<karolherbst>
jekstrand: before I start diging into why some generic pointers won't get resolved, any suggestions in terms of nir passes I have to run in order to resolve those successfully?
ybogdano has quit [Ping timeout: 480 seconds]
eukara has quit [Ping timeout: 480 seconds]
jewins1 has joined #dri-devel
jewins has quit [Ping timeout: 480 seconds]
naveenk2 has joined #dri-devel
blue_penquin has quit []
blue_penquin has joined #dri-devel
blue_penquin has quit []
tales_ has joined #dri-devel
Company has quit [Quit: Leaving]
eukara has joined #dri-devel
bluepenquin has joined #dri-devel
LexSfX has joined #dri-devel
tales_ has quit []
ppascher has joined #dri-devel
elongbug__ has quit [Read error: Connection reset by peer]
eukara has quit [Ping timeout: 480 seconds]
shankaru has joined #dri-devel
mclasen has quit [Ping timeout: 480 seconds]
heat has quit [Ping timeout: 480 seconds]
Duke`` has joined #dri-devel
mattrope has quit [Quit: Leaving]
jewins1 has quit [Read error: Connection reset by peer]
jewins has joined #dri-devel
fxkamd has quit []
mhenning has quit [Quit: mhenning]
itoral has joined #dri-devel
jewins has quit [Ping timeout: 480 seconds]
rbrune has joined #dri-devel
Duke`` has quit [Ping timeout: 480 seconds]
sdutt has quit [Read error: Connection reset by peer]
lemonzest has joined #dri-devel
tomeu has joined #dri-devel
mszyprow has joined #dri-devel
ahajda has joined #dri-devel
mvlad has joined #dri-devel
itoral has quit [Remote host closed the connection]
itoral has joined #dri-devel
danvet has joined #dri-devel
aravind has joined #dri-devel
shankaru has quit [Quit: Leaving.]
jkrzyszt_ has joined #dri-devel
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
Guest574 is now known as shadeslayer
MajorBiscuit has joined #dri-devel
lynxeye has joined #dri-devel
tursulin has joined #dri-devel
shankaru has joined #dri-devel
craftyguy_ has joined #dri-devel
ngcortes has quit [Remote host closed the connection]
craftyguy has quit [Ping timeout: 480 seconds]
rasterman has joined #dri-devel
pcercuei has joined #dri-devel
bnieuwenhuizen has quit [Server closed connection]
bnieuwenhuizen has joined #dri-devel
dottedmag has quit [Server closed connection]
dottedmag has joined #dri-devel
bnieuwenhuizen is now known as Guest640
urja has quit [Server closed connection]
urja has joined #dri-devel
BobBeck has joined #dri-devel
Major_Biscuit has joined #dri-devel
Major_Biscuit has quit []
Major_Biscuit has joined #dri-devel
MajorBiscuit has quit [Ping timeout: 480 seconds]
MrCooper has quit [Server closed connection]
MrCooper has joined #dri-devel
xerpi[m] has quit []
ella-0[m] has quit []
HayashiEsme[m] has quit []
dafna2[m] has quit []
<danvet>
shashank_sharma,
mclasen has joined #dri-devel
leah has quit [Server closed connection]
leah has joined #dri-devel
itoral has quit [Remote host closed the connection]
leah is now known as Guest647
itoral has joined #dri-devel
jernej has quit [Server closed connection]
jernej has joined #dri-devel
anarsoul has quit [Ping timeout: 480 seconds]
lplc has quit [Ping timeout: 480 seconds]
znullptr[m] has quit [Server closed connection]
znullptr[m] has joined #dri-devel
q66 has quit [Remote host closed the connection]
anarsoul has joined #dri-devel
lplc has joined #dri-devel
tdsea^ has joined #dri-devel
imre has joined #dri-devel
devilhorns has joined #dri-devel
itoral has quit [Remote host closed the connection]
zigotonus has quit [Read error: Connection reset by peer]
mlankhorst has quit [Server closed connection]
mlankhorst_ has joined #dri-devel
ced117 has quit [Server closed connection]
ced117 has joined #dri-devel
libv has quit [Server closed connection]
libv has joined #dri-devel
emersion has quit [Server closed connection]
emersion has joined #dri-devel
<karolherbst>
jfalempe: cool
<karolherbst>
maybe somebody wants to benchmark llvmpipe and see if it regresses perf or not
angerctl has quit [Server closed connection]
angerctl has joined #dri-devel
kchibisov has quit [Server closed connection]
kchibisov has joined #dri-devel
sdutt has joined #dri-devel
<jfalempe>
I can try to run some benchmarks, but I don't know which one will actually use this code.
