How many HDD are you running? Set to spin down or no? Those spinning all the time add up quickly.
Sleep state and power States are different things… I’ve never heard of power profiles causing issues. I’d try keeping sleep disabled in BIOS and then look into what you need to change to allow the processor to idle/downclock. There’s no reason this shouldn’t work I’m aware of.
Could try undervolting as well.
I have a very similar use case so here is my opinion.
HARDWARE
-No dGPU unless this is your PRIMARY gaming computer. (Reason: better battery life, lighter laptop, with recent AMD iGPU you have decent performance for non-VR/not massive openworld AAA games.)
-recent AMD CPU. (Reason: better performance to watt ratio than Intel which makes a big difference for most of your use cases. Better multi-core performance which makes compiling code much faster. Massively better iGPU for light-medium duty gaming.)
-atleast 16GB ram if not expandable but as much as you can reasonably budget.
-16:10 or taller aspect ratio screen (16:9 sucks on laptop size devices, the extra height makes a big difference for school, coding, browsing, pretty much everything but watching 16:9 movies)
-Resolution: personal preference. IMO 1080p or 1920*1200 for 16:10 is ideal for 14" and below laptops. Lower resolution means better battery and on a small screen the PPI is high enough. If you are OK with a trade off of battery life and want a super crisp display then 2K is the highest I would go. 4K is retarded on laptop sized screens unless you are plugged in 90% of the time and you’ll have to fuck with scaling then.
-metal body for stiffness and durability
-decent key travel (usually longer travel means better IME)
If you want to do machine learning/AI work professionally I use and recommend investing in a dedicated desktop with a large memory nvidia (cuda cores) GPU and installing the cuda drivers. Trying to cram commercially viable ai hardware into a laptop is a losing battle and you’ll end up with a worse experience for both use cases, wont be able to fit large models in the memory anyways, and end up buying a desktop for AI while being stuck with a laptop that is worse for laptop use)
SOFTWARE
#1 Nobara OS KDE - best OOB experience for gaming IMO. Easy transition from windows. Has kernel fixes and many laptop specific fixes (asusctrl for example) by default which means you have a good chance of extra features like LEDs, fingerprint, etc working without tinkering). Fedora based.
#2 Linux Mint Debian Edition (LMDE6) - best non-gaming distro to learn and grow into IMO. Access to deb packages. Stable. (nobara has been stable for me as well, but it is LMDE’s bread and butter). Ease of transition from windows. Can game just as well if you are capable of following simple instructions to configure the stuff done by default on nobara and pop (may need to manually change kernels, drivers, etc to get the best performance on new hardware)
#3 Pop_OS - used it for years, but I prefer Nobara after comparing. Ubuntu based so you have access deb packages without ubuntu’s bullshit. Setup out of the box for gaming. I got fed up with failed updates, broken packages, and sluggishness so I swapped to nobara which has been a treat.
EDIT: you can snag some good deals on amazon warehouse deals (used-like new) laptops. These are usually just open box returns and if there is anything wrong you have 30 days to return it.
I recently upgraded to an Asus vivobook S 14x OLED (M5402R) for $780 CAD ($580USD) with a ryzen 7 6800H, 16GB DDR5, a 1TB gen 4 nvme, and it has zero signs of use, slight coil whine under load that I can only hear if I put my ear next to the keyboard and don’t have any sound or music on (I suspect this was the reason for the return on mine since its a common complaint for this model. That’s what I was hoping for since I’m not that picky and its worth the steep discount IMO.) Everything works oob on Nobara. I believe lenovo also regularly heavily discounts their previous gen thinkpads which are a great option, although the AMD configs are rare. Good luck!
Edit: whoops, just realized you said freezing not crashing, and probably have a separate issue. I’ll leave this here in case it helps anyone that finds this thread with crashes a couple minutes into videos.
Had this issue ages ago, then my dad did too a year later on a different client version. Manually changing the “preferred media player” option fixed it on my firestick 4k, 4k Max, and my dads standard firestick.
Jellyfin app>settings menu>Playback>Video section>preferred media player>libVLC (in my case, Exoplayer seemed to be causing the crashes approx 2 years ago but you can try both, I just tried exoplayer again and it doesn’t seem to be crashing either when set manually now so it may have been patched)
Damn FOSS Android Auto development is starting the new year off strong! First grapheneOS successfully implementing it on a non-stock OS and now this too. Too bad I got rid of my vehicles last year and no longer have a use for it on my ebike.
Most (all) of the creators I watch have patreon/buymeacoffee/merch/sponsored videos. I’ve heard from many of them that the amount of ad revenue they receive from YouTube is a rounding error compared to that. No reason every single creator on peertube cant put a link in the description to donate, sell merch, etc and its not conditional on tiptoeing around youtube’s ai fuelled demonetization that constantly steals the little ad revenue they would have otherwise received
Clearly just upset you got conned into paying hundreds for an inferior product
I’ve used both extensively and stand by my statement, from a functional standpoint as well.
For the downloading media part:
The *arr stack is what you’re looking for + Jellyfin for streaming (Opensource, 100% free, and much better than Plex).
Prowlarr: manage your indexers
Radarr: find/automatically download movies
Sonarr: find/automatically download tv shows
Jellyfin: streaming your media
Look up trashguides for setting up all this stuff, very detailed guides. They are compatible with torrents and Usenet. I like using docker with portainer for easy management and if you use a VPN container you can selectively route these containers through the VPN so your other services that dont require the VPN dont need to route through it.
You could try llamaCPP, I think its configured to run better on cpu’s