stikonas has quit [Remote host closed the connection]
pendingchaos has quit [Ping timeout: 252 seconds]
pendingchaos has joined #panfrost
kaspter has joined #panfrost
kaspter has quit [Excess Flood]
kaspter has joined #panfrost
pendingchaos has quit [Ping timeout: 246 seconds]
pendingchaos has joined #panfrost
robmur01 has quit [Ping timeout: 260 seconds]
pendingchaos has quit [Read error: Connection reset by peer]
pendingchaos has joined #panfrost
pendingchaos has quit [Ping timeout: 252 seconds]
pendingchaos has joined #panfrost
pendingchaos has quit [Client Quit]
<robclark>
alyssa: huh, I call dcd devcoredump (also if you can get useful state out of the gpu, devcoredump is a thing I highly recommend wiring up on the kernel side to make gpu hangs more debuggable)
davidlt has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 252 seconds]
camus is now known as kaspter
vstehle has joined #panfrost
vstehle has quit [Quit: WeeChat 3.0]
vstehle has joined #panfrost
vstehle1 has joined #panfrost
vstehle has quit [Ping timeout: 240 seconds]
vstehle1 has quit [Read error: Connection reset by peer]
vstehle has joined #panfrost
nlhowell has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 265 seconds]
camus is now known as kaspter
catfella_ has joined #panfrost
alpernebbi has joined #panfrost
catfella has quit [Ping timeout: 265 seconds]
<bbrezillon>
alyssa: yes, it's a bit annoying to have to rebase and amend commits to add tags (I wish Marge could automate that somehow), but it's also not the end of the world, and when you're a reviewer it also helps sorting the commits that still need to be reviewed and those that are ready (assuming a multi-round review here)
patrik has joined #panfrost
warpme_ has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 245 seconds]
camus is now known as kaspter
anholt has quit [Ping timeout: 252 seconds]
felipealmeida has quit [Ping timeout: 252 seconds]
felipealmeida has joined #panfrost
kaspter has quit [Read error: Connection reset by peer]
stikonas has joined #panfrost
kaspter has joined #panfrost
raster has joined #panfrost
kaspter has quit [Ping timeout: 252 seconds]
kaspter has joined #panfrost
pendingchaos has joined #panfrost
kaspter has quit [Quit: kaspter]
kaspter has joined #panfrost
pendingchaos has quit [Read error: Connection reset by peer]
pendingchaos has joined #panfrost
catfella_ has quit [Remote host closed the connection]
anholt has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
robmur01 has joined #panfrost
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
chewitt has quit [Quit: Zzz..]
chewitt has joined #panfrost
BenG83 has joined #panfrost
BenG83 has quit [Quit: Leaving]
indy has quit [Ping timeout: 268 seconds]
<alyssa>
bbrezillon: Oh, that's true.
kaspter has quit [*.net *.split]
zkrx has quit [*.net *.split]
thecycoone has quit [*.net *.split]
phh has quit [*.net *.split]
thecycoone has joined #panfrost
kaspter has joined #panfrost
zkrx has joined #panfrost
phh has joined #panfrost
enty has joined #panfrost
leper` has quit [Ping timeout: 240 seconds]
ente has quit [Read error: Connection reset by peer]
leper` has joined #panfrost
zkrx has quit [Ping timeout: 245 seconds]
zkrx has joined #panfrost
tomboy64 has quit [Ping timeout: 246 seconds]
tomboy64 has joined #panfrost
kaspter has quit [Ping timeout: 245 seconds]
kaspter has joined #panfrost
cowsay_ has joined #panfrost
cowsay has quit [Ping timeout: 260 seconds]
<bbrezillon>
daniels: wicast is having problems with panvk + wayland-WSI, maybe you'd be able to help
<bbrezillon>
IIUC, drm_handle_format() and/or dmabuf_handle_modifier() never gets called, but it's not clear why
<wicast>
at wsi_common_wayland.c:435 branch#panvk, my `display->drm.wl_drm` is empty, so the display->formats stay unfilled.
