marcan changed the topic of #asahi-gpu to: Asahi Linux GPU development (no user support, NO binary reversing) | Keep things on topic | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-gpu
dorkbutt has joined #asahi-gpu
nate8_ has joined #asahi-gpu
nate8 has quit [Ping timeout: 480 seconds]
Kaz has joined #asahi-gpu
Kaz has quit []
dorkbutt has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dorkbutt has joined #asahi-gpu
user982492 has quit [Quit: Textual IRC Client: www.textualapp.com]
mental has joined #asahi-gpu
zalyx has quit [Quit: later alligator]
zalyx has joined #asahi-gpu
mental is now known as lugpocalypse
dorkbutt has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dorkbutt has joined #asahi-gpu
lugpocalypse has quit [Quit: Konversation terminated!]
<lina> jannau: For gnome terminal you need to unset the use system color scheme thing, else the background breaks. I don't know what's up with that...
pjakobsson has joined #asahi-gpu
pjakobsson_ has quit [Ping timeout: 480 seconds]
dorkbutt has quit [Read error: Connection reset by peer]
captainmurphy has joined #asahi-gpu
captainmurphy has left #asahi-gpu [#asahi-gpu]
captainmurphy has joined #asahi-gpu
captainmurphy has left #asahi-gpu [#asahi-gpu]
cr1901_ has joined #asahi-gpu
cr1901 has quit [Read error: Connection reset by peer]
cr1901_ is now known as cr1901
WindowPain_ has joined #asahi-gpu
WindowPain has quit [Ping timeout: 480 seconds]
<jannau> all good, I was just testing mpv across a couple of window managers/compositors
SSJ_GZ has joined #asahi-gpu
captainmurphy has joined #asahi-gpu
captainmurphy has left #asahi-gpu [#asahi-gpu]
LinuxM2 has joined #asahi-gpu
captainmurphy has joined #asahi-gpu
SSJ_GZ has quit [Ping timeout: 480 seconds]
SSJ_GZ has joined #asahi-gpu
SSJ_GZ has quit [Ping timeout: 480 seconds]
pip[m]1 has joined #asahi-gpu
pip[m]1 has left #asahi-gpu [#asahi-gpu]
SSJ_GZ has joined #asahi-gpu
captainmurphy has left #asahi-gpu [#asahi-gpu]
MajorBiscuit has joined #asahi-gpu
LinuxM1 has joined #asahi-gpu
LinuxM1 has quit [Quit: Leaving]
LinuxM2 has quit [Quit: Leaving]
dordoka has joined #asahi-gpu
dordoka has quit [Quit: Konversation terminated!]
cylm has joined #asahi-gpu
dordoka has joined #asahi-gpu
dordoka has quit []
minecrell8 has quit []
minecrell has joined #asahi-gpu
DarkShadow44 has quit [Quit: ZNC - https://znc.in]
DarkShadow44 has joined #asahi-gpu
mkurz has joined #asahi-gpu
mkurz has quit []
mkurz has joined #asahi-gpu
mkurz has quit [Remote host closed the connection]
mkurz has joined #asahi-gpu
bcrumb has joined #asahi-gpu
bcrumb has quit [Quit: WeeChat 3.7.1]
yuyichao_ has joined #asahi-gpu
bcrumb has joined #asahi-gpu
yuyichao has quit [Read error: Connection reset by peer]
nate8_ has left #asahi-gpu [Leaving]
LinuxM1 has joined #asahi-gpu
WindowPain_ is now known as WindowPain
bcrumb has quit [Quit: WeeChat 3.7.1]
bcrumb has joined #asahi-gpu
bcrumb has quit []
bcrumb has joined #asahi-gpu
bcrumb has quit []
LinuxM1 has quit [Quit: Leaving]
LinuxM1 has joined #asahi-gpu
systwi has quit [Ping timeout: 480 seconds]
systwi has joined #asahi-gpu
WindowPain has quit [Quit: ZNC 1.8.2 - https://znc.in]
WindowPain has joined #asahi-gpu
MajorBiscuit has quit [Ping timeout: 480 seconds]
WindowPain has quit [Quit: ZNC 1.8.2 - https://znc.in]
WindowPain has joined #asahi-gpu
mkurz has quit [Read error: Connection reset by peer]
LinuxM1 has quit [Quit: Leaving]
bcrumb has joined #asahi-gpu
bcrumb has quit []
cylm has quit [Quit: WeeChat 3.6]
mkurz has joined #asahi-gpu
WindowPain has quit [Quit: ZNC 1.8.2 - https://znc.in]
WindowPain has joined #asahi-gpu
cylm has joined #asahi-gpu
WindowPain_ has joined #asahi-gpu
WindowPain_ has left #asahi-gpu [#asahi-gpu]
WindowPain has quit [Ping timeout: 480 seconds]
WindowPain has joined #asahi-gpu
bcrumb has joined #asahi-gpu
WindowPain_ has joined #asahi-gpu
WindowPain has quit [Read error: Connection reset by peer]
WindowPain_ is now known as WindowPain
LinuxM1 has joined #asahi-gpu
bcrumb has quit [Quit: WeeChat 3.7.1]
LinuxM2 has joined #asahi-gpu
LinuxM2 has quit [Quit: Leaving]
mkurz has quit [Read error: Connection reset by peer]
ariel has joined #asahi-gpu
ariel is now known as Guest1386
Guest1386 is now known as facekapow
facekapow has quit []
bcrumb has joined #asahi-gpu
bcrumb has quit []
bcrumb has joined #asahi-gpu
bcrumb has quit []
mkurz has joined #asahi-gpu
erik has joined #asahi-gpu
<corion> The gpu driver is ridiculously nice.
