ exceeded]
indy has joined #panfrost
nlhowell has joined #panfrost
indy has quit [Remote host closed the connection]
indy has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
cbueno has joined #panfrost
hyrc has joined #panfrost
cbueno has quit [Remote host closed the connection]
guillaume_g has quit []
camus has joined #panfrost
camus1 has quit [Read error: Connection reset by peer]
atler is now known as Guest1090
atler has joined #panfrost
Guest1090 has quit [Ping timeout: 480 seconds]
q4a has quit []
amazingfate_ has quit []
bbrezillon has quit [Ping timeout: 480 seconds]
bbrezillon has joined #panfrost
digetx has quit [Ping timeout: 480 seconds]
digetx has joined #panfrost
erle has quit [Ping timeout: 480 seconds]
erle has joined #panfrost
MajorBiscuit has joined #panfrost
MajorBiscuit has quit [Quit: WeeChat 3.5]
<cphealy> alyssa: Some week or two back you posted this showing perf counter data: https://rosenzweig.io/mh.txt Is there any public code for this that I can experiment with?
<alyssa> all of my code is public because i can't be trusted not to dd if=/dev/urandom of=my rootfs
<alyssa> note the license
<cphealy> I must be missing the license
<cphealy> Or is that the point? ;-)
<cphealy> alyssa: Thank you!
<cphealy> this looks very useful
<FLHerne> I guess if there's no specific license, it's covered by the default MIT one in docs/license.rst?
rasterman has quit [Quit: Gettin' stinky!]
erle has quit []
hyrc has left #panfrost [#panfrost]