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
ckeepax has quit [Ping timeout: 240 seconds]
ckeepax has joined #panfrost
MastaG has joined #panfrost
<icecream95> alyssa: My incredibly bad elisp to get to a commit from a patch file was only 1 << 5 lines: https://gitlab.freedesktop.org/-/snippets/1518
<alyssa> icecream95: Didn't I already say that yes, I'd include the names from now on? :P
<icecream95> alyssa: I know how short-term your memory is :P
<alyssa> Hnggggggggh
<alyssa> You got me there...
warpme_ has quit [Quit: Connection closed for inactivity]
ente has joined #panfrost
anarsoul has quit [Quit: ZNC 1.7.5 - https://znc.in]
anarsoul has joined #panfrost
enty has joined #panfrost
ente has quit [Read error: Connection reset by peer]
stikonas has quit [Remote host closed the connection]
atler has quit [Killed (kornbluth.freenode.net (Nickname regained by services))]
atler has joined #panfrost
vstehle has quit [Ping timeout: 260 seconds]
kinkinkijkin_ has joined #panfrost
robher_ has joined #panfrost
kinkinkijkin has quit [Ping timeout: 256 seconds]
robher has quit [Ping timeout: 256 seconds]
robher_ is now known as robher
kinkinkijkin_ is now known as kinkinkijkin
atler has quit [Ping timeout: 256 seconds]
atler has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 244 seconds]
camus1 is now known as kaspter
rando25892 has joined #panfrost
felipealmeida has quit [Quit: ZNC 1.8.1 - https://znc.in]
felipealmeida has joined #panfrost
davidlt has joined #panfrost
rando25892 has left #panfrost [#panfrost]
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 260 seconds]
camus1 is now known as kaspter
chewitt has quit [Quit: Adios!]
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 256 seconds]
camus1 is now known as kaspter
vstehle has joined #panfrost
tgall_foo has quit [Ping timeout: 240 seconds]
tgall_foo has joined #panfrost
kaspter has quit [Ping timeout: 256 seconds]
kaspter has joined #panfrost
<bbrezillon> alyssa: thanks for the ufind_msb() patch. Still trying to find a way to get rid of the 64b divs in split_div(), but no success so far :-/
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 256 seconds]
camus1 is now known as kaspter
kaspter has quit [Remote host closed the connection]
kaspter has joined #panfrost
warpme_ has joined #panfrost
karolherbst has joined #panfrost
raster has joined #panfrost
<icecream95> It might be easiest if you just cherry-pick those commits. That way, they'll actually get tested before merge :)
<icecream95> "clause ending up with 8 or fewer quardwords." 8 * 16 = 128, so the max clause size is equal to the prefetch buffer size, which makes sense
karolherbst has quit [Ping timeout: 272 seconds]
karolherbst has joined #panfrost
stikonas has joined #panfrost
robmur01 has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 246 seconds]
camus1 is now known as kaspter
ids1024 has quit [Ping timeout: 240 seconds]
ids1024 has joined #panfrost
nlhowell has quit [Ping timeout: 264 seconds]
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
alpernebbi has joined #panfrost
warpme__ has joined #panfrost
warpme_ has quit [Ping timeout: 256 seconds]
warpme__ is now known as warpme_
<alyssa> ah, yes, magic numbers
raster has quit [Ping timeout: 240 seconds]
<HdkR> That's some dense Panfrost usage!
raster has joined #panfrost
chewitt has joined #panfrost
<macc24> rk3328?
<macc24> mali 400 never dies
<HdkR> In the year 3000 we will still be talking about Utgard being around in fossilized space ships
<macc24> not fossilized, those that are still in service
<HdkR> :)
<alyssa> bbrezillon: Well, talk me through it - what are the 64b divs for?
<bbrezillon> alyssa: implementing panfrost_compute_magic_divisor()
<amonakov> bbrezillon: can you give me a link for context, I've studied something related recently
<bbrezillon> the dividend is a 64b number which happens to also be a power of 2
<bbrezillon> the divisor is a 32b number
rak-zero has joined #panfrost
<amonakov> bbrezillon: thanks, I read that and panfrost.h, but I don't quite get what's split_div and what's the issue you're facing. what am I missing?
<amonakov> s/panfrost/&_job/
<bbrezillon> it's just easier to read in C :)
jolan has quit [Quit: leaving]
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
jolan has joined #panfrost
orkid has quit [Quit: leaving]
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus1 is now known as kaspter
orkid has joined #panfrost
rak-zero has quit [Ping timeout: 240 seconds]
warpme_ has quit [Quit: Connection closed for inactivity]
guillaume_g has quit [Quit: Konversation terminated!]
robmur01 has quit [Remote host closed the connection]
robmur01 has joined #panfrost
davidlt has quit [Ping timeout: 256 seconds]
alpernebbi has quit [Quit: alpernebbi]
raster has quit [Quit: Gettin' stinky!]
<alyssa> bbrezillon: Starting to agree the 64b div is required after all, this is a shame
<alyssa> Playing with the algebra now but this is..awkward
<alyssa> I do have an idea, though
karolherbst has quit [Ping timeout: 260 seconds]
macc24 has quit [Ping timeout: 240 seconds]
leper` has quit [Quit: .]
leper` has joined #panfrost
leper` has quit [Max SendQ exceeded]
leper` has joined #panfrost
<cwabbott> iirc i ran an indirectdraw test with the blob once and the shader was absolutely massive
<cwabbott> the compute shader to generate the job descriptor, that is
<cwabbott> don't think there's a way around it
lvrp16 has quit [Ping timeout: 246 seconds]
austriancoder_ has joined #panfrost
austriancoder has quit [Ping timeout: 260 seconds]
austriancoder_ is now known as austriancoder
lvrp16 has joined #panfrost
<alyssa> cwabbott: Yeah, doesn't mean we need to be sloppy about it on top of it.
<alyssa> bbrezillon: Ignore my comment about 64b div though, it's going to suck regardless, the good news is it doesn't really matter.
raster has joined #panfrost
<alyssa> cwabbott: Run a test with indirect draw + transform feedback + geometry shader + tessellation, if you want to see the DDK cry
<alyssa> and make sure you have some blend shaders in there for good measure