alyssa changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - https://gitlab.freedesktop.org/panfrost - Logs https://freenode.irclog.whitequark.org/panfrost - Transientification is terminating. Memory reductions in progress.
<Lyude> robclark: hm, alright. hoping to probably have part of the install on the emmc
<alyssa> Slow day, huh
<alyssa> So, it's writing out 0x7F60. Lovely
<alyssa> (=nan)
<alyssa> But... why
<alyssa> Lovely, it's --deliberately-- writing the nan out. Okay.
<alyssa> It's possible that we need to match a st_vary_16 with the fp16 varying but that seems... I'm sorry, that just seems improbable given that main varyings are already RGBA16F and that's fine
<alyssa> Swizzle was wrong
<alyssa> Aaand there we go :)
<alyssa> gl_PointSize works now, nice nice
<alyssa> Rather miffed I spent all day on this, but I digress
<alyssa> Aaa seeing us actually pass some dEQP tests is nice :P
<alyssa> HdkR: When we going to do regression testing? :P
<alyssa> dEQP-GLES2.functional.rasterization.*
<alyssa> Passing 35/52, failing 17/52
<alyssa> Hardly something to be proud of but days ago we were failing 52/52 so :P
<HdkR> alyssa: Hm. We either need to fix midgard first gen or I need to get some more rockpro boards
<alyssa> Maybe both
<alyssa> :p
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
<alyssa> Woaaaaah
<alyssa> dEQP actually helped me identify and fix a bug?!
<alyssa> Who'da thunk testing would actually be _useful_!
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
stikonas has quit [Remote host closed the connection]
<alyssa> Up to an 8/10 pass rate on functional.rasterization.primitives.*
NeuroScr has quit [Quit: NeuroScr]
NeuroScr has joined #panfrost
<alyssa> HdkR: gl_PointSize MR is sent off
<alyssa> ezequielg: With that dealt with, taking a look at your patch
<HdkR> alyssa: Do you know how points deal with clipping on Mali?
<alyssa> HdkR: I didn't try..
<HdkR> I wonder if Piglit has tests for it
<alyssa> "fdph"... is this just a dot product?
<HdkR> I don't remember if behaviour of point clipping is defined in GL anyway
<HdkR> alyssa: Do we need to change the tracker state to show minimum/maximum point sizes?
<HdkR> Bifrost vulkan driver claims to support point sizes from 1-1024 with a granularity of 0.0625
<HdkR> Midgard is 1-64 with the same granularity
<HdkR> Also. Is this for gl_PointSize or glPointSize? :P
<alyssa> We probably do need to set some PIPE_CAPs, yeah
anarsoul has quit [Remote host closed the connection]
<alyssa> HdkR: If I really want to groan, I'd remember there's also gl_PointCoord..
<HdkR> alyssa: I completely forgot about that
<alyssa> I only discovered it existed today while reading the PointSize section of the spec
<alyssa> HdkR: Do I handle it in this MR too?
<HdkR> neh
<alyssa> Aight
<alyssa> In that case I think we're good to merge..?
<alyssa> (Atm, we report a max point width of 255.0. I haven't actually tested so shrug)
<HdkR> Probably. In the middle of cooking atm though :P
<alyssa> Sure :P
<alyssa> Oh, lovely, we have to do a fair bit of compute ourselves (like, in the shader) to do gl_PointCoord
<alyssa> Never a dull moment
<HdkR> It isn't commonly used so it makes sense to have to simulate it
<HdkR> Biggest use case is simulation of point sprites
anarsoul has joined #panfrost
<alyssa> Mmyeah
<alyssa> HdkR: Guess I'll push it off until we're actually dealing with an app that uses it, or we're doing well on conformance tests
<HdkR> :P
<alyssa> HdkR: What? Plenty of actual bugs to fix..
<HdkR> I believe it is a good choice to skip it for now
<alyssa> +1
* alyssa really wants to squash some of these GNOME or STK related bugs but
<alyssa> Not totally sure how to approach them
<alyssa> I could just work on memory leak resolution etc but I'm pending on the DRM prep branch to be merged so I don't cause tomeu conflicts galore
<alyssa> ...Why do I have the scissor test disabled? :P
NeuroScr has quit [Quit: NeuroScr]
<alyssa> ...Ugh, it looks like I'm going to need to do clipping plane lowering for desktop GL support
<alyssa> The lack thereof is possibly (probably?) breaking STK/Neverball/etc
NeuroScr has joined #panfrost
_whitelogger has joined #panfrost
paulk-leonov has quit [*.net *.split]
ric96 has quit [*.net *.split]
ric96 has joined #panfrost
paulk-leonov has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
pH5 has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
BenG83 has quit [Quit: Leaving]
stikonas has joined #panfrost
<rtp> anarsoul: yeah, even if I've noticed you on other chans. I've not been really active lately. btw, I thought you were playing with lima and not panfrost ? :)
stikonas has quit [Remote host closed the connection]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
raster has joined #panfrost
mateo` has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
afaerber has quit [Quit: Leaving]
afaerber has joined #panfrost
raster has quit [Remote host closed the connection]
Guest3786 has joined #panfrost
Guest3786 is now known as shadeslayer[m]
<HdkR> Time for more on flight programming. Sadly won't be on panfrost
<shadeslayer[m]> raster: hey, any progress on the mail I sent?
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
<alyssa> Hm?
cwabbott has quit [Quit: cwabbott]
cwabbott has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
cwabbott has quit [Client Quit]
cwabbott has joined #panfrost
chewitt has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
pH5 has quit [Quit: bye]
anarsoul|2 has joined #panfrost
stikonas has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
pH5 has joined #panfrost
sy has joined #panfrost
afaerber has quit [Quit: Leaving]
afaerber has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
rhyskidd has joined #panfrost
<alyssa> ezequielg: I was talking to someone who pointed out why NV12's format code is probably 0xc
<alyssa> Look at it in decimal! :P
<anarsoul|2> 12?
<alyssa> anarsoul|2: Yup =P
<alyssa> For NV12
<anarsoul|2> coincidence?
<alyssa> Possibly :P
<alyssa> Or since the codes are arbitrary it was a helpful mnemonic for someone :P
<anarsoul|2> btw, how do you test it?
<anarsoul|2> i.e. how do you make blob to sample from nv12 texture?
<anarsoul|2> kmscube won't work since blob doesn't have gbm_bo_map()
<alyssa> anarsoul|2: EGLImages, practically, but we don't know of blobs that have that
<anarsoul|2> right
<alyssa> ezequielg: figured it out by bruteforce, but it's easy enough to test, since... you can't "accidentally" decode it correctly :p
marcodiego has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]