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
cphealy has joined #panfrost
WoC has quit [Remote host closed the connection]
WoC has joined #panfrost
Green has quit [Quit: ...]
Green has joined #panfrost
WoC has quit [Remote host closed the connection]
WoC has joined #panfrost
Green has quit [Quit: ...]
Green has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
sphalerite has quit [Ping timeout: 258 seconds]
Green has quit [Quit: ...]
Green has joined #panfrost
kaspter has joined #panfrost
vstehle has quit [Ping timeout: 246 seconds]
WoC has quit [Ping timeout: 260 seconds]
stikonas has quit [Ping timeout: 260 seconds]
Green has quit [Quit: ...]
Green has joined #panfrost
<macc24> oh no...
<macc24> devfreq broke
<macc24> again
Green has quit [Client Quit]
Green has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 260 seconds]
camus is now known as kaspter
<urjaman> hm?
archetech has joined #panfrost
Green has quit [Quit: ...]
Green has joined #panfrost
warpme_ has quit [Quit: Connection closed for inactivity]
Green has quit [Quit: ...]
Green has joined #panfrost
Green has quit [Quit: ...]
Green has joined #panfrost
felipealmeida has quit [Ping timeout: 268 seconds]
felipealmeida has joined #panfrost
Green has quit [Quit: ...]
Green has joined #panfrost
archetech has quit [Quit: Textual IRC Client: www.textualapp.com]
vstehle has joined #panfrost
davidlt has joined #panfrost
kaspter has quit [Ping timeout: 256 seconds]
kaspter has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus has joined #panfrost
camus is now known as kaspter
yawniek has quit [Ping timeout: 272 seconds]
kaspter has quit [Ping timeout: 264 seconds]
kaspter has joined #panfrost
raster has joined #panfrost
yawniek has joined #panfrost
icecream95 has quit [Ping timeout: 240 seconds]
yann has joined #panfrost
yawniek has quit [Quit: WeeChat 2.9]
yann has quit [Ping timeout: 260 seconds]
kaspter has quit [Remote host closed the connection]
kaspter has joined #panfrost
sphalerite has joined #panfrost
yawniek has joined #panfrost
chewitt has quit [Ping timeout: 256 seconds]
camus has joined #panfrost
kaspter has quit [Ping timeout: 272 seconds]
camus is now known as kaspter
chewitt has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
Green has quit [Read error: Connection reset by peer]
stikonas has joined #panfrost
raster has joined #panfrost
warpme_ has joined #panfrost
gcl has quit [Ping timeout: 240 seconds]
kaspter has quit [Quit: kaspter]
gcl has joined #panfrost
gcl_ has joined #panfrost
gcl has quit [Ping timeout: 246 seconds]
robmur01 has joined #panfrost
* robmur01 chuckles at HdkR "I completely expect Nvidia to kill the ARM display IP"
<robmur01> Arm already killed Arm's display IP
<robmur01> or from another perspective, Arm's customers did. By not buying it :P
<macc24> why ._.
<macc24> oh
<HdkR> The truth is crueler than my imagination
<macc24> at least they triedf
<cphealy> NXP uses the ARM display IP in the LS1028A (Oddly, it's one of the few IP cores in the SoC design that has zero detail in the ~7000 page reference manual...)
<alyssa> Fascinating
<cphealy> The Vivante GPU in the LS1028A has at least 20 pages dedicated to it. The ARM display IP = "This information is under NDA. Contact local NXP sales representative for more details." ;-)
<macc24> now that i used linux on kevin for some time
<macc24> i never realized that the bar was so low for usability xD
yann has joined #panfrost
<robmur01> cphealy: yeah, there doesn't appear to be a Non-Confidential version of the DP500 TRM (or indeed any other DPU TRMs) - TBH it's probably mostly that there was just never any justification for the effort of releasing one
<robmur01> from an end-user (i.e. not SoC integrator) perspective there's probably not much in there that you couldn't figure out from the Linux driver anyway; I don't think they're massively complex beasts
yann has quit [Ping timeout: 260 seconds]
<alyssa> I have to write unit tests.
<macc24> when supertuxkart crashes under panfrost but u still want to play it: https://i.imgur.com/BYCWwGH.png (that's under llvmpipe)
warpme_ has quit [Quit: Connection closed for inactivity]
<kinkinkijkin> pcsxr (not rearmed) works fine full-speed on duet with panfrost
<kinkinkijkin> despite being interpreter only
<macc24> wow
<macc24> i wonder if ps2 emulation would work
<kinkinkijkin> probably yes, but also probably very slow
<robmur01> but then you could use the PS2 emulator to run Linux :D
archetech has joined #panfrost
<macc24> kinkinkijkin: i don'
<macc24> i don't think that second computer is needed to recover from bad cadmium update
<alyssa> DEVLEOPERS
<macc24> alyssa: ???
