Disappointing Vegas performance with new AMD RX6600XT

Comments

Reyfox wrote on 7/28/2022, 11:10 AM

@Howard-Vigorita thanks for the tips. I know the 3X performance difference was just rendering a clip with nothing done to it. And as long as I can work around it.

Hoping too that VP20 will work out with less minor bugs. I gave the card back to the owner. I did try the .veg file with the PiP scaling... it worked as it should with my aging RX480 8GB. The same with the other .veg.

I'm keeping those files and will try them when the card gets here and is installed.

 

Last changed by Reyfox on 7/28/2022, 11:22 AM, changed a total of 1 times.

Newbie😁

Vegas Pro 22 (VP18-21 also installed)

Win 11 Pro always updated

AMD Ryzen 9 5950X 16 cores / 32 threads

32GB DDR4 3200

Sapphire RX6700XT 12GB Driver: 25.3.2

Gigabyte X570 Elite Motherboard

Panasonic G9, G7, FZ300

Former user wrote on 7/28/2022, 11:42 AM

@Howard-Vigorita Hi, i just downloaded those 2 .veg files, 'The small vp19 sample projects linked demonstrate the bugs', can you mention what the bugs are?

Thanks.

Howard-Vigorita wrote on 7/28/2022, 12:22 PM

@Former user The pip test shows black border lines on the top and right-side of the pip windows that come and go depending on the scale or rotation. The Track Motion bug blacks-out portions of the background screen to the left of the scaled window that varies with rotation. Only affects amd 5000 and 6000 series gpus so tarck#1 and track#2 should look the same when soloed on a different gpu. Changing the amd 5000 and 6000 series gpu to anything else in Video preferences would also make the bug go away... you could still render vce but would not be using amd processing for the timeline or fx.

Reyfox wrote on 7/28/2022, 12:27 PM

@Howard-Vigorita ahh... now I understand. So you have to lose the GPU for processing and timeline while editing.

Newbie😁

Vegas Pro 22 (VP18-21 also installed)

Win 11 Pro always updated

AMD Ryzen 9 5950X 16 cores / 32 threads

32GB DDR4 3200

Sapphire RX6700XT 12GB Driver: 25.3.2

Gigabyte X570 Elite Motherboard

Panasonic G9, G7, FZ300

Howard-Vigorita wrote on 7/28/2022, 12:30 PM

@Reyfox I don't ever do that. I keep it and use a workaround.

Reyfox wrote on 7/28/2022, 12:38 PM

@Howard-Vigorita then I will have to "hit" you up on what you do! 😉

Newbie😁

Vegas Pro 22 (VP18-21 also installed)

Win 11 Pro always updated

AMD Ryzen 9 5950X 16 cores / 32 threads

32GB DDR4 3200

Sapphire RX6700XT 12GB Driver: 25.3.2

Gigabyte X570 Elite Motherboard

Panasonic G9, G7, FZ300

Former user wrote on 7/28/2022, 1:28 PM

@Howard-Vigorita Thanks for explaining that, I don't have AMD 5000 or 6000 series GPU so that's why i see no prob with those .veg's 👍

Cielspacing wrote on 7/28/2022, 6:07 PM

Those affected might like to check latest AMD software update Adrenalin Edition 22.7.1 that came out this week.
It brings updates/fixes for longstanding issues that could be related.
Release Notes:

Highlights
.Swordsman™ Remake.
.Radeon™ Boost using Variable Rate Shading with Elden Ring™, Resident Evil Village™ and VALORANT™.
.Microsoft® Windows® 11 version 22H2.
.Microsoft® Agility SDK Release 1.602 including new minor features.
.Microsoft® Agility SDK Release 1.606 including Microsoft® Shader Model 6.7.
.Additional Vulkan® extensions.

OpenGL® Optimizations
.Up to 79% increase in performance in Minecraft™ @ 4k Fabulous settings, using Radeon™ Software Adrenalin 22.7.1 on the Radeon™️ RX 6950XT, versus the previous software driver version 22.6.1 RS-491
.Up to 75% increase in performance in Minecraft™ @ 4k Fabulous settings, using Radeon™ Software Adrenalin 22.7.1 on the Radeon™️ RX 6400, versus the previous software driver version 22.6.1 RS-495

Radeon™ Super Resolution
.Expanded support for discrete Radeon™ RX 5000 and 6000 series GPUs on AMD Ryzen™ processor notebooks with hybrid graphics.
.RSR has been improved to provide a more seamless experience in borderless fullscreen mode with a performance/quality slider to personalize your gaming experience.

