Yesterday, 08:24 AM
Hello im trying to enter the self hosted family and i started with installing ubuntu server on a random 3.2 usb to start pratcing with jellyfin and other self hosted stuff.
So the computer that i used, to boot the ubuntu instance, is my laptotp and its rocking an Ryzen 5 5600H and rtx 3060 6GB desktop with 8gb ram.
I activated hardaware acelarartion for my nvidia driver in ubuntu, did all the process installed drivers and all thoes shenanigans.
I tought of the possible worst case scenario for me, that is a 1080p hevc with a very conbulated .ASS subtitle, and when i went in another instace of jellyfin to the transcode process, to my nvidia shield, it was transcoding wuat i expect since it will need to burn in the subs, but wuat i wasent expecting was an almost 70-85% cpu usage on only 1 stream, and yes i did the nvidia-msi comando to check if the gpu was being used and it was using the ffmpeg with 220 MB i think it was MB.
So i wuanted to do this post to check if the hardware that im planning to use for my server will be fine, im planning on using an i5-13400 ou i5-14400 with an intel Arc a310 4gb with a 16GB Ram, by the way on my desktop with was using like 30% of the 8GB of ram so i think its normal.
Im still on the ropes between this cpu and gpu combo, or just go with a more powerfull CPU with integrated graphics, and im planning on having 2 to 3 streams at the same time, but not all transcoding files like this and not the same cliente, this cliente is just worst case scenarion for 1 user, i have some 4K files hevc but i think, if i go with gpu route, the gpu can do them easly.
Sry for bad English on some parts.
So the computer that i used, to boot the ubuntu instance, is my laptotp and its rocking an Ryzen 5 5600H and rtx 3060 6GB desktop with 8gb ram.
I activated hardaware acelarartion for my nvidia driver in ubuntu, did all the process installed drivers and all thoes shenanigans.
I tought of the possible worst case scenario for me, that is a 1080p hevc with a very conbulated .ASS subtitle, and when i went in another instace of jellyfin to the transcode process, to my nvidia shield, it was transcoding wuat i expect since it will need to burn in the subs, but wuat i wasent expecting was an almost 70-85% cpu usage on only 1 stream, and yes i did the nvidia-msi comando to check if the gpu was being used and it was using the ffmpeg with 220 MB i think it was MB.
So i wuanted to do this post to check if the hardware that im planning to use for my server will be fine, im planning on using an i5-13400 ou i5-14400 with an intel Arc a310 4gb with a 16GB Ram, by the way on my desktop with was using like 30% of the 8GB of ram so i think its normal.
Im still on the ropes between this cpu and gpu combo, or just go with a more powerfull CPU with integrated graphics, and im planning on having 2 to 3 streams at the same time, but not all transcoding files like this and not the same cliente, this cliente is just worst case scenarion for 1 user, i have some 4K files hevc but i think, if i go with gpu route, the gpu can do them easly.
Sry for bad English on some parts.