@Howard-Vigorita 

Indeed this might be the case. At the Preferences>Video>tab the system sees my XFX Radeon RX 6800XT card as AMD (gfx 1030).

i/O sees the AMD UVD/VCN in the HW decoder box.

will run the Prime95. thanks for the suggestion

Based on the card appearing in both places and the fact you get better times doing VCE renders I think it's safe to say Vegas is recognizing the card.

Why it and the CPU aren't performing better is the question.

I ran stres tests with prime95 with no errors and excellent performance.

Here is the very distintive CPU usage pattern - reproducing during the Mainconcept AVC rendering of i9 10980XE + RX 6800XT

And here is the very distintive CPU usage pattern - reproducing during the VCE rendering of i9 10980XE + RX 6800XT

Glad to hear your big navi is at least recognized and that the rest of your system is stable ... I think you're the 1st here to report on trying a 6000-series gpu with Vegas. I get the impression it's not being well utilized, however. Might be your source footage but its disturbing that your i9 is yielding lower performance than an older i7. Might just need better cooling.

Just looked up your mobo on asus and it doesn't seem to have hdmi connectors so I assume no Intel igpu to split off the decoding load. I have an older x299 chipset asus motherboard with a little navi (5700xt) and I found adding an Nvidia 1660 for decoding sped things up a bit for me. That depends on what kind of footage I'm decoding, however. Not all formats are supported by gpu decoders. I get the best performance from 4K 4:2:0 10-bit hevc footage (mp4 or mov) from a z-cam e2. Little more of a struggle with HD 4:2:2 footage (mxf) from a Canon. Looking at the Task manager gpu charts might give you a better idea of what's going on there. Also, if you put your test clip and project up somewhere, some folks here might try it out for you.

@MP- Just noticed your post in the Benchmarking thread of your results running the Sample Project benchmark. You might want to compare what I get running that same benchmark on a 5700xt which is the little navi, little brother to your 6800xt. In an older x299 system with a lower performance Xeon cpu as well as in a higher performance i9 9900k machine.

See: http://www.rtpress.com/roundup2020.htm

@MP- Just noticed your post in the Benchmarking thread of your results running the Sample Project benchmark. You might want to compare what I get running that same benchmark on a 5700xt which is the little navi, little brother to your 6800xt. In an older x299 system with a lower performance Xeon cpu as well as in a higher performance i9 9900k machine.

See: http://www.rtpress.com/roundup2020.htm

I know... I compared and discussed with other community members in that thread. No clue still..

The problem is that rendering is much slower even compared to my previous older CPU - i7 5820K and E5 2699CV4.

The footage I render is MXF from Canon XC10, but I achieved much better times before with inferior CPUs

I tried all possible settings in the mobo.

Since the CPU does manage to reach consistently 100% of usage in the last seconds of the rendering (in the very first seconds, and then after 25% of the clip and until completion), the issue must be in the Vegas use of the CPU - eventually in combination with the new big Navi - and I wait for the Magix support to provide some insights and solutions.

Will inform of course this post.

Thanks again!

I tried also separate parts rendering of the benchmarking project and I realised that in different sections of rendering the CPU usage is very different - corresponding to the different stages of the CPU usage pattern (pics I uploaded above). However, I don't see any particular differencies in the project FX.

03.18 (Work fast appears) to 07.15 (NEW, Dynamic... apepars) --> CPU usage drops to 5-7% with spikes pattern)

07.15 - end --> CPU goes to 86% until the end.

No matter at which point I start the rendering I get this CPU usage in different parts of the project.

Can this be an indication of specific issues with rendering elements? I cannot see very different FX applied. In fact, more FX are applied in the 2nd very fast rendering part.

Turn off Hardware Decoding Under File IO. That should max out your CPU.

Under preferences/video, do you have two of the same cards listed there? If you switch from one to the other does performance changed at all? That's the oddest thing I see with your screenshots.

Beyond that it's possible Vegas just doesn't scale well across so many cores or take advantage of the GPU power either.

One other thing to try is to use the Vegas to send frames to Voukoder for encoding. Its x.264 implementation may make better use of your CPU (well, worth a try). At low bitrates it still give very good looking files. https://www.voukoder.org/

Install both the program and connector. It's freeware with a donation if you like it. I use it extensively.

@MP- “Tried VCE. Renders same or slower.”

As I mentioned in the Benchmarking thread, its odd that you are not getting better HW render times from your VCE HW encoding than your CPU encoding. The large up and down spikes also occur with Nvidia Nvenc encoding, with some stop start, but nvenc encoding is usually multiples of cpu encoding.

The only access I have to amd is a mates, underpowered laptop. It may be more an indication of the slow cpu rather than the asic, but when I compared render times on a sample avc clip, I get about a 13x speed improvement using his vce 3.1 HW encoding vs his CPU encoding.

How about this, instead of using the Benchmarking project just use an avc clip straight out of camera test clip, no FX or anything else added to the timeline.

Then render this clip out using CPU and VCE to FHD. Use the same FHD render template as for the benchmarking project, but make sure the fps matches the source clips fps. Then let us know here what the results are.

Post here the sample clips duration, the CPU and VCE render times. Perhaps make available the sample clip for download, then other users, especially those with an AMD HW encoding capability can also test it and give you feedback. This eliminates the FX etc in the Benchmarking thread project.

 

How about this, instead of using the Benchmarking project just use an avc clip straight out of camera test clip, no FX or anything else added to the timeline.