Fixed Issues
.Lower than expected Folding@home™ compute performance with OpenCL™ API on some AMD Graphics Products such as the Radeon™ RX 6800.
.Auto Undervolt may disable Zero RPM fan feature.
.Hitman 3™ may freeze when rapidly switching between windows in Fullscreen Exclusive mode.
.Video upscaling in browsers appears blurry with some AMD Graphics Products such as the Radeon® RX 6900 XT Graphics.
.Enhanced Sync may cause games to lock to 15FPS with video playback on extended monitors.

 

Vincent-Brice wrote on 7/29/2022, 3:14 PM

@JamBam @Howard-Vigorita @Reyfox Hi JamBam, I was a little disappointed with the performance of my RX 6800 XT until I enabled "Legacy AVC Decoding". Performance is better with this setting. By default I get around 4fps in the Vegas 4k project loop section but it will rise to around 22fps with Legacy enabled. Also the render times are slightly faster. But like yours, my GPU doesn't "break a sweat". I have recorded the Vegas 4k project being rendered with legacy enabled and without and recorded framerates on the loop section with legacy enabled and without. See what you think 😉

Render:

https://1drv.ms/v/s!AiKR8DL4bKWahbhu64_X-E4co16Geg?e=1JjmAl

Loop FPS:

https://1drv.ms/v/s!AiKR8DL4bKWahbhvec5JtOsJUfk4Gw?e=UgAxhi

Ah, SORRY for adding you guys in the comment, I've just realised that the comments that caused me to add you were a year old 😬

 

Last changed by Vincent-Brice on 7/29/2022, 4:59 PM, changed a total of 1 times.

Gigabyte Aorus Pro V2 motherboard, AMD Ryzen 9 5950X, Gigabyte AMD RX 6800XT, 64GB Corsair Vengeance 3600MHz ("Ryzen tuned", whatever that is), 500GB Corsair Force MP600 Gen 4 M.2 C:drive for windows and programs, 500GB Samsung Sata SSD EVO D:drive for video files, 1TB Samsung Sata SSD EVO E:drive for all other data, Gigabyte Aorus Waterforce 360 CPU cooler, Thermaltake Core P5 TG open case, Contour ShuttleXpress.

Windows 11 Pro 24H2, Vegas Pro 22 (Build 248)

Reyfox wrote on 8/6/2022, 9:25 AM

@Reyfox Just be aware that Vegas 19 and earlier running amd series 6000 gpus still have a few minor bugs with pip (op showed it) and track motion scaling. The small vp19 sample projects linked demonstrate the bugs. And my work-arounds if you mute the 1st track. Hopefully they'll be history in vp20 when it gets released.


@Howard-Vigorita, I just installed the RX6700XT and see the bug in the PiP that you mentioned. I guess we will have to keep an eye out when VP20 is released. I didn't see it with my "old" RX480 8GB with the latest drivers 22.5.1, not the 22.7.1.

There is a big bug (at least for me) with B643 in the amount of RAM a project uses compared to VP18 loading in the same project (copy from VP19 and paste in VP18). Not sure when this crept in.

 

Newbie😁

Vegas Pro 22 (VP18-21 also installed)

Win 11 Pro always updated

AMD Ryzen 9 5950X 16 cores / 32 threads

32GB DDR4 3200

Sapphire RX6700XT 12GB Driver: 25.3.2

Gigabyte X570 Elite Motherboard

Panasonic G9, G7, FZ300

Vincent-Brice wrote on 8/6/2022, 1:16 PM

@Reyfox Hey, you got your new GPU! 😀No crashes yet? Does it perform better with Vegas settings on default or with Legacy AVC Decoder enabled? Just when you have a spare minute, have a look at my video clips above and see if you have the same experience on the Loop FPS. As we have very similar PC specs, I'd be really interested to know how you get on 😉

Gigabyte Aorus Pro V2 motherboard, AMD Ryzen 9 5950X, Gigabyte AMD RX 6800XT, 64GB Corsair Vengeance 3600MHz ("Ryzen tuned", whatever that is), 500GB Corsair Force MP600 Gen 4 M.2 C:drive for windows and programs, 500GB Samsung Sata SSD EVO D:drive for video files, 1TB Samsung Sata SSD EVO E:drive for all other data, Gigabyte Aorus Waterforce 360 CPU cooler, Thermaltake Core P5 TG open case, Contour ShuttleXpress.

