StaxRip 2.0 Support Thread

[Atak_Snajpera]

BTW. Do you even need prefetch on such small numbers of cores? It would be funny if you got even higher fps without those extra AviSynth threads

[Bahamuth]

I just stumbled over this:
http://www.chaneru.com/Roku/HLS/X264...gs.htm#threads
"Default: auto (frame based threads: 1.5 * logical processors, rounded down; slice based threads: 1 * logical processors)"

Could it be, that x264 is limiting itself? Should I tamper with this setting?

[poisondeathray]

Originally Posted by Atak_Snajpera

BTW. Do you even need prefetch on such small numbers of cores? It would be funny if you got even higher fps without those extra AviSynth threads

Hahaha funny guy

Atak I believe what you're saying is true for your case your HW, and your scripts.

I've never seen a script or situation where "2" was worse speedwise than no prefetch (on a dual core at least, I guess) . It's just a bit of a pain to find the "ideal" prefetch value

Some people look at CPU usage, without measuring speed, it's a big mistake . Sometimes high CPU usage is thrashing, not "faster" processing or encoding

[Atak_Snajpera]

Some people look at CPU usage, without measuring speed, it's a big mistake

I look at fps reported by AVSMeter...

I just stumbled over this:
http://www.chaneru.com/Roku/HLS/X264...gs.htm#threads
"Default: auto (frame based threads: 1.5 * logical processors, rounded down; slice based threads: 1 * logical processors)"

Could it be, that x264 is limiting itself? Should I tamper with this setting?

vspipe is a problem (Does not send frames for encoder fast enough). For science run the same file in ripbot264 and show screenshot from process hacker just like before.

[poisondeathray]

Originally Posted by Bahamuth

I just stumbled over this:
http://www.chaneru.com/Roku/HLS/X264...gs.htm#threads
"Default: auto (frame based threads: 1.5 * logical processors, rounded down; slice based threads: 1 * logical processors)"

Could it be, that x264 is limiting itself? Should I tamper with this setting?

It should be ok autodetected; You can manually set --threads if you want

Can you switch to avs+ in staxrip ? Worth a try .

Atak is right in that something is bottlenecking in your vpy setup.

[poisondeathray]

Originally Posted by Atak_Snajpera

I look at fps reported by AVSMeter...

Useful tool Atak, but synthethic tests do not necessarily reflect actual encoding performance.

If you look at avsmeter, varying prefetc, sometimes a much higher CPU% results in only a marginally higher FPS with lots of threads . That's probably similar to the "choking" you're referring to or "thrashing" I'm referring to. CPU usage% and FPS do not always relate 1:1 . Thrashing could be a bottleneck for CPU encoder.

[stax76]

Can you switch to avs+ in staxrip ? Worth a try .

By design, you can change any setting at any time (Filters Menu > Filter Setup > AviSynth).

[Bahamuth]

With the old staxrip and avisynth (the setting I used with my old computer) x264 also does stay at around 40% of CPU usage.

With the Ripbot the encoding speed is around same (~140FPS) when setting preset slower. But CPU usage is again not fully utilized.
Please see image here (don't want to spam the thread)

[Bahamuth]

And with AviSynth+ it's not going really better: Image.

Contents of the AviSynth script:

LoadPlugin("C:\apps\video\StaxRip2\Apps\Plugins\Both\FFMS2\ffms2.dll")
FFVideoSource("C:\Users\Ezatoka\Desktop\Arpeggio of Blue Steel Ars Nova DC.m2v", colorspace = "YV12", \
              cachefile = "C:\Users\Ezatoka\Desktop\Arpeggio of Blue Steel Ars Nova DC_temp\Arpeggio of Blue Steel Ars Nova DC.ffindex")
#AssumeFPS(25)

[poisondeathray]

Probably what Atak said about too small resolution and QTGMC bottleneck . You can try 2 simultaneous encodes (or try GPU(CUDA) QTGMC variant if that is the bottleneck).

[Atak_Snajpera]

Originally Posted by Bahamuth

With the old staxrip and avisynth (the setting I used with my old computer) x264 also does stay at around 40% of CPU usage.

With the Ripbot the encoding speed is around same (~140FPS) when setting preset slower. But CPU usage is again not fully utilized.
Please see image here (don't want to spam the thread)

Did you activate deinterlacing? CPU usage by ffmpeg is too low.

[Z'Hadum]

Thank you for your work!

Originally Posted by stax76

Thanks to everybody for testing and help with the updates.

2.0.5.3 Beta
------------

- update: mkvtoolnix 39
- update: mpv.net 5.4.3
- update: NVEnc 4.55

- fix: few UI issues
- fix: avs MCTemporalDenoise/GradFun2DBmod
- fix: VUI luminance issue

https://www.dropbox.com/sh/4ctl2y928...dd3yqcAHa?dl=0

https://1drv.ms/f/s!ArwKS_ZUR01gySQIE1QCL2i9F_uR

Can also confirm that the luminance issue is fixed.
The rest seems also to work for me....

[jlw_4049]

Ready for that new stable build

[stax76]

The stable release 2.0.6.0 is online. Thanks to Patman and everybody else for posting tool updates, bug reports and suggestions.


Changelog:

https://github.com/staxrip/staxrip/b...r/changelog.md


Download:

https://github.com/staxrip/staxrip/releases

[Atlantis]

No change log for 2.0.6.0

[Atlantis]

AviSynth+ has not been updated?

There is AviSynth+ 3.4.0

[Atlantis]

Thanks for updating FFT3DFilter. I find it to be the fastest and best temporal noise reduction. It's more practical.

[stax76]

Originally Posted by Atlantis

AviSynth+ has not been updated?

There is AviSynth+ 3.4.0

The main reason for why it wasn't updated is there is no small enough installer because the only installer available includes vcredist which is more than 20 MB, it's too large to include. I don't know if 3.4.0 is different from 2.7.7.2, it's likely that for staxrip both will work well.

[Atlantis]

It's only 27.2 MB. I think we can handle it with today's internet speeds even in countries with ADSL only.
For comparison, Firefox installer is 50 MB and VLC is 40 MB.

I don't understand the technical details maybe someone with knowledge should contact the AviSynth+ team and ask for a better installer? We don't need vcredist?

[Patman]

Originally Posted by Atlantis

No change log for 2.0.6.0

Here is right changelog