Then render this clip out using CPU and VCE to FHD. Use the same FHD render template as for the benchmarking project, but make sure the fps matches the source clips fps. Then let us know here what the results are.

Post here the sample clips duration, the CPU and VCE render times. Perhaps make available the sample clip for download, then other users, especially those with an AMD HW encoding capability can also test it and give you feedback. This eliminates the FX etc in the Benchmarking thread project.

Do this, then disengage the GPU, by going to "GPU processing of video" and choose OFF, and restart, Now do the same 2 renders again. If your CPU is playing fine with Vegas, turning off GPU processing should make both your CPU and VCE render faster and use a lot more of your cpu. If this happens we know the problem is with how Vegas is using your GPU, and we blame Magix for Vegas not having full compatibility with your card, or we blame AMD for once again having terrible drivers for a new gpu launch. My preference would be for you to do a straight transcode, no 4K to 1080P conversion as scaling of this type is done much more efficiently by a GPU, if you engage a sick GPU you get slowness, if you demand a CPU do a task that a GPU is much more efficient at you get a slow down.

* I do not have an AMD card but with Nvidia cards you can turn off gpu processing, but still have your gpu decoding work as normal and also your hardware encoding, so I hope that is true as what we will then achieve is disengaging GPU processing from the rendering engine but not changing anything else. If you don't get the expected results then try disengaging the gpu decoder from the render engine. You are looking for a bottleneck

I rendered the becnhmarking project in UHD

- without all FX and add ons (muted), just the clip. I got 2.58 (instead of 3.01 at AVC and 2.31 in VCE rendering)

- with GPU processing off (decoder off in i/o tab)

I also tested a previous test of mine I cannot share - and got 7 min compared to under 5 mins with my older i7 5820K...

This shows that I might face a problem with the CPU or the mobo and power?

Later in the weekend I will test the i9 9980XE and if not resolved, with another mobo.

Here you can see that during VCE rendering the benchmarking project, the CPU and the GPU card show exactly the same pattern of usage. So the same issue affects both the CPU and the card

Also the GPU is around o until the end of the project, when it reaches 83+% of GPU usage. Below screenshots from the task manager (left) and the AMD Radeon monitor screen (right)

I rendered the becnhmarking project in UHD

- without all FX and add ons (muted), just the clip. I got 2.58 (instead of 3.01 at AVC and 2.31 in VCE rendering)

- with GPU processing off (decoder off in i/o tab)

I also tested a previous test of mine I cannot share - and got 7 min compared to under 5 mins with my older i7 5820K...

This shows that I might face a problem with the CPU or the mobo and power?

Later in the weekend I will test the i9 9980XE and if not resolved, with another mobo.

 

I don't have a fast high core cpu like you, so I am only using information sourced from other vegas users here with cpu such as Ryzen3900x/3950x. Normally turning off GPU allows the highest utilisation of CPU and much faster encode with a straight transcode. But you are not seeing that. My spidey senses felt sure it was the GPU that was causing the slowdown, but did not get the performance increase expected with turning GPU off.

You don't have a cpu problem with benchmarks or other games/programs, but have the slowdown with vegas. You have reset vegas. As I don't have first hand experience with 18 core cpu's only my 6core I can't think of anything else for you to try

the GPU is around o until the end of the project, when it reaches 83+% of GPU usage. Below screenshots from the task manager (left) and the AMD Radeon monitor screen (right)

That is another reason why I thought it could be GPU related. What is happening at the end, I guessed could be the result of creating a progressive download encode. The indexing meta data is moved from the end of the file to the front of the file, and require a 2nd pass. This is possibly why you're seeing the maxing out of your CPU, it no longer has to engage the gpu and any latency that could introduce. The GPU use would be higher as it is scanning your file so decode should be high and encode at it's highest in VCE mode when it's writing the indexes

1. I checked the benchmarking project without any FX and it rendered much faster (in 1.07min) without spikes

2. I also rendered other footage straight from camera, no FX.

It does render faster (~30fps steady), although in AVC (with GPU off) CPU usage is still rather limited = ~50% however without spikes or any fluctuations (I saw another post on this foum about 50% maximum CPU usage)

VCE renders at half the time but the card gives about 6% of usage in the relvant monitors - might be the issue mentioned above about a new amd card and its drivers.

So as I understand it it might be the (recntly launched) 6800XT card that - without proper drivers - might function as a 'brake' to the rendering perfomance. But still the max 50% CPU usage when GPU is off and no FX are present in the project, is still an issue/problem.

Just out of curiosity, did the % utilization of GPU and CPU change now that you're seeing better render times?

Curious as well.

Just out of curiosity, did the % utilization of GPU and CPU change now that you're seeing better render times?

Very good point. Indeed, while there are no spikes and no visible bottlenecks any more, the CPU usage is steady and smoothly running rendering UHD at 20-30% only! (40-50% when rendering without FX and at FHD)

Similarly the GPU usage does not seem to go above 20%.

I would wait for further test (see below) and for a response to my support tickets to Magix, and to AMD (for the 6800XT in this sense, regarding possible better drivers more compatible with Vegas)

Regarding the issue and its unexpected resolution:

Apparently, the i9 10980XE when in heavy load requires a lot of power (from 165 goes to 345W I think) and the motherboard prior to flashing was not able to provide it smoothly - there go the spikes.

I suspect, the figures can be even better with another more power distribution efficient MB like the Asus WS X299 Sage II CEB, which I will be trying in the next days.

Any other recommendation is very welcome