Windows 11 Pro 24H2, Vegas Pro 22 (Build 248)

Former user wrote on 8/6/2022, 5:23 PM

@Vincent-Brice Hi, don't be too disappointed with those benchmark results, here's a vid of mine, check my specs in Signature & on the benchmark results spreadsheet there's people with the full fps of 25fps, UHD render of 28secs-ish, (can't remember exactly, haven't looked in a while) , either they are lying, not telling the whole truth, they've tweaked Vegas in a way that they're no sharing, Vegas &/or GPU drivers has changed for the worse or it simply doesn't work well with AMD CPU's, the best render time i've had for UHD is 1.04secs, & avg playback of Region 1 is about 13-14fps. that would put me about 13th on that list. ! 😒😒

Vegas seems a little slow today Region 1 looks like 10-12fps, Legacy AVC made render slower

 

 

Former user wrote on 8/6/2022, 6:19 PM

@JamBam @Howard-Vigorita @Reyfox Hi JamBam, I was a little disappointed with the performance of my RX 6800 XT until I enabled "Legacy AVC Decoding". Performance is better with this setting. By default I get around 4fps in the Vegas 4k project loop section but it will rise to around 22fps with Legacy enabled. Also the render times are slightly faster. But like yours, my GPU doesn't "break a sweat".

 

 

@Vincent-Brice Did you set dynamic ram preview to zero for that test?

It's one of the requirements of that benchmark. I have found depending on media setting DRP to zero can reduce playback speed by up to 25% I don't think I noticed that change with this benchmark though, with rtx3080 I get 12-14fps mainly with 17-18fps near end of loop.

Howard-Vigorita wrote on 8/6/2022, 7:14 PM
@Howard-Vigorita, I just installed the RX6700XT and see the bug in the PiP that you mentioned. I guess we will have to keep an eye out when VP20 is released. I didn't see it with my "old" RX480 8GB with the latest drivers 22.5.1, not the 22.7.1.

@Reyfox The 6700xt should be a real step up from the rx480. The pip bug is more an annoyance. The workaround is just to throw a bezier mask behind the pip to get rid of the border. I implemented it in the test project on track#2 which will show in the preview if you mute track #1.

Former user wrote on 8/6/2022, 9:33 PM

@Former user I just ran it again with dynamic ram preview to zero, on Region 1 playback didn't change much, 9-10-12 were the fps numbers mainly, the biggest i saw very briefly was 17,

Former user wrote on 8/6/2022, 11:25 PM

@Former user We're about the same, I probably took my reading after too many passes. Even with Dynamic Ram set to zero there still looks to be some caching. In a playback test I had a 4K60 that could only just hold 60fps, could see very tiny deviations from 60. I set dynamic ram cache to zero, and playback went down to around 45fps.

Luckily that problem isn't really seen here, the creator of benchmark didn't want caching to affect results. I show it happening here with DRP turn on, towards the end it's very obvious, but it may not always be that obvious. You can see the VRAM use go down, decoding turn off and then the GPU processing go very low.

Reyfox wrote on 8/7/2022, 1:34 PM

@JamBam I certainly will give it as shot tomorrow. Went kayaking today... arms a little sore.

Newbie😁

Vegas Pro 22 (VP18-21 also installed)

Win 11 Pro always updated

AMD Ryzen 9 5950X 16 cores / 32 threads

32GB DDR4 3200

Sapphire RX6700XT 12GB Driver: 25.3.2

Gigabyte X570 Elite Motherboard

Panasonic G9, G7, FZ300

Vincent-Brice wrote on 8/7/2022, 1:37 PM

@JamBam @Howard-Vigorita @Reyfox Hi JamBam, I was a little disappointed with the performance of my RX 6800 XT until I enabled "Legacy AVC Decoding". Performance is better with this setting. By default I get around 4fps in the Vegas 4k project loop section but it will rise to around 22fps with Legacy enabled. Also the render times are slightly faster. But like yours, my GPU doesn't "break a sweat".

