<karolherbst>
so uhm... I know we have this agreement with GL and VK conformance, but how does that work with CL?
<bnieuwenhuizen>
karolherbst: would be worth poking the X.Org board. Basically the Khronos board needs to waive adopter fees and then SPI needs to sign the adopter agreement
<karolherbst>
I guess
<bnieuwenhuizen>
in my experience we don't have a pre-signed agreement for all Khronos standards (i.e. we needed to poke for every new Vulkan version up till now) so we probably don't have an easy path to skip things, though last time both sides were pretty supportive (though it can be slow)
<karolherbst>
who is this khornos CL person again?
<karolherbst>
or did that change in the mean time..
flacks has quit [Quit: Quitter]
flacks has joined #dri-devel
ella-0 has quit [Read error: Connection reset by peer]
neonking has quit [Remote host closed the connection]
Company has joined #dri-devel
maxzor has joined #dri-devel
ella-0 has joined #dri-devel
itoral has quit []
pnowack has joined #dri-devel
<javierm>
tzimmermann: reviewing your latest series, could you please remind me what was the difference between drm_WARN_ONCE() and drm_warn_once() ?
<javierm>
looking at the macros it seems to just be two implementations that do the same ?
<tzimmermann>
javierm, i don't see much of a difference either. AFAIK macros with capital letters are deprecated
<tzimmermann>
formatting appears to be different
<javierm>
tzimmermann: that's not what drm_print.h says though...
<javierm>
only the DRM_WARN_ONCE() and similar are deprecated
<javierm>
i.e: the ones that don't take a drm or parent device
<javierm>
tzimmermann: yeah, that's why I said two implementations that do (basically) the same
<javierm>
it's confusing...
<tzimmermann>
oh, well, drm_WARN_ONCE() depends on a condition
<javierm>
tzimmermann: ahh, indeed
iive has joined #dri-devel
apinheiro has quit [Ping timeout: 480 seconds]
icecream95 has quit [Ping timeout: 480 seconds]
pnowack has quit [Quit: pnowack]
<karolherbst>
jekstrand: ehh.. we have to support that test_half stuff :(
Daanct12 has quit [Ping timeout: 480 seconds]
jewins has joined #dri-devel
tzimmermann has quit [Quit: Leaving]
<karolherbst>
airlied: we can't use opencl-h-base.h sadly :(
<karolherbst>
ehh opencl-c-base.h
<karolherbst>
somehow with that we don't get the vstore_half and vload_half definitions, which we actually require
<tlwoerner>
out of curiosity, is the language used to represent GLSL IR scheme? (or some other variant of Lisp?)
sdutt has joined #dri-devel
iive has quit [Ping timeout: 480 seconds]
<FLHerne>
tlwoerner: No, it just happens to use a lot of parens
<tlwoerner>
FLHerne: lol
<FLHerne>
I'd hesitate to call it a 'language' either, it's just a human-readable representation of the IR nodes
laolmtdea^ has quit [Remote host closed the connection]
fxkamd has joined #dri-devel
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
Haaninjo has joined #dri-devel
sdutt has quit []
sdutt has joined #dri-devel
Haaninjo has quit [Ping timeout: 480 seconds]
<jekstrand>
karolherbst: What tests a I fixing?
<karolherbst>
jekstrand: things seem to just work on gen12 though, so that's fine
<karolherbst>
but.. llvm is kind of broken there :( and we might have to focus on fixing all the bugs there
<jekstrand>
karolherbst: test_half works on gen12?
<karolherbst>
yeah
<jekstrand>
That's annoying believable. :-/
* karolherbst
wants to use external crates
JohnStultz[m] has quit []
LaughingMan[m] has quit []
cleverca22[m] has quit []
Wallbraker[m] has quit []
<krh>
karolherbst: I can't believe you made it this far without...
<karolherbst>
krh: ehh.. it's not too bad
<krh>
karolherbst: yeah, the mesa tradition is to reinvent the wheel
* krh
looks at src/util ...
<jekstrand>
Not all of that is re-invention. Did we really re-invent the kernel list? Or did we just re-type it to avoid GPL?
<karolherbst>
don't we have like 5 lists impls?
<jekstrand>
2
<karolherbst>
what? no, I am sure there are more :P
<jekstrand>
We used to have 3 and I deleted one
<jekstrand>
There's 2 that I know of that are used everywhere. I'm sure someone's got a hand-rolled one somewhere.