<jfalempe>
it depends on color format of the "blend buffer" and framebuffer. (there are fast paths with memcpy if they are equal).
heat has quit [Ping timeout: 480 seconds]
mclasen has quit []
mclasen has joined #dri-devel
Lynne has quit [Server closed connection]
Lynne has joined #dri-devel
gio has quit [Server closed connection]
gio has joined #dri-devel
<karolherbst>
yeah.. maybe some of the llvmpipe folks know
<karolherbst>
jfalempe: or maybe look at the generated assembly for x86 and make a judgement? dunno
oneforall2 has quit [Remote host closed the connection]
oneforall2 has joined #dri-devel
<jfalempe>
hum I don't know much x86 assembly, not sure I will be able to tell if it's significantly slower.
shoragan has quit [Server closed connection]
shoragan has joined #dri-devel
<karolherbst>
yeah.. me neither
sdutt has quit []
sdutt has joined #dri-devel
pinchartl has quit [Server closed connection]
jewins has joined #dri-devel
pinchartl has joined #dri-devel
oneforall2 has quit [Remote host closed the connection]
oneforall2 has joined #dri-devel
dsrt^ has quit [Remote host closed the connection]
oneforall2 has quit [Remote host closed the connection]
rbrune has quit [Ping timeout: 480 seconds]
oneforall2 has joined #dri-devel
oneforall2 has quit [Remote host closed the connection]
oneforall2 has joined #dri-devel
elongbug has joined #dri-devel
Surkow has quit [Server closed connection]
Surkow has joined #dri-devel
aravind has quit [Ping timeout: 480 seconds]
<pinchartl>
Lyude: has there been a known issue with gabe.freedesktop.org and events@lists.x.org ? I've just received 7 e-mails related to the elections in one go, with the oldest dating from Feb the 4th
i-garrison has quit [Read error: Connection reset by peer]
i-garrison has joined #dri-devel
Guest389 has quit [Server closed connection]
dreda has joined #dri-devel
dreda is now known as Guest670
iive has joined #dri-devel
<danvet>
pinchartl, #freedesktop, ping emmersion
<pinchartl>
done
rkanwal has joined #dri-devel
mattrope has joined #dri-devel
ella-0_ has joined #dri-devel
fxkamd has joined #dri-devel
khfeng has quit [Ping timeout: 480 seconds]
ella-0 has quit [Read error: Connection reset by peer]
<jfalempe>
karolherbst, I tested with "glmark2 -b desktop" on s390x with 16bits vnc. I get 26 FPS with and without my patch. So at least performance impacts should be minimal.
<jekstrand>
karolherbst: nir_opt_deref is your friend.
<tjaalton>
jenatali: do you have an idea why things like starting gnome-chess would segfault when running on intel hw and with xorg
<ajax>
o_O
<tjaalton>
not verified by me, but someone who knows packaging and reverted that and it was still ok
<ajax>
that's wild
i-garrison has quit [Read error: Connection reset by peer]
<jenatali>
Whoops, wasn't identified on IRC
<jenatali>
I don't know anything about the tlsdesc portion of that change, I was just interested in the initial exec bits which are needed for correctness on Android
i-garrison has joined #dri-devel
<jenatali>
I'd be interested to see if it was the tlsdesc bits causing the problem
<karolherbst>
jekstrand: I already call that one in like three places :D
<tjaalton>
jenatali: right, this just enables that
<karolherbst>
but yeah... at least good to know that nir_opt_deref is handling those
<jekstrand>
karolherbst: Well, that's the one that restricts modes. Of course, it can only do so if copy-prop has gotten rid of any ALU in the way.
