Selur's Little Message Board

Tl;dr: How much faster will my software encoding become, if I switch to a same gen CPU with more cores\threads?

Currently I'm encoding HEVC (with some VapourSynth filters) on an old Haswell with 4 cores\8 threads. And I've been eyeballing some Xeon cpus of the same gen with up to 18 cores (but at a lower frequency). There is no hardware HEVC support in this generation, and I guess the multithreading boost alone might actually save me a lot of time.
I didn't find any relevant comparisons except a "Premiere Pro CC 2019 8K to 4K Encoding Time" benchmark on cpuagent.com which let's me hope for around x4 speed enhancement in the best case. So maybe someone can advise, is there a point in such an "upgrade" singularly from the encoding speed point of view.
On the other hand, as I've read there seems to be an inverse relationship between thread count and encoding\compression quality. Any feedback from this point of view will be also greatly appreciated.

For HD and higher resolutions x265 should scale relatively linear. So doubling the core count while keeping the frequency would double the fps.
Vapoursynth can vary a lot.
If you post more details about your workflow I can run it with my Ryzen R9 3950x and report back.

Cu Selur

(22.12.2021, 13:20)Selur Wrote: [ -> ]For HD and higher resolutions x265 should scale relatively linear. So doubling the core count while keeping the frequency would double the fps.
Vapoursynth can vary a lot.
If you post more details about your workflow I can run it with my Ryzen R9 3950x and report back.

Cu Selur

Since those old Xeons are really cheap and after some more thinking I went and ordered a mobo+cpu kit. So your answer is very reassuring, thank you.
I can't even think about troubling you with any test runs, but I will try to post my own comparison results for future reference when I'll set everything up.