bcrumb has joined #asahi-gpu
bcrumb has quit []
Venemo_ is now known as Venemo
erik has quit [Remote host closed the connection]
cylm has quit [Quit: WeeChat 3.6]
LinuxM1 has quit [Quit: Leaving]
LinuxM2 has joined #asahi-gpu
c10l has quit [Quit: Bye o/]
c10l has joined #asahi-gpu
LinuxM2 has quit [Remote host closed the connection]
LinuxM2 has joined #asahi-gpu
amw has quit [Ping timeout: 480 seconds]
LinuxM2 has quit [Quit: Leaving]
mkurz has quit [Ping timeout: 480 seconds]
alyssa has joined #asahi-gpu
<alyssa> I think this mechanism might be busted.
<alyssa> One bug jumps out:
<alyssa> we need to add AGX_VDM_STREAM_LINK_LENGTH to space at the beginning
<alyssa> so that when we ensure there's room each time, there's actually room + enough for a final stream link for the *next* draw
<alyssa> that assert then is wrong
<alyssa> We actually need to assert something stronger:
<alyssa> assert(batch_encoder_current + AGX_VDM_STREAM_LINK_LENGTH <= batch->encoder_end && "...")
<alyssa> i.e. we need to make sure we left enough room for the next draw to link to a new stream
<alyssa> without overflow on the link
<alyssa> It's unclear whether that bug can be hit in practice, but if not it's only by accident
<alyssa> I woudl also check whether we need any special allocation for the stream
<alyssa> i.e. low 32-bit perhaps?
<alyssa> I don't remember what macOS does
<alyssa> it's also possibly the XML for VDM_STREAM_LINK itself is busted though I don't think so, because otherwise we couldn't trace across stream links (and we can)
<alyssa> ensure_cmdbuf_has_space should probably also assert its own precondition
<alyssa> assert(batch_encoder_current + AGX_VDM_STREAM_LINK_LENGTH <= batch->encoder_end && "...")
<alyssa> this indicates that the control stream ("encoder") is allocated as regular memory. so not the 32-bit thing
<alyssa> but it's possible there is a heavier alignment requirement
<alyssa> Here we allocating with 256 alignment, but maybe we need more than that
<alyssa> this code path only triggers at high draw call counts
<alyssa> which explains how I missed it
<alyssa> 10k fish at aquarium would probably hit it, 1k would probably not
<alyssa> and come to think of it I've actually hit this bug twice and not debugged in either case because I thought it was something else
<alyssa> once in a patch to shrink the default encoder size, such that instead of allocating 0x80000 we only allocate 64k and then we end up hitting this issue at much smaller draw call counts, once in drawoverhead
SSJ_GZ has quit [Ping timeout: 480 seconds]
<alyssa> anyway I'm supposed to be on break right now but I was eating dinner and it clicked what was going wrong
<alyssa> so there's the relevant info, if jannau or lina or anyone else wants to write the fix
thevar1able has quit [Remote host closed the connection]
thevar1able has joined #asahi-gpu
thevar1able has quit [Remote host closed the connection]
thevar1able has joined #asahi-gpu
thevar1able has quit [Remote host closed the connection]
thevar1able has joined #asahi-gpu
thevar1able has quit [Remote host closed the connection]
thevar1able has joined #asahi-gpu