<karolherbst>
jfalempe: ohh, I meant on x86
<karolherbst>
not really caring about perf on s390x, but on other platforms it's important
<karolherbst>
jekstrand: yeah.. I just copied the opt loops from microsoft/clc*, but maybe there is something missing
<karolherbst>
need to figure out why they don't get resolved
<karolherbst>
but... you know.. it's a luxmark 3.1 kernel :(
<karolherbst>
those things are huge
<ajax>
tjaalton: possibly the code in src/mapi/entry_x86-64_tls.h needs to learn about the gnu2 style?
<ajax>
shooting entirely in the dark here
<karolherbst>
I have to implement inline samplers before that anyway
<jekstrand>
karolherbst: Yeah, luxmark is hopeless
mclasen has joined #dri-devel
<karolherbst>
jekstrand: it did run with nouveau at some point though... not correctly, but it did run
<karolherbst>
:D
<karolherbst>
anyway.. should be able to figure it out and make it run
<tjaalton>
beta coming up, need to revert this now and worry more later :)
<ajax>
indeed
<ajax>
or at least hide it behind a meson option that defaults off
<tjaalton>
yeah for upstream that's better, I'll revert in the packaging
<jenatali>
jekstrand: saw the email, seems reasonable
<karolherbst>
jekstrand: I think that's fine
<karolherbst>
I saw some mangling failures, and if that fixes those I am fine with that :D
<jenatali>
karolherbst: I'm pretty sure I got luxmark 3.1 (whatever is current - 1) running correctly. Latest version is hopeless though
<karolherbst>
jenatali: yeah, I tried 3.1
<karolherbst>
but something wasn't working correctly with generics
<karolherbst>
I am on llvm 15 though
<jenatali>
Yeah you have to disable generics on new llvm
<karolherbst>
:(
<jekstrand>
karolherbst: Yup, it'll fix the mangling failures with libclc
<jenatali>
I think Jason's patch fixes that though
<jekstrand>
karolherbst: In particular, test_bruteforce frexp
<karolherbst>
ahh
<karolherbst>
will try it out then
<jenatali>
Btw I'm still not really around. I'll be back from leave in another ~week
<karolherbst>
jenatali: no worries
<karolherbst>
I have big internal things to work on anyway
linkmauve has quit [Server closed connection]
<karolherbst>
just filling breaks from waiting compiles to finish with CL work :D
mszyprow has quit [Ping timeout: 480 seconds]
mripard_ has quit [Server closed connection]
mripard has joined #dri-devel
kts has joined #dri-devel
Duke`` has joined #dri-devel
ybogdano has joined #dri-devel
linkmauve has joined #dri-devel
heat has joined #dri-devel
pjakobsson_ has joined #dri-devel
Major_Biscuit has quit [Ping timeout: 480 seconds]
lemonzest has joined #dri-devel
pjakobsson has quit [Ping timeout: 480 seconds]
oneforall2 has quit [Remote host closed the connection]
<jekstrand>
zmike: I'm looking at this semaphore patch and it makes me terribly tempted to rewrite lavapipe to use common sync....
Major_Biscuit has joined #dri-devel
<jekstrand>
zmike: Especially with the work I just did to let queue threads block.
<zmike>
jekstrand: I won't stop you, but it seems like a waste of effort since none of the code would be shared
MajorBiscuit has joined #dri-devel
<zmike>
on the cpu syncobj side
<jekstrand>
zmike: That's fine. We don't need to use syncobj.
<jekstrand>
But I'll try to review this first. It'll get me used to the code a bit.