 

 

@Vincent-Brice Did you set dynamic ram preview to zero for that test?

It's one of the requirements of that benchmark. I have found depending on media setting DRP to zero can reduce playback speed by up to 25% I don't think I noticed that change with this benchmark though, with rtx3080 I get 12-14fps mainly with 17-18fps near end of loop.


@Former user @Former user Ah Todd, no I didn't. I was careful to try and get the settings right but didn't realise DRP had to be set to zero for the test. So rendered it at zero and the render went from 1:13 to 3:11!! Good job I didn't record the 1:13 on the spreadsheet Gid or this would have been true of me: "either they are lying, not telling the whole truth, they've tweaked Vegas in a way that they're no sharing" 😂. But I'm still getting FPS in the upper teens into twenties for the marked region with Legacy AVC Decoder enabled (4-5 FPS without). So there was no cashing (however it's spelt) I started Windows fresh and straight into Vegas to do the test. Just fast forward through the vid of it to check I'm doing it right this time, cheers:

https://1drv.ms/v/s!AiKR8DL4bKWahbkCV7bedxeBIGK6hA?e=TIAwKG

 

Last changed by Vincent-Brice on 8/7/2022, 1:41 PM, changed a total of 1 times.

Gigabyte Aorus Pro V2 motherboard, AMD Ryzen 9 5950X, Gigabyte AMD RX 6800XT, 64GB Corsair Vengeance 3600MHz ("Ryzen tuned", whatever that is), 500GB Corsair Force MP600 Gen 4 M.2 C:drive for windows and programs, 500GB Samsung Sata SSD EVO D:drive for video files, 1TB Samsung Sata SSD EVO E:drive for all other data, Gigabyte Aorus Waterforce 360 CPU cooler, Thermaltake Core P5 TG open case, Contour ShuttleXpress.

Windows 11 Pro 24H2, Vegas Pro 22 (Build 248)

Former user wrote on 8/7/2022, 4:55 PM

@Reyfox Hows your new GPU. after reading this & looking at the benchmark results, for a long time i've wondered about whether i got the right GPU, I have considered getting the RTX4000 when it comes out but i might just go for an RX

So so tempted even for my own curiousity

RogerS wrote on 8/7/2022, 10:08 PM

Hi JamBam, I was a little disappointed with the performance of my RX 6800 XT until I enabled "Legacy AVC Decoding". Performance is better with this setting. By default I get around 4fps in the Vegas 4k project loop section but it will rise to around 22fps with Legacy enabled. Also the render times are slightly faster. But like yours, my GPU doesn't "break a sweat".

 

 

@Vincent-Brice Did you set dynamic ram preview to zero for that test?

It's one of the requirements of that benchmark. I have found depending on media setting DRP to zero can reduce playback speed by up to 25% I don't think I noticed that change with this benchmark though, with rtx3080 I get 12-14fps mainly with 17-18fps near end of loop.


Ah Todd, no I didn't. I was careful to try and get the settings right but didn't realise DRP had to be set to zero for the test. So rendered it at zero and the render went from 1:13 to 3:11!! Good job I didn't record the 1:13 on the spreadsheet Gid or this would have been true of me: "either they are lying, not telling the whole truth, they've tweaked Vegas in a way that they're no sharing" 😂. But I'm still getting FPS in the upper teens into twenties for the marked region with Legacy AVC Decoder enabled (4-5 FPS without). So there was no cashing (however it's spelt) I started Windows fresh and straight into Vegas to do the test. Just fast forward through the vid of it to check I'm doing it right this time, cheers:

https://1drv.ms/v/s!AiKR8DL4bKWahbkCV7bedxeBIGK6hA?e=TIAwKG

 

Just to clarify, setting dynamic ram preview to 0 was recommended only for the fps preview part of the benchmark as that can be affected by caching. This part is hard to do well as values can fluctuate a lot so just pick a median number.

For rendering, use whatever you normally use for dynamic ram preview (200MB, 5%, 10%, etc.) and there is a field to mark the amount. If you would like to run a test of DRP and try 0 and non zero values you are welcome to report both.