<karolherbst>
seems like we indeed only have 4 remaining
<karolherbst>
r300 has a handrolled one, and global ones are glsl/list, list.h and simple_list.h
MajorBiscuit has quit [Ping timeout: 480 seconds]
<jekstrand>
simple_list.h still exists?
<karolherbst>
yeah
<jekstrand>
We need to kill it. I thought I already had.
<jekstrand>
Oh, no. It's dead.
<jekstrand>
We just need to delete the header and everything that includes it.
<karolherbst>
ohh.. nothing uses it
<jekstrand>
Yeah. That's an easy cleanup
* jekstrand
types a patch
* karolherbst
only comes up with bad solutions to deal with C pointers :(
hikiko has quit [Ping timeout: 480 seconds]
Duke`` has joined #dri-devel
ybogdano has joined #dri-devel
<dcbaker>
jekstrand: I have pathces to kill simple_list
<karolherbst>
llvm-15 begins to sound like a nightmare
<jekstrand>
karolherbst: \o/
<karolherbst>
let's see if spawning a new thread for even status notifiers increases speed without having to use something complex as thread pools, although I'd really like to use thread pools :D
<bnieuwenhuizen>
karolherbst: sending an email to the board is a good next step yes
<karolherbst>
mhh shit stalls even with a thread :(
<karolherbst>
mhh shit stalls even with a thread :(
frankbinns has quit [Ping timeout: 480 seconds]
heat has quit [Remote host closed the connection]
<karolherbst>
my door rang in the exact same moment I wanted to send my last message :D
<karolherbst>
jekstrand: another idea I had was to have shareable csos
<karolherbst>
so create the compute stuff on the helper context and then once it gets into the worker, there is no compilation or anything funny going on anymore
<karolherbst>
but not sure if we want to make that the default for gallium or if drivers need to opt in
mattrope has joined #dri-devel
LexSfX has quit [Ping timeout: 480 seconds]
LexSfX has joined #dri-devel
<dcbaker>
jekstrand: no more simple list!
<dcbaker>
fortunately my bug was super simple, cut and paste error of trying to free a `tcs` shader in a `tes` pass...
<jekstrand>
dcbaker: \o/
<dcbaker>
next maybe I'll start ripping out the glsl list...
<jekstrand>
that's going to be hard
<jekstrand>
I've wanted to but it's not as easy as one would like
<jekstrand>
util/list.h and glsl/list.h are fundamentally different.
<jekstrand>
util/list.h is circular while exec_list isn't
<jekstrand>
And that means operations like "is this the last node" are harder with util/list
<jekstrand>
And things like nir_instr_next() rely on it
apinheiro has quit [Ping timeout: 480 seconds]
<karolherbst>
to no surprise starting a new thread each time is expensive :D
slattann has quit []
maxzor has quit [Ping timeout: 480 seconds]
cheako has joined #dri-devel
lemonzest has quit [Quit: WeeChat 3.4]
iive has joined #dri-devel
ahajda has quit [Quit: Going offline, see ya! (www.adiirc.com)]
alanc has quit [Remote host closed the connection]
ybogdano has quit [Ping timeout: 480 seconds]
alanc has joined #dri-devel
ybogdano has joined #dri-devel
ahajda has joined #dri-devel
<karolherbst>
cursed use of nested tuples: "item.0.0(item.0.1, item.1, item.0.2)"
ybogdano has quit [Ping timeout: 480 seconds]
<karolherbst>
ahhhhh
<karolherbst>
why is this callback business so terrible
Daanct12 has joined #dri-devel
ngcortes has joined #dri-devel
<karolherbst>
yewah nice.. now I serialize on the locking for the callback queue :)
<karolherbst>
ohh no
<karolherbst>
I am just stupid
<karolherbst>
what does this threadpool help, if the threads are locking while calling the cb :)
ybogdano has joined #dri-devel
neonking has joined #dri-devel
rasterman has quit [Quit: Gettin' stinky!]
* karolherbst
thinks about wrapping simple_mtx_t
mvlad has quit [Remote host closed the connection]
rgallaispou1 has joined #dri-devel
rgallaispou1 has quit [Read error: Connection reset by peer]
xroumegue has quit [Ping timeout: 480 seconds]
rgallaispou has quit [Ping timeout: 480 seconds]
ybogdano has quit [Ping timeout: 480 seconds]
nchery has quit [Read error: Connection reset by peer]