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
rasterman has quit [Quit: Gettin' stinky!]
camus1 has quit []
camus has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
Danct12 has quit [Quit: Quitting]
Lyude has joined #panfrost
Danct12 has joined #panfrost
Danct12 has quit [Quit: Quitting]
Danct12 has joined #panfrost
Danct12 has quit [Quit: Quitting]
Danct12 has joined #panfrost
rcf has quit [Quit: WeeChat 3.3-dev]
hanetzer has quit [Quit: WeeChat 3.2]
rcf has joined #panfrost
rasterman has joined #panfrost
warpme_ has joined #panfrost
alpernebbi has joined #panfrost
jambalaya has quit [Remote host closed the connection]
jambalaya has joined #panfrost
anarsoul|2 has joined #panfrost
anarsoul has quit [Remote host closed the connection]
Net147_ has joined #panfrost
Net147 has quit [Read error: Connection reset by peer]
rcf has quit [Quit: WeeChat 3.3-dev]
rcf has joined #panfrost
camus has quit [Remote host closed the connection]
camus has joined #panfrost
HdkR_ has joined #panfrost
HdkR has quit [Ping timeout: 480 seconds]
indy has quit [Ping timeout: 480 seconds]
<robmur01> Neat, looks like the Baikal-M SDK is actively providing panfrost/mesa alongside the Arm driver
<robmur01> surprisingly minimal hacks for T628
mrjay has joined #panfrost
mrjay has quit []
<chewitt> @robmur01 are the sources somewhere public?
<chewitt> I'm interested in those minimal hacks :)
<robmur01> yup, linked straight from the SoC product page - no idea what format the archive is (.run?) but 7-zip understood it
<robmur01> they've bodged the core group coherency issue by restricting job affinity to the half of the GPU where the tiler is
<robmur01> pretty big hammer, but I guess it does make things behave
<chewitt> ahh, not in git then
<chewitt> (which would just be sensible)
indy has joined #panfrost
<robmur01> oh, that's the one indeed
<robmur01> the mesa patch is just allowing the GPU ID through and adding their kmsro entrypoint
<chewitt> I'm interested in getting Kodi to run on an Odroid XU4 .. also T628
<chewitt> Current status is Kodi starts, the home-screen starts to render, then the GPU faults and the board deadlocks
<chewitt> It will be ~2 weeks before I can test whether this makes a difference .. but I'll add to my post-vacation fiddling list
<robmur01> certainly the affinity hack should help - at worst just apply it to all jobs to pretend it's a T624 :)
<robmur01> last time I checked actual rendering was working fairly well, but IIRC there are still a few parts of missing SFBD support
* robmur01 has a growing irrational desire for one of those mini-workstations
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
<alyssa> robmur01: fascinating
<alyssa> "IIRC there are still a few parts of missing SFBD support"
<alyssa> t720 supports gles2 pretty well on panfrost
<alyssa> so it's supposed to work on t6xx...
<alyssa> ```
<alyssa> if (bo->is_heap)
<alyssa> prot |= IOMMU_CACHE;
<alyssa> ```
<alyssa> Uhmm
<robmur01> yeah, I want to sit down at some point and make sense of all that they've fiddled with around memory attributes
<robmur01> since it's certainly an area we have scope for improvement
<alyssa> that heap thing seems like the sort of the thing that'll work great until it really really doesn't.
<robmur01> it has me wondering if they're I/O coherent and also bodging around that without the more recent patches
<robmur01> and I guess if you want to nitpick, s/SFBD/SFBD with a proper tiler/ :P
<alyssa> Pf :p
<robmur01> also pan_emit_sfbd() still hints at raw formats and MSAA being TODO items
<alyssa> sure, but those are boh ES3.0 items
<alyssa> and ES3.0 requires MRT
<alyssa> and MRT on SFBD is something up with which I shall not put
<robmur01> fair enough
jambalaya has quit [Ping timeout: 480 seconds]
jambalaya has joined #panfrost
nlhowell has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
<alyssa> _AU formats used as the data conversion descriptor of opaque_output on Bifrost fails regardless of the round_to_fb_precision flag.
<alyssa> can't tell if this is a hw bug or not.
camus has joined #panfrost
camus1 has quit [Remote host closed the connection]
jernej has quit [Remote host closed the connection]
jernej has joined #panfrost
alpernebbi has quit [Quit: alpernebbi]
HdkR_ is now known as HdkR
hanetzer has joined #panfrost
hanetzer has quit []
rasterman has quit [Quit: Gettin' stinky!]
chewitt has quit [Remote host closed the connection]