01:05cynicfm: hey i use intel hd620 kaby lake with open source drivers is there any tweaking available for this gpu?? thanks :)
01:06HdkR: This is the channel for the open source driver for Nvidia graphics, but I'd presume no for that
01:08HdkR: That's a 24EU GPU, which isn't exactly a powerhouse
05:07imirkin: skeggsb: ping
05:07skeggsb: imirkin: hey
05:08imirkin: welcome back
05:08imirkin: i sent some patches
05:08imirkin: and there was the fan speed patch for nv4x
05:08imirkin: just want to make sure they haven't fallen off the edge of your inbox
05:21skeggsb: imirkin: thanks :)
05:22skeggsb: and yeah, they kinda had, thanks for the reminder
12:37karolherbst: l4mRh4X0r: were you running into further minecraft issues with my branch?
12:43l4mRh4X0r[m]: Nope, haven't found any
12:44l4mRh4X0r[m]: I haven't played using nouveau for too long though, it has about half the frame rate i965 has
12:44karolherbst: l4mRh4X0r[m]: maxwell2 gpu?
12:45karolherbst: can be reclocked on a laptop, just needs a couple of patches... (should really get those upstream)
12:46l4mRh4X0r[m]: maxwell 1, according to wikipedia :)
12:53l4mRh4X0r[m]: Can that one be reclocked as well, and if so, how?
13:43karolherbst: l4mRh4X0r[m]: it can: you should have a /sys/kerne/debug/dri/1/pstate file
13:43karolherbst: cat it to get the perf states
13:43karolherbst: and select the perf state by echoing it into the file as root
13:44karolherbst: like "echo 0xf > pstate"
13:44karolherbst: * /sys/kernel/debug/dri/1/pstate
13:50l4mRh4X0r[m]: Yeah, that's a much more comparable framerate
13:50l4mRh4X0r[m]: Still lower though, as far as I can tell
14:11l4mRh4X0r[m]: karolherbst: just ran into a bug:
14:11l4mRh4X0r[m]:sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/zSuqOQccncVRhCkMLtqhxeOf >
14:12karolherbst: that more or less means the device isn't fast enough working through all the commands afaik
14:12karolherbst: and we don't throttle on the kernel side
14:13l4mRh4X0r[m]: And now this again:
14:13l4mRh4X0r[m]:sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/vDRJFDJXaOvOdbVeXJNHdzye >
14:14l4mRh4X0r[m]:posted a file: hs_err_pid3757.log (187KB) < https://matrix.org/_matrix/media/v1/download/canarymod.net/unuvFhbEfrlQjOkrLYeOBUoc >
14:14l4mRh4X0r[m]: That's the whole error log java produced
14:15karolherbst: l4mRh4X0r[m]: well, after a pushbuf was rejected it can lead to all kind of weird issues
14:16l4mRh4X0r: That error is from a new process by the way
14:16karolherbst: ahh, I see
14:16l4mRh4X0r: Killed the other one with sigkill
14:17karolherbst: *sigh* those java reports are also totally useless for debugging mt issues
14:19karolherbst: but I think I know what this issue is, it is just relativly hard to trigger with random applications
14:20karolherbst: I simply don't have a nice solution for this yet
14:22l4mRh4X0r[m]: Alright. If you need me to test something, let me know.
14:23karolherbst: what GPU is that btw? wondering why the highest pstate is not faster than the intel GPU :/
14:23karolherbst: (except you are already exceeding 100 fps, but I highly doubt the intel gPU would be able to do that either)
14:24l4mRh4X0r: Hah, nope
14:24l4mRh4X0r: According to lspci: NVIDIA Corporation GM107GLM [Quadro M1000M] (rev a2)
14:26karolherbst: okay, that GPU is slow
14:26karolherbst: could be worse though
14:29karolherbst: faster then intel GPUs, but not much... I guess we could do a better job inside nouveau to close the gap
14:36l4mRh4X0r[m]: Now I got this message and a call trace: `fifo: fault 00 [READ] at 0000000024bd6000 engine 1b [CE2] client 18 [HUB/GR_CE] reason 42  on channel 2 [007fb24000 java]`
14:37l4mRh4X0r[m]: The call trace after "Asynchronous wait on fence timed out"
14:38l4mRh4X0r[m]: In my kernel log by the way, if that wasn't obvious
21:12pmoreau:Fingers crossed... 10th time’s the charm?
22:50pmoreau: :-( I was certain I could use Nouveau just fine on my GP102, but that is no longer the case it seems, getting some "gr: intr 0x00000040" and "disp: outp 03:0006:0144: training failed".