Replaced RTX 2070 with RTX 3080 TI and the encoding takes longe

vlad-gh wrote on 12/22/2021, 11:16 AM

I'm using Vegas Post 18 and my system is:

Win10 x64, i9-10900k, 32GB ram

I replaced the RTX 2070 8gb with a RTX 3080 Ti 12gb and now the render takes longer.

I thought that maybe Vegas 18 wasn't capable on using the RTX 3000 series so I've upgraded to Vegas Post 19.

Now, Vegas 18 renders faster than Vegas 19 (the same project, the same settings).

So there are 2 problems/questions now :)

- why the render with RTX 2070 takes less time than with RTX 3080 Ti in Vegas 18 ??? (using nvenc, of course)

- why the render in Vegas 18 takes less time than in Vegas 19 (same project, same render settings) And it's like 30-35% faster the render with Vegas 18 :(

Please help.

Back to post

Comments

Former user wrote on 12/29/2021, 4:25 PM

Rendering out at MAGIX AVC/AAC MP4 - Internet UHD 2160p 25fps (NVENC), The best i can get is 1:05mins,

Can you go to your icon at the top, click it - My Profile, & fill in your Signature with your Vegas version, Windows version & system specs, CPU, GPU, RAM etc. this will then show at the bottom of the comments & will help people help you 👍

Former user wrote on 12/29/2021, 4:33 PM

Former user wrote on 12/29/2021, 6:19 PM

@vlad-gh I think it would be interesting for you to encode this benchmark project,

https://www.dropbox.com/s/3i4xgnhl9fmc0je/BenchmarkProject4K.zip?dl=0

click download, extract the archive, load the sample project .veg file

Encode using CPU first, encode 4K25 MagixAVC, then encode the same using NVENC . List the encode time for both.

Great idea :)

I used Vegas18 and Vegas19 for the renders and I got the exact same durations in both programs:

CPU - 1m38s

NVenc - 1m00s

They are some of the fastest results I've seen for the 4K25 MagixAVC hardware encode, so for this benchmark your system works fine

https://docs.google.com/spreadsheets/d/1b3ggVifKsuT-cp2kQHjum_KnQ4-2jBmUIvzmu7BQZ34/

RogerS wrote on 12/29/2021, 10:36 PM

@vlad-gh I'm curious, on the Benchmark link the're pics showing render settings,

In the one in the link the bitrate is Max 50,000 Avg 28,000

but on the default now it's Max 48,000 Avg 24,000

presuming you used one of these templates your render times for (NVENC) are quite good,

If you [Vlad] submit results to this benchmarking survey, please do use the standard bitrates for this benchmark rather than the defaults. It's a minor difference so I don't expect you'll see much different times.

Since you're mainly interested in comparing VP 18 and 19 it doesn't matter.

Former user wrote on 12/29/2021, 11:05 PM

@RogerS Standards = Max 50,000 Avg 28,000 ? Default = Max 48,000 Avg 24,000 ?

RogerS wrote on 12/29/2021, 11:22 PM

Benchmark UHD: is 50/28 (default is 48/24).

Former user wrote on 12/29/2021, 11:49 PM

@RogerS I have a template saved where i've changed it to 50/28, that's what I've been using 👍

vlad-gh wrote on 12/31/2021, 8:50 AM

@vlad-gh I think it would be interesting for you to encode this benchmark project,

https://www.dropbox.com/s/3i4xgnhl9fmc0je/BenchmarkProject4K.zip?dl=0

click download, extract the archive, load the sample project .veg file

Encode using CPU first, encode 4K25 MagixAVC, then encode the same using NVENC . List the encode time for both.

I figure you might get the similar results as I do using VP17 which isn't compatible with my 3070 card

CPU - 1m41s

NVENC - 4m43s

I do not have the slow Nvenc problem you have with VP19, but this test may prove that your particular model of 30 series GPU is not compatible with VP19, and a patch is required to make 3080ti's compatible.(If it's not a driver problem)

Hi,

Great idea :)

I used Vegas18 and Vegas19 for the renders and I got the exact same durations in both programs:

CPU - 1m38s

NVenc - 1m00s

Hi,

I've rendered with 50mbps with 28mbps. The same durations:

CPU - 1m37s

NVenc - 00:59s

Former user wrote on 12/31/2021, 6:57 PM

As said that benchmark shows there's nothing wrong with your graphics card or compatibility with VP18 and VP19, which was your initial claim, do you have another project you can upload for others to try and shows this problem you were talking about with slow NVENC encoding?