Nouveau Open-Source NVIDIA Tests On Linux 4.8, Mesa 12.1-dev

[dungeon]

AMD is unbelievable, you can Pick and Choose parts even of two different drivers and made one you like

[duby229]

AMD is unbelievable, you can Pick and Choose parts even of two different drivers and made one you like

But, now I'm lost again.. I have no idea how to interpret your meaning. spell it out for me better so I can understand more clearly.

[bridgman]

AMD is unbelievable, you can Pick and Choose parts even of two different drivers and made one you like

But, now I'm lost again.. I have no idea how to interpret your meaning. spell it out for me better so I can understand more clearly.

People are playing "driver Lego" aka making frankenbuilds combining userspace drivers from the hybrid/PRO stack with kernel driver/libdrm from upstream, in order to get PRO stack features on distros we are not yet supporting. I think that's what dungeon is talking about.

[bridgman]

Sigh... moderated again. My post quoted both dungeon and duby229 then tried to explain what dungeon meant, so that's probably why it got moderated.

[dungeon]

I am thinking of that too, but also various other combinations... even on Windows it is posibile to inject EG/NI support even when support for those was "dropped"

So i think of, when AMD drop support it is in some ways not actually dropped, while when nVidia claim longterm support they actually does not do it Example here:

https://www.phoronix.com/forums/forum/phoronix/latest-phoronix-articles/893124-nvidia-367-44-stable-linux-driver-released?p=893184#post893184

That 340 serie is "supported" till end of 2019. but actually nVidia does not update driver in last... how many... near 10 months

[DanL]

You mean lowest for high quality roasting.

Oh, that's not good, What's the French word for that? Do they have one or should we just call it French Roast? "GPU Holocaust?"

[Enverex]

For many people, usage of closed blobs is out of question

Who are these people and why? I mean I get that the OSS idea of drivers is better, but I'm only seeing this as a preference, not a requirement for those people.

[wpupkin]

I have 750 Ti , and it runs as expected (except one error in dmesg: bus: MMIO read of 00000000 FAULT at 3e6684 [ IBUS ], but maybe not error at all)
It's very special and cold device, so I can safely set nouveau.config=NvBoost=2 kernel option to reach boosted frequencies, and it runs like GeForce GTX 760 in Unigine Valley (or like GeForce GTX 780 Ti with antialiasing disabled)

hwmon never shows temperature more than 50+/- degrees and fun - 40% , so (I think) it's safe to set higher boost frequency
for example here is some info from official site (Palit):
GeForce^® GTX 750 Ti StormX (1024MB GDDR5) : Base Clock : 1020MHz / Boost Clock : 1085MHz
GeForce^® GTX 750 Ti StormX OC (2048MB GDDR5) : Base Clock : 1085MHz / Boost Clock : 1163MHz <- my model
GeForce^® GTX 750 Ti StormX Dual (2048MB GDDR5): Base Clock : 1202MHz / Boost Clock : 1281MHz <- 2x fun model

pstate 07: # cat /sys/kernel/debug/dri/0/pstate :

07: core 405 MHz memory 810 MHz AC DC *
0f: core 270-1306 MHz memory 5500 MHz
AC: core 405 MHz memory 810 MHz

hwmon: core:931/temp:34/fan:30

pstate 0f: # cat /sys/kernel/debug/dri/0/pstate :

07: core 405 MHz memory 810 MHz
0f: core 270-1306 MHz memory 5500 MHz AC DC *
AC: core 1215 MHz memory 5499 MHz

hwmon: core:1175/temp:45/fan:37 (on load temp:51/fan:40)

so, I think it's safe for this model to run boosted as StormX Dual, maybe up to 80-85 degrees