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
nlhowell has quit [Ping timeout: 258 seconds]
stikonas has quit [Remote host closed the connection]
buzzmarshall has quit [Remote host closed the connection]
vstehle has quit [Ping timeout: 258 seconds]
<alyssa> bbrezillon: u_blitter-based wallpapering isn't going to cut it for MRT
<alyssa> The blob handles by doing N blits and then masking off only one RT at a time, and the shader has to be MRT aware and do some goofiness accordingly
<alyssa> The upshot is that's a Midgard-only hack
<alyssa> and only needed at all due to errata on t760/t860
<alyssa> (Of course, <T760 there's no mrt at all)
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 272 seconds]
camus1 is now known as kaspter
vstehle has joined #panfrost
nlhowell has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 272 seconds]
camus1 is now known as kaspter
guillaume_g has joined #panfrost
guillaume_g has quit [Client Quit]
camus1 has joined #panfrost
guillaume_g has joined #panfrost
kaspter has quit [Read error: Connection reset by peer]
camus1 is now known as kaspter
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 256 seconds]
camus1 is now known as kaspter
afaerber has quit [Ping timeout: 240 seconds]
nlhowell has quit [Quit: WeeChat 2.8]
nlhowell has joined #panfrost
guillaume_g has quit [Quit: Konversation terminated!]
guillaume_g has joined #panfrost
afaerber has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 256 seconds]
camus1 is now known as kaspter
yann|work has joined #panfrost
kaspter has quit [Ping timeout: 258 seconds]
camus1 has joined #panfrost
camus1 is now known as kaspter
raster has joined #panfrost
nlhowell has quit [Ping timeout: 272 seconds]
nlhowell has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 246 seconds]
camus1 is now known as kaspter
kaspter has quit [Ping timeout: 264 seconds]
kaspter has joined #panfrost
stikonas has joined #panfrost
gcl has quit [Ping timeout: 256 seconds]
gcl_ has joined #panfrost
kaspter has quit [Ping timeout: 260 seconds]
kaspter has joined #panfrost
<icecream95> Hack to force 4x MSAA: https://gitlab.freedesktop.org/snippets/1090/raw
kaspter has quit [Ping timeout: 246 seconds]
kaspter has joined #panfrost
nlhowell has quit [Ping timeout: 256 seconds]
nlhowell has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 246 seconds]
camus1 is now known as kaspter
guillaume_g has quit [Quit: Konversation terminated!]
guillaume_g has joined #panfrost
icecream95 has quit [Quit: leaving]
rcf has joined #panfrost
<alyssa> icecream95: Scary =p
<alyssa> Not sure if that's efficient
<alyssa> For context, mali supports 4x MSAA in two "modes", one writing out all the samples to main memory (4x write bw) and then needing a dedicated resolve (4x read bw + 1x write bw) so 9x higher bw in total
<alyssa> The other keeping all the samples in the tilebuffer and doing the resolve automatically so same b/w as no MSAA if you limit to 4x
<alyssa> MSAA enabled via GLES2 will do the fast one but that's iirc winsys only
<alyssa> MSAA via GLES3 on an FBO will be the slow one, you're supposed to use EXT_multisampled_render_to_texture
<alyssa> anyway, not sure which one that patch will route to in the backend, since I was going for conformance
guillaume_g has quit [Quit: Konversation terminated!]
kaspter has quit [Remote host closed the connection]
guillaume_g has joined #panfrost
kaspter has joined #panfrost
TheMojoMan has joined #panfrost
TheMojoMan has quit [Ping timeout: 272 seconds]
cwabbott_ has joined #panfrost
cwabbott has quit [Ping timeout: 272 seconds]
cwabbott_ is now known as cwabbott
rah has joined #panfrost
<rah> do the Mali GPUs require any binary firmware blobs?