WoC has joined #panfrost
<alyssa> Developers developers developers
* robmur01 nervously glances around for throwable chairs...
<macc24> hypothesis: duet has no usb on pogo pins on bottom
<alyssa> 🙈
<macc24> kinkinkijkin: what's the output of 'lsusb -v | grep iConfiguration' on your duet?
yann has joined #panfrost
<alyssa> nothing like bifrost's famous 78-bit instructions :<
<macc24> cursed
yann has quit [Ping timeout: 260 seconds]
<WoC> anyone who knows if there is a OpenCL that can be used in combination with panfrost on mali-t860 ?
<macc24> hypothesis 2: duet keyboard is i2c device, and pads on bottom of board are connected to pins on bottom of device
<alyssa> macc24: means we have to implement int128 operations to handle packing :V
<macc24> alyssa: oh my god
<alyssa> macc24: oh! and instead of having things be 16-bit aligned....
<alyssa> everything is 15-bit aligned
<alyssa> yes. really
<macc24> this is beyond cursed
<alyssa> (* except for an 8-bit shift at the start because of course)
<alyssa> 128-bit quadword = 8-bit tag + 7 * 15-bit subwords
<alyssa> er, 8
<alyssa> so a 78-bit instructions fits snugly in 5 15-subwords
<alyssa> with 3-bits thrown in some other field where they can fit it
<macc24> i'm happy that i don't have to deal with this ;D
<alyssa> >:
robmur01 has quit [Quit: Leaving]
* alyssa is down a debug hellhole
<macc24> i sure do love solving screens of merge conflicts
<alyssa> me_irl
yann has joined #panfrost
yann has quit [Ping timeout: 256 seconds]
<alyssa> So I burned enough time trying to get *good* test cases and...
<alyssa> TL;DR whipped up a Python script to hackily parse disassemblies from clauses in the blob and generate unit tests :p
<alyssa> Lyude: Guess who's working on the packing code ^^
<alyssa> oh gosh
<alyssa> Code invoked subtle undefined behaviour causing "impossible test failures"
<alyssa> Since gcc's constant folding differed from arm64's implementation. ughh
tomboy64 has quit [Ping timeout: 240 seconds]
davidlt has quit [Ping timeout: 240 seconds]
warpme_ has joined #panfrost
tomboy64 has joined #panfrost
WoC has quit [Remote host closed the connection]
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
WoC has joined #panfrost
archetech has quit [Quit: Konversation terminated!]
<Lyude> alyssa: cool!
<alyssa> Lyude: I have it _almost_ working
<alyssa> A minute ago i'd've said I had it
<macc24> almost?
<alyssa> added another test case, and we're failing this one
<macc24> i may have working 5.11 kernel for duet
<alyssa> or not, looks like the test case is wrong
<alyssa> so I do have it working
<alyssa> 👍
<Lyude> cool :)
dstzd has quit [Read error: Connection reset by peer]
dstzd_ has joined #panfrost
dstzd_ is now known as dstzd
raster has quit [Quit: Gettin' stinky!]
<alyssa> still need to deal with constants
<alyssa> including a genuinely -awful -- mchanism that never got r/e'd
lvrp16 has quit [Ping timeout: 260 seconds]
lvrp16 has joined #panfrost
icecream95 has joined #panfrost
<icecream95> So *of course* the faults don't happen when I set PAN_MESA_DEBUG=trace...
yann has joined #panfrost
<alyssa> >_<
<alyssa> anything I can help with?
* alyssa has exhausted her ability to do scheduling stuff today
yann has quit [Ping timeout: 256 seconds]
icecream95 has quit [Ping timeout: 264 seconds]
icecream95 has joined #panfrost
<icecream95> alyssa: Shouldn't the lane0 modifier of LOAD.i16 be for the instruction destination, not src[0] as it currently is?
<alyssa> 👀
<alyssa> Yes, it should... that's a really annoying edge case :|
popolon has joined #panfrost
<alyssa> Same goes for LOAD.i8...
<alyssa> (and LOAD.i32 but it's not clear how widening 32 to 64 makes any sense)
<alyssa> Bifrost is just full of surprises.
<macc24> icecream95: got usb-c ccd thing?
<icecream95> macc24: no
* macc24 got pinout of duet's keyboard
<macc24> it's usb after all
<icecream95> macc24: What's the fifth pin for, then?
<macc24> PINT
<icecream95> Ahh, I always wondered why the keyboard sometimes acted drunk :P
<macc24> my guess is "id":
<macc24> from left to right: vbus, d+, gnd, d-, d- and pint
<macc24> it still boots after reassembling
* macc24 lets out sigh of relief
<macc24> and it looks like there is no mt8183 uart on testpads