Add 4k in benchmark

suggest a way to improve Neat Video
Post Reply
jpsdr
Posts: 221
Joined: Mon Aug 11, 2008 7:33 am

Add 4k in benchmark

Post by jpsdr »

Hello.

I think it would be very interesting to add a 4k profile in the benchmark test, because today it stop at 1080p.

4k will become more and more used. I don't say everyone will use it, but the people using it actualy increase.

I have this WE upgraded my dedicated PC video from a 8 cores Haswell to a 10 cores Broadwell.
I've allready noticed on the Haswell, in the benchmark, that there was almost no significant increase/differences in the last steps (14,15 and 16 CPUs).
It was even more obvious with the Broadwell, speed was almost identical between 14 and 20 CPUs, to finaly choose 16 CPUs as best result. This was in 1080p.
I think i obviously hit some kind of saturation limitation. But... Is this still true for 4k...? I don't know, i can't benchmark at it.

Of course, i may be wrong, but i have the feeling that adding a 4k profile in the benchmark may not be a too much big work, and this is something that can be very interesting for the next version.

As i said, 4k will become more and more used, and people working on 4k with high-end CPU should have the possibility to accurately select the best working profile for them.
NVTeam
Posts: 2745
Joined: Thu Sep 01, 2005 4:12 pm
Contact:

Post by NVTeam »

If you open a 4k clip in your video editor and then open Neat Video, it should offer 4k in that list too, because it always includes the current frame size in that list.

Thank you,
Vlad

P.S.
> Is this still true for 4k
I think so.
jpsdr
Posts: 221
Joined: Mon Aug 11, 2008 7:33 am

Post by jpsdr »

In fact, it's GPU the guilty, it's more obvious with 4k than 1080p.

With CPU only, each core added increase speed until the end, even with 20 cores. Of course, last steps increase speed are lower than first.

But, with GPU and CPU, speed increase some kind of stabilize around 10-12 cores. After, until the end, it's slighty fluctuating, resulting a best combination changing between each benchmark tests from 10 to 16 cores with GPU (result is somehow random between these two values).

Nevertheless, GPU+CPU is still faster than CPU only (GTX950 card).
Post Reply