As Intel QS and plex’s use of hardware acceleration basically offloads the work from the CPU, is the passmark still relevant?
For example, i know that the Intel Celeron J5005 can do up to 4 simultaneous 1080p transcodes (lon.tv youtube review) yet it has a passmark of 2904.
Is this score only relevent for direct CPU transcoding? If so, how can we find out the performance of QS on such CPUs to predict how many streams can be transcoded?
it may not be directly accurate, but the cpu performance is still very relevant;
even with a gpu (like a p2000 quadro) that is often highly recommended, a server will run out of CPU (or disk/network IO) before saturating the gpu.
so, long story short, gpu can be a huge help in transcoding tasks.
the cpu still does everything else, and if your cpu is slow/low powered, then your server is going to perform accordingly for non transcoding tasks.
whether that is sufficient for anyone’s particular usage, is entirely dependent on their library size and number of active users.
Cheers for the reply…
Its hard trying to actually pick the right spec/solution! I would like a consumer NAS, but I dont think they are quite there yet, perhaps when they start using the J5005, as I consider 4x 1080p streams to be my max requirement.
Im also considering self build, which could come in about 2/3 the cost of the NAS (excluding storage), but I like all of the other features that the NAS offer, in a neat little box. Im not sure the likes of freenas/unraid could rival them, but requires further research.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
