Is Clear Linux in early stages still? (Clear Linux & GPU switching for Machine Learning)

I switched from mint to clear linux discovering mint is not as fast as windows for my intel i7. I though it was because of processor optimization stuff because I don’t think windows can be a good OS ever especially a fast one lol. However after switching to CL improvement was visible but the issues started with nvidia drivers. Idk which application it was but I was able to use CUDA on demand in linux mint.

When trying to install I was unsucessful probably because I couldn’t disable a GPU, as a result I ended up being in stuck in the black CLI without GUI. Instead of trying to solve from the cli I just did another clean install but this time I used lts kernel. my pulse audio driver stuff was misconfigured somehow then switched to native kernel again and it was solved but now I have scary red lines on my dmesg output.

An example from dmesg

traps: gnome-photos[4763] trap int3 ip:7f0c7dad918e sp:7ffec65e0650 error:0 in[7f0c7da6f000+f0000]
[ 1302.216801] crashprobe[4858]: segfault at 10 ip 00007f6cdb6cc728 sp 00007ffdeac490a0 error 4 cpu 7 in[7f6cdb68e000+45000]
[ 1302.216808] Code: ff c0 48 39 45 18 75 e9 eb e3 31 c0 80 7d c4 02 75 18 2e 2e 49 8b 14 c0 48 0f ca 49 89 14 c4 48 ff c0 48 39 45 18 75 ea eb c3 <49> 8b 14 c0 49 89 14 c4 48 ff c0 48 39 45 18 75 ef eb b0 b0 01 48
[ 1302.216818] Process 4858(crashprobe) has RLIMIT_CORE set to 1

My Device: Dell G3 15, Nvidia 1660Ti
I use frequent Machine Learning and I want to be able to switch my GPU with Bumblebee/Primus or something.

My Question Is: Should I use clear linux*. I can learn & modify stuff and solve this issues the problem is that I don’t have the time and it feels unstable honestly. If I can contribute for my devices here on clear linux it would become more stable I think, at least the code will be reviewed. Is CL at this stage where I can install/use bumblebee from swupd in the following 3 months or so?

If thats the case I would like to make the effort.

CL, as any rolling Linux distros, can totally crash your system at any time when you run swupd update.

As a workaround I use 2 servers for ML … 1 production server, and one test server, only if the test server is stable, it becomes production and the previous production server becomes next test server.

So the downtime will be zero and I can always integrate new hardware and software updates without getting white hair, and still use the speed benefit of CL.

1 Like

Ok thanks. I might coinsider that kind of setup. Right now my work was urgent so II installed pop_os for nvidia support. Nice to have in mind to use servers, not sure how they will work but I willl look into it.

Another Question: Do you use bumblebee for GPU/dGPU switch. It would be nice to see if it is possible.