In the past there was no standalone column for drp so many people reported their value in the comments field. I added one for dynamic ram preview as I thought it was important and manually copied some of the values into it.

Reyfox wrote on 8/8/2022, 5:33 AM

@Former user and @Vincent-Brice, today I ran a quick test on a project that I did a couple of months ago.

Backstory. I beta tested for the US Army's gaming studio for the last 12 years. Some, classified, but what the US Army did in 2002 was release an online multi-player game free to the public, called America's Army. I did a trilogy of sorts, 3 videos with my making fun of the other beta testers and developers as I destroy them. Yes, I talk smack. The final video, I had to rush because the gaming studio was closing shop and ending America's Army: Proving Grounds development.

In the video I used NBFX Titler Pro 7 a lot. BorisFX Title Studio. Vegas titling software. BorisFX many transitions, GL transitions, NBFX transitions. Compositing a lot within Vegas using .mov files, Ignite Pro 5, BorisFX effects, and just about anything "stupid" I could add, including music from Smartsound, and yes, some copyright stuff, but I didn't care about monetization. The video was 19 tracks.

Before installing, I rendered it with the RX480 installed. Render time was 13:42 and 33.67fps average. With the RX6700 XT, 12:35 and 42.12fps. Mind you, a lot of the time, neither the CPU or GPU was being used much with effects, overlays, and transitions, which there were a lot. But one thing I did notice. When it was pour video being rendered with nothing on it, with the RX480, top fps was around 100. With the 6700xt, 250fps. A 3x increase in speed.

I will run the benchmark later today and report my findings. @Former user, @Howard-Vigorita can tell you a lot about the AMD cards since he uses them.

@JamBam, I monitored the performance of the 6700xt, watching the GPU clock speed, etc. with MSI Afterburner and AMD. I did not overclock the GPU. But what I did notice while watching the Vegas Render Process window is how much faster the 6700xt was compared to the RX480 8GB overclocked card on straight video rendering. And while the 480 would be at 100% max at times (according to Task Manager), the 6700xt was loafing along at a max of 20%. CPU usage looked similar. Everything else I added in the video didn't make any real use of the hardware I have listed in my signature below. Is the cost of the "upgrade" worth it? Yes and no, for what I do. My RX480 has seen extreme heavy duty use for years. I didn't want to take a change of failure and not being able to get a replacement. So yes. No, because saying a few minutes is no big deal to me compared to the cost.

 

Howard-Vigorita wrote on 8/8/2022, 10:13 AM

@Reyfox My first step up from an rx480 was an rx580 and found even that made a significant difference to editing in Vegas. At the time I was shooting primarily with an Canon 3ccd xf305 which records 8-bit HD 4:2:2 color in an mxf format that cannot be decoded by gpu. But the improved gpu performance for everything else freed up the cpu enough to make everything better. Btw, had allot of fun working on an est once upon the turn of a century which I imagine was right up your alley.

Reyfox wrote on 8/8/2022, 10:33 AM

@Howard-Vigorita, for sure, things are much better. It's when you add "stuff" to the videos and whether that "stuff" is optimized for the hardware.

Back in the "day", the process of editing was exciting. It was new to me and amazing putting videos together. But my mind was quicker and patience a lot longer. Glad when I went NLE and no more tape!

Newbie😁

Vegas Pro 22 (VP18-21 also installed)

Win 11 Pro always updated

AMD Ryzen 9 5950X 16 cores / 32 threads

32GB DDR4 3200

Sapphire RX6700XT 12GB Driver: 25.3.2

Gigabyte X570 Elite Motherboard

Panasonic G9, G7, FZ300

fr0sty wrote on 8/8/2022, 2:30 PM

@Reyfox Hows your new GPU. after reading this & looking at the benchmark results, for a long time i've wondered about whether i got the right GPU, I have considered getting the RTX4000 when it comes out but i might just go for an RX

So so tempted even for my own curiousity

Having come from dual Radeon VII's to a RTX 3090, I don't regret making the switch, at all.

Systems:

Desktop

AMD Ryzen 7 1800x 8 core 16 thread at stock speed

64GB 3000mhz DDR4

Geforce RTX 3090

Windows 10

Laptop:

ASUS Zenbook Pro Duo 32GB (9980HK CPU, RTX 2060 GPU, dual 4K touch screens, main one OLED HDR)