<wicast>
I'm checking how wl_drm is empty.
<wicast>
I get it...
<wicast>
at `registry_handle_global()`, my `interface` is `wl_compositor` not `wl_drm` or `zwp_linux_dmabuf_v1`
<daniels>
wicast: which compositor are you using?
<daniels>
note that registry_handle_global() will be called many times, one per extension
<daniels>
you can also start your client with WAYLAND_DEBUG=client to get a full protocol dump including the advertised extensions
<wicast>
sorry, I just too rush, just like you said it has several extersions. I'm doing a full check.
<daniels>
which compositor you're using is an important question
<wicast>
I use waston, boot with sddm
<daniels>
and weston's log shows it using GL? with the Panfrost driver?
<daniels>
(it logs this to stdout)
<wicast>
'DISPLAY' environment variable not set... skipping surface info
<wicast>
WARNING: panvk is not a conformant vulkan implementation, testing use only.
<wicast>
ERROR at /build/vulkan-tools/src/Vulkan-Tools-1.2.172/vulkaninfo/vulkaninfo.h:24
<wicast>
8:vkGetPhysicalDeviceSurfaceFormatsKHR failed with ERROR_SURFACE_LOST_KHR
<bbrezillon>
alyssa: hehe, it was on my TODO, glad you did it! :-)
<anholt>
blend colors in shader keys? ouch.
<bbrezillon>
the other option I had in mind was to reserve 4 push constants when compiling blend shaders, so we can pass them to the blend shader this way
<bbrezillon>
but that means reserving those push constants without being sure they'll even be used, unless we keep frag shader variants for the blend_shader case
kaspter has quit [Ping timeout: 260 seconds]
kaspter has joined #panfrost
taowa has quit [Read error: Connection reset by peer]
<alyssa>
anholt: It's not as bad as it looks -- that's the key for the blend shader, which is only for the blend CSO
<alyssa>
the same blend shader is used for all the frag shaders
<alyssa>
so that means not nearly as many variants, and more importantly bounded shader size (and a single basic block, no spilling, etc -- basically the cheapest shader compileable)
<alyssa>
Like, it still sucks, but I've never seen blend shader recompiles actually make a blip in profiles.
<anholt>
yeah, but the blend constants are supposed to be effectively uniforms.
<anholt>
so when some app uses them as factors to do some smooth blend transition in an animation, now you have a giant pile of these shaders
<alyssa>
"supposed to" is a funny word in GL..
taowa has joined #panfrost
adjtm has quit [Remote host closed the connection]
<alyssa>
I get it, it's really not ideal. But we've had far bigger fires to fight, and handling this optimally is seriously nontrivial.
adjtm has joined #panfrost
<alyssa>
It's worth noting the DDK works the same way.
<alyssa>
(as Panfrost, I mean0
<alyssa>
admittedly the DDK isn't, er, known for CPU side performance or low memory usage...
<alyssa>
When Panfrost only supported Midgard, we played tricks where we compile an inline vec4(0.0) constant, and just patched the shader binary at draw time if needed.
<alyssa>
That doesn't work on Bifrost because Bifrost's constant packing is ludicrous and so minimally we'd need to do a full repack of the shader, and patch the backend IR directly (with different paths for Midgard vs Bifrost)
<alyssa>
It was easier to just do the key in NIR and forget about it.
<alyssa>
If it makes you feel better, for AGX I'll use actual uniforms :p
WoC has joined #panfrost
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
zkrx has quit [Ping timeout: 252 seconds]
zkrx has joined #panfrost
urjaman has quit [Read error: Connection reset by peer]
urjaman has joined #panfrost
warpme_ has quit [Quit: Connection closed for inactivity]
archetech has quit [Quit: Konversation terminated!]
kaspter has quit [Ping timeout: 265 seconds]
kaspter has joined #panfrost
nlhowell_ has quit [Ping timeout: 246 seconds]
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
stikonas has quit [Remote host closed the connection]