On a serious note:
This feature is actually very useful. Libraries can use it create neat error messages. It is also needed when logging information to a file.
You should however never ever parse the source code and react to it differently.
You know that this is acutally working right??? 😊
deleted by creator
My memory sticks are all DDR4 with 32GB@2133MT/s.
Each card has 24GB so 48GB vram total. I use ollama it fills whatever vrams is available on both cards and runs the rest on the CPU cores.
My specs because you asked:
CPU: Intel(R) Xeon(R) E5-2699 v3 (72) @ 3.60 GHz
GPU 1: NVIDIA Tesla P40 [Discrete]
GPU 2: NVIDIA Tesla P40 [Discrete]
GPU 3: Matrox Electronics Systems Ltd. MGA G200EH
Memory: 66.75 GiB / 251.75 GiB (27%)
Swap: 75.50 MiB / 40.00 GiB (0%)
What are you asking exactly?
What do you want to run? I assume you have a 24GB GPU and 64GB host RAM?
I regularly run llama3 70b unqantesized on two P40s and CPU at like 7tokens/s. It’s usable but not very fast.
True multiple drives speed up reads significantly. As long as the videos are sequential read speeds can be very fast (600MB/s) even on one drive though. Results may vary.
I have a ~40TB HDD array and jellyfin is super fast. Just put the database and cache files on a SSD.
For bulk storage of 4k videos with high bitrates HDDs are way cheaper.
Which os are you running?
Try to partition it with free space at the end and see if it makes a difference.
Try to trim the drive and see if it speeds up again.
Do you use any disk encryption?
Llama3.1 33b would be so cool. It would be a nice middle ground for my machine.
At least on linux rm is very fast
I use tubearchivist. It has a jellyfin addon but it could really use some improvements on how it exposes the videos.
What did you use to get the information for the current internet throughput?
Full pass through has no advantage when my reverse proxy terminates ssl and internal services are http only right?
Regardless of fqdn nginx has to decrypt and restream anyways.
I think mixing RAM sticks is mostly fine today. Maybe you won’t get 100% performance but I don’t think it will be very noticeable. You may still run into issues with some capacity combinations depending on the mainboard/cpu. Regarding clock speeds usually all run on the clock of the slowest one.
Matching RAM latency also matters for performance.
When using different capacity RAM channels matter so take care on the order of population.
Let’s say it counts. What made you STOP?
Imagine if nerv had invested some of its money in the mental health of its employees. I like to think many issues could have been solved by hiring a few therapists.
Well now that causes breakage two dependencies down the line. Good luck with that. 😅