MajorBiscuit has quit []
oneforall2 has joined #dri-devel
<zmike>
I won't stop you, but it feels like finding an untested corner case of the spec isn't grounds for spending however much of your time this is going to demand
kts has quit [Quit: Konversation terminated!]
fxkamd has quit []
fxkamd has joined #dri-devel
oneforall2 has quit [Remote host closed the connection]
naveenk2 has quit []
oneforall2 has joined #dri-devel
mszyprow has joined #dri-devel
Major_Biscuit has quit [Ping timeout: 480 seconds]
lynxeye has quit [Quit: Leaving.]
tobiasjakobi has joined #dri-devel
tobiasjakobi has quit [Remote host closed the connection]
gouchi has joined #dri-devel
jkrzyszt_ has quit [Ping timeout: 480 seconds]
<Lyude>
pinchartl: if you haven't been receiving anything from there I think we need to figure out the password for the admin interface on that list, since I've been noticing all my emails to it have been going to the moderation queue
* Lyude
had been hoping someone was actually at the other end of that queue but apparently not it sounds like…
cphealy_ has quit []
<pcercuei>
lkml.org is down?
<mceier>
at least http doesn't load here, but it still responds to pings
<karolherbst>
jekstrand: we are slowly getting there: Pass 2042 Fails 88 Crashes 46 Timeouts 0: 100%|
<karolherbst>
jenatali: I saw you had a fdiv scaling pass, was that to fix failing tests like math_brute_force exp?
<jenatali>
karolherbst: I think it was for the fdiv tests but I don't remember offhand
rkanwal has quit [Ping timeout: 480 seconds]
<karolherbst>
I do have a couple of tests failing due to low fp64 precision
<karolherbst>
mhh, I think you just bail with doubles? :/
<karolherbst>
but come on "ERROR: exp: 3.032227 ulp error at 0x1.0000000000400p+2 (0x4010000000000400): *0x1.b4c902e2755a5p+5 vs. 0x1.b4c902e2755a8p+5" :(
<karolherbst>
mhh but that's all lowered inside libclc
CATS has quit [Remote host closed the connection]
<karolherbst>
it uses fdiv though.. let's see
<karolherbst>
yeah.. sadly doesn't fix things
<karolherbst>
jenatali: do you have any downstream fixes left for fixing up nir passesn like algebraic opt?
<jenatali>
karolherbst: nope
<karolherbst>
mhh
<karolherbst>
even fma fails here though :/ I hope it's not a bug inside libclc
<pinchartl>
Lyude: it would be a good idea to ensure that your e-mails don't get moderated :-)
<karolherbst>
mhh but most of the double conversion tests are also failing
<karolherbst>
yeah.. maybe I focus on that first, so conversions are alright before digging into why those other things fail
<karolherbst>
huh? rror for vector size 1 found at 0x0000ff86: *-0x1p-149 vs -0x1p-126
<karolherbst>
I kind of thought that -0x1p-126 is the lowest possible value for a 32 bit float
<karolherbst>
ohh wait
<karolherbst>
that's just the exponent
<karolherbst>
jenatali, jekstrand: do we even handle denorms in our convert stuff yet?
CATS has joined #dri-devel
<karolherbst>
it sounds annoying, but CL actually requires CL_FP_DENORM for fp64 :(
<jekstrand>
karolherbst: Not sure
<karolherbst>
mhh doesn't look like it though
nchery has joined #dri-devel
<karolherbst>
or maybe it's just llvmpipe
<karolherbst>
mhh
<karolherbst>
airlied: we do disable denorms in llvmpipe for x86, don't we?
<karolherbst>
is there a good way of enabling it?
<ajax>
nerf util_fpstate_set_denorms_to_zero ?
cphealy has joined #dri-devel
<karolherbst>
mhh doesn't change a thing
<jekstrand>
airlied: Do we have pointer-sized atomics on every platform lavapipe supports?
<jekstrand>
ajax: Maybe you can answer for s390. I know we do on x86 and ARM.
<jekstrand>
And I'm pretty sure POWER is covered too
<jekstrand>
Hrm... nvm. I need a lock.
