3900x vs 3950x - Better for recording?

[TheWhiteRose000]

The question of the hour is, what would be generally better for recording.

The Ryzen 3900x or the Ryzen 3950x.

 

Now I am aware that Nvidia has Nvenc which can yield a better encode but the problem here is the system would have

2 Elgato 4k 60 capture cards and as far as I'm aware Nvenc has issues capturing from multiple sources.

 

It could also be noted that this could be split between the GPU's Nvenc encounter and the CPU but for practicality if a person wanted to switch

between recording and streaming what would be the best option for a high quality encoding.

 

3900x or 3950x

Software of choice here is OBS, Resolution is 1080p and FPS 60 at Placebo speed.

(would it be possible to achieve Placebo speed?)

 

So in this instance what would be the best option?

For high quality editing that can encode 4K footage from two sources and scale it down to 1080p for the CPU's and would it even be possible to achieve the Placebo speed on a 3950x?

Is the 3950x worth the extra bank or is it simply a wiser decision to buy a 3900x?

[TheWhiteRose000]

4 minutes ago, Abyssal Radon said:

Really, if you can afford a 3950X, why not just get the 3950X? 

Trying to future proof, but on a tight budget.
If I could find a 3950x for $580-650 on ebay I'd get that but the $470 3900x is looking more affordable.

[TheWhiteRose000]

10 minutes ago, Abyssal Radon said:

Well then shop around then. You probably won't see a massive difference between a 3900X and 3950X.

I know in terms of gaming, yea there wont be a big difference but I'm focusing more on encoding.

[Enderman]

The 3950X will be better by quite a bit, so just get that and don't upgrade it for a long time.

[mariushm]

There's almost NEVER a reason to use placebo preset when encoding. It's pointless.

Those presets are about enabling or disabling features of the encoder in order to improve speed of encoding when you have other constraints like bitrate (maximum bits that can be used within a second)

For recording to disk (Archiving) you can use quality factor mode where you can tell the encoder to compress every frame of the video at a minimum quality level... for example it's like saying save each frame as JPG at 95% quality (usually you should not notice quality loss between 100% and 92-95%). In such mode, the profile you choose just determines how many features or modes the encoder can use to reduce the amount of disk space is gonna be used, it doesn't result in more quality.

So for example, you could use veryfast with qp 5 and you'll get super fast compression but maybe 100 mbps bitrate for 1080p 60fps ... using "veryslow" will result in same quality (because quality is locked through the qp 5 parameter) but maybe bitrate would average 90 mbps and you'll use more cpu.

So you could save the stream to disk in near lossless quality using very little cpu (ultrafast or veryfast profile in x264) and high bitrate (let's say 100mbps for 1080p 60fps and 200-400mbps for 4K) and then you could recompress using "veryslow" preset and more reasonable bitrates like variable 40 mbps for 1080p 60fps (and keep that for upload to youtube or further processing/rendering etc)

 

For streaming, generally the quality improvements are minimal once you go to "slower"... "slow" and "veryslow" presets usually bring very little improvements in quality.90% of the time . Maybe then next 8..20 techniques go within 95-98% of the best result but only happens less than 5% of the time and the other techniques and the last

[TheWhiteRose000]

18 minutes ago, mariushm said:

There's almost NEVER a reason to use placebo preset when encoding. It's pointless.

Those presets are about enabling or disabling features of the encoder in order to improve speed of encoding when you have other constraints like bitrate (maximum bits that can be used within a second)

For recording to disk (Archiving) you can use quality factor mode where you can tell the encoder to compress every frame of the video at a minimum quality level... for example it's like saying save each frame as JPG at 95% quality (usually you should not notice quality loss between 100% and 92-95%). In such mode, the profile you choose just determines how many features or modes the encoder can use to reduce the amount of disk space is gonna be used, it doesn't result in more quality.

So for example, you could use veryfast with qp 5 and you'll get super fast compression but maybe 100 mbps bitrate for 1080p 60fps ... using "veryslow" will result in same quality (because quality is locked through the qp 5 parameter) but maybe bitrate would average 90 mbps and you'll use more cpu.

So you could save the stream to disk in near lossless quality using very little cpu (ultrafast or veryfast profile in x264) and high bitrate (let's say 100mbps for 1080p 60fps and 200-400mbps for 4K) and then you could recompress using "veryslow" preset and more reasonable bitrates like variable 40 mbps for 1080p 60fps (and keep that for upload to youtube or further processing/rendering etc)

 

For streaming, generally the quality improvements are minimal once you go to "slower"... "slow" and "veryslow" presets usually bring very little improvements in quality.90% of the time . Maybe then next 8..20 techniques go within 95-98% of the best result but only happens less than 5% of the time and the other techniques and the last

You think long run with two elgato cards it would be better to use one for the game rendered through nvenc and the other through the CPU?
Process wise do you think a 3900x would be able to handle all of that, or would it be better to just aim for a 3950x?