<rah> I couldn't see anything obvious in linux-firmware.git
<robmur01> no, they do not :)
<rah> sweet
<rah> Mali 1 : Adreno 0
<alyssa> rah: \o/
<alyssa> rah: Then again if you go for Adreno, you get robclark
<alyssa> whereas with Mali you get, well, me
<alyssa> ;P
<alyssa> rah: The closest thing to firmware blobs are some tiny programs that get uploaded for things like blits. But we do that all in free s/w in Mes
<alyssa> Mesa
<alyssa> This holds for Utgard through Bifrost, I don't know what Valhall (or beyond) may bring but I'm optimistic
<alyssa> Out of curiousity, what's the interest level (for users and developers) for Vulkan support on T760/T860?
guillaume_g has quit [Quit: Konversation terminated!]
davidlt has joined #panfrost
adjtm has joined #panfrost
cwabbott has quit [Quit: cwabbott]
cwabbott has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
raster has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
raster has joined #panfrost
* alyssa just dipped her toes in the water learning about the vk api if patches are going to start coming
<alyssa> this afternoon's task will be to setup a pan_pool abstraction in common code so we can move stuff like job descriptor uploads into common code :)
Elpaulo has quit [Quit: Elpaulo]
Elpaulo has joined #panfrost
Ntemis has joined #panfrost
davidlt has quit [Ping timeout: 246 seconds]
Ntemis has quit [Read error: Connection reset by peer]
Depau_ has quit [Quit: ZNC 1.8.1 - https://znc.in]
Depau has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
<HdkR> alyssa: Vulkan will become more important as VKD3D and vulkan native apps start running :P
<alyssa> HdkR: Yeah, will that happen on arm64 before bifrost takes over
<HdkR> :)
<alyssa> uh
<alyssa> ?
<alyssa> *
<HdkR> Here's hoping timelines are favourable
<alyssa> HdkR: where's fex at?
<HdkR> We're hooking thunking up atm to get a good speed boost and I'm working on getting Linux signals working correctly
<alyssa> nice!
<alyssa> running any big titles yet? :P
<HdkR> CSGO might work with some hacks to work around VAC, but haven't been trying bigger titles yet
<HdkR> FTL is the target game for me :P
<Lyude> HdkR: what's the state of that like on ppc
<alyssa> ---oh, right, RH :)
<HdkR> Lyude: PPC LE or BE?
<HdkR> :P
<alyssa> Ouch :p
<Lyude> HdkR: the good one
<HdkR> Ah, so LE
<Lyude> (LE)
<Lyude> yeah
<Lyude> hey-aarch64 has BE too!
* HdkR throws X-gene under the rug
<HdkR> Nope, never existed
<HdkR> :D
* alyssa gaslighting intensifies
<HdkR> I never actually looked at POWER past the Gecko, does it support a TSO memory model and real memory atomics at this point?
<Lyude> HdkR: seeing how much work with svm has gone into ppc i'd be amazed if it didn't
<Lyude> no idea about TSO thoughy, but it definitely should at least have memory atomics
<HdkR> It might work with the interpreter, haven't made a JIT though
<HdkR> Would need to poke their latest ISA to see if it would even be worth it
raster has joined #panfrost
icecream95 has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
raster has joined #panfrost
lvrp16 has quit [Ping timeout: 244 seconds]
robher has quit [Ping timeout: 272 seconds]
daniels has quit [Ping timeout: 260 seconds]
ric96 has quit [Ping timeout: 246 seconds]
ezequielg has quit [Ping timeout: 256 seconds]
warpme_ has quit [Ping timeout: 260 seconds]
krh has quit [Ping timeout: 246 seconds]
robclark has quit [Ping timeout: 244 seconds]
narmstrong has quit [Ping timeout: 260 seconds]
steev has quit [Ping timeout: 240 seconds]
jstultz has quit [Ping timeout: 272 seconds]
cwabbott has quit [Quit: cwabbott]
cwabbott has joined #panfrost
yann|work has quit [Ping timeout: 260 seconds]