<ajax>
if you mean does the ISA have them, almost certainly. if you mean does gallivm know how to emit them for that arch, probably, i would think that's an acceptance criteria for llvm itself
<ajax>
but i do not actually know s390
<jekstrand>
Does anyone/
<jekstrand>
?
pcercuei has quit [Quit: brb]
pcercuei has joined #dri-devel
mszyprow has quit [Ping timeout: 480 seconds]
rbrune has joined #dri-devel
Haaninjo has joined #dri-devel
rkanwal has joined #dri-devel
rkanwal has quit [Quit: rkanwal]
sdutt has quit [Remote host closed the connection]
sdutt has joined #dri-devel
<karolherbst>
sooo my favourite feature: inline samplers
<airlied>
jekstrand: yes I think we do
<airlied>
I don't think I've cared about non-64bit s390 at all
<karolherbst>
so better to have your opinion on that change
<zmike>
jekstrand: for sure
<zmike>
remember, lavapipe passes all tests now so if anything at all fails it's your fault and you have to fix it no matter how long it takes or how unrelated it might seem
<jekstrand>
zmike: Just applying my patented coding methodology. :P
<zmike>
it's a strong patent, that's for sure
<karolherbst>
fixing one bug causing two new ones? :P
gawin has joined #dri-devel
heat_ has quit [Remote host closed the connection]
heat_ has joined #dri-devel
asocialblade has quit []
mclasen has quit [Ping timeout: 480 seconds]
rkanwal has joined #dri-devel
dviola has quit [Quit: WeeChat 3.4]
* jekstrand
turns on a fan because it's getting warm in here with lavapipe running. :D
* jekstrand
should probably only do lavapipe development during the winter.
<zmike>
the name is apt
<jekstrand>
:D
mclasen has joined #dri-devel
shankaru1 has joined #dri-devel
shankaru has quit [Read error: Connection reset by peer]
<marex>
I really seriously didn't except there can be anything worse than ChipOne in terms of datasheet availability
<marex>
there is
<marex>
Lontium
<jekstrand>
zmike: Well, my CTS run came back clean except for conformance_version and that div-by-zero test that crashes for me but not you.
<zmike>
nice
<jekstrand>
zmike: And some timeouts but that's fine
<zmike>
I guess I have to review it then don't I
<jekstrand>
zmike: Well, it is dependent on another MR that's maybe better if dj-death reviews.
<zmike>
okay so i'll wait until you tell me it's safe
<jekstrand>
But he just sent a giant pile of MRs today so maybe I can trade some reviews with him.
<jekstrand>
Debating whether or not I want to dig into v3dv next.
<jekstrand>
Or maybe I can fight with this here ARM chromebook
mvlad has quit [Remote host closed the connection]
Duke`` has quit [Ping timeout: 480 seconds]
asocialblade has joined #dri-devel
asocialblade has quit []
mszyprow has joined #dri-devel
Guest640 has quit []
bnieuwenhuizen has joined #dri-devel
rkanwal has quit [Quit: rkanwal]
gouchi has quit [Remote host closed the connection]
eukara has joined #dri-devel
gawin has quit [Ping timeout: 480 seconds]
mszyprow has quit [Ping timeout: 480 seconds]
asocialblade has joined #dri-devel
danvet has quit [Ping timeout: 480 seconds]
rbrune has quit [Ping timeout: 480 seconds]
ahajda has quit [Read error: Connection reset by peer]
* jekstrand
starts deleting chunks of v3dv....
sdutt has quit [Ping timeout: 480 seconds]
Haaninjo has quit [Quit: Ex-Chat]
sdutt has joined #dri-devel
elongbug has quit [Remote host closed the connection]
tzimmermann_ has joined #dri-devel
<jekstrand>
Ok, this thing is a bit weird...
tzimmermann__ has quit [Ping timeout: 480 seconds]
rasterman has quit [Quit: Gettin' stinky!]
pcercuei has quit [Quit: dodo]
orbea has quit [Remote host closed the connection]
orbea has joined #dri-devel
ybogdano has quit [Ping timeout: 480 seconds]
camus has quit [Ping timeout: 480 seconds]
shankaru1 has quit []
edrex[m] has left #dri-devel [#dri-devel]
tursulin has quit [Read error: Connection reset by peer]