@Mindmatter

Sony AVC is not a GPU-supported renderer. Try one of the Nvenc templates using MagicAVC or MagicHEVC. You could also try an Nvenc template in Voukoder or HOS.

Hi all

 

I don't understand why Vegas is not using the full power and possibilities of my new PC. I wonder... what's the point in investing in a 12core CPU and an Nvidia 3070?

I was watching a Intel 12600K review, and they were talking about who should buy it vs AMD equivalents or earlier generation Intel's in same price range. One of the things he said that people who run old software or poorly optimized software will see gains.

That's probably the answer, underneath all the add-ons and improvements and features it's very old software initially built without GPU support, and GPU decoder support, when people were using 2-6 core cpu's.

Some of the old users swear best config isTurn OFF GPU entirely, and decode OFF for trouble free use during editing, turn on again when encoding(or keep off if it crashes). Should see less crashes and your cpu will more regularly run at 100% when GPU is OFF. It is as if it's because the software was not purpose built to run with a GPU, a GPU decoder, and CPU's with a lot of cores it has efficiency and stability problems.

If your main issue is playback performance I am hoping a 12600K will make a difference, overclocked if need be because it can do more work per cycle your CPU may be able to deliver 20-30 percent more power to Vegas which will increase GPU use and reduce render time, . Have to see how that turns out.

People were saying the same about 11900K, I think it's faster in stats/benchmarks, but doesn't really improve Vegas as hoped, but it was a very poorly received chip in comparison to Alder Lake CPU's.

If you want performance stop using old encoder options. Magic AVC supports GPUs.

Give frameserving to Voukoder a try too for more even CPU and GPU loading.

Some Fx will prevent full hardware usage too, FYI, as they're just not optimized.

Sony AVC is not a GPU-supported renderer.

Also, AVC is optimal at 6-8 cores. Distributing that among 12 cores won't make it faster.

Also, you need to take a close look at what filters / effects are in the chain. Your render can only be as fast as your slowest bottleneck. But you know that.

I suggest you run your tests with machine-encoded HEVC and GPU-accelerated filters, then begin to draw conclusions.

@Mindmatter I understand your frustration as I came here to search that question.. again. I really do like Vegas & I've tried to work this out but to no avail.

@walter-i. Not helpful, why answer so sarcastically & without adding something useful or a link to a relevant post?.

This isn't a rant, just an observation, i just thought i'd share 😉

Thankyou to everyone else for your input, I've tried several formats & i can't turn GPU off as i have no internal GPU, often Vegas barely scratches the surface of the available power, although playback & rendering out a normal project of just edited/cut clips with Magix AVC/AAC (NVENC) is fairly quick, if it has effects added that can slow things down dramatically.

Here i have a 1.44min clip, exported from Vegas @ Magix AVC/AAC (NVENC), then re-imported, the properties of the project are exactly the same, I sometimes use the Magix option without (NVENC) at the end when there's jpgs/pngs or text in the timeline, it gives a more reliable export, it uses a lot more CPU but is a little slower to render out.

I have BCC S_EdgeRays with the fx's built-in Mocha creating a mask around the sq screen to limit where the rays project from in the screen capture clip on the timeline ,

I know Mocha is heavy on usage (whatever the right word is)

I wanted it to play at Good - Full on the preview to get the true effect because Good - Auto blurs the image a bit,

but it wouldn't play so i tried Prerender & Build Dynamic Ram, both were extremely slow & used very little of my PC's potential. & took 40mins-ish to complete the Prerender

So i decided to just Render out the 1.44min project,

1hr it estimated!! i got bored after 15mins of watching it crawl along

I tried it in HitFilm Pro,playback & export no better, & i couldn't get the Mocha that's built into the S_EdgeRays effect to load, it just came up with a blank screen, i had to add the other HitFilm-Mocha to mask the area, then apply another track of my clip because that Mocha masked out the whole clip not just the rays effect,, there's plenty of ways to mask an area but i wanted to try replicate what i had in Vegas,

I tried it on Free! DaVinci Resolve,

S_EdgeRays played well, the inbuilt Mocha loaded no prob to isolate the effect, Resolve has a Smart Cache, which prerenders the timeline automatically while it's playing or while you're away from the keyboard/screen, the cache took a little time but way way faster than Vegas, about 5mins vs 40mins. although it didn't use a huge amount of CPU, RAM etc. It exported the clip in 8.32mins even tho i mistakenly set it at 60fps not 29.97fps.

I thought i'd have another go with a different clip, pretty much the same results

Vegas estimated to render the clip in 55mins, i declined that offer,

I skipped Hitfilm

& DaVinci exported it in 9.44mins.

Both paid-for software's just got beaten by a Free one. 😏

Walter's reply wasn't sarcastic and this topic has been covered extensively.

There is nothing you can do at a user level to change how Vegas works beyond the obvious of not using a CPU-only render template.

What am I missing here?

The search function.
 

Sarcasm!

So you admit that not even Resolve maxes out your CPU and GPU, but you still insist on using that as a metric for telling if you are getting the most out of your system? Some tasks get slowed down by things other than CPU and GPU utilization... computers are more complex than that, bottlenecks can occur anywhere along the chain.

We've beaten that horse to death, as have we the "VEGAS vs. Resolve" debate. If you have questions about which codecs render fastest, or ways to improve render speeds, we'd be glad to help.

I'm sure we all appreciate all the help given 👍,

but I wasn't emphasising the 'maxing' of my CPU or 'as a metric for telling if you are getting the most out of your system', I think you said it yourself 'computers are more complex than that' & i don't think Mindmatter would be bothered about how much his CPU or GPU were maxing out but he stated 'renders are unbelievably slow'

I was just showing that -

Vegas prerendered at 40mins, Resolve 'cache'd' in 5mins,

Vegas rendered out in 1hr, Resolve exported in 8mins,

If renders were fast & CPU/GPU use were low there'd be no prob,

Yes i stated 'although it didn't use a huge amount of CPU, RAM etc. It exported the clip in 8.32mins', I don't know how you can twist that honesty, you're being very defensive & in a little bit of denial, reading into my post what you want.

 

Vegas rendered out in 1hr, Resolve exported in 8mins,

If renders were fast & CPU/GPU use were low there'd be no prob,

Yes i stated 'although it didn't use a huge amount of CPU, RAM etc. It exported the clip in 8.32mins',

I actually found when you insert a sapphire filter into the render process Resolve and Vegas are about the same in render speed, any advantage Resolve has in encoding is lost due to the delay caused by the filter. I used S_sharpen on it's default setting (best) because I knew it uses more GPU then the couple I've seen you use, so I predicted Resolve may have an advantage in render speed but did not. This is a 1080p29.97 project encoding to 1080P AVC 1minute in length

VP18 1:22 MagixAVC (software encode)
VP18 1:18 x264 via Voukoder (software encode)

Resolve 1:20 h.264 AVC (software encode)
Resolve 1:25 x264 via Voukoder (software encode)

You might think but how is this possible, the GPU graphs don't correlate but Vegas is using 2 engines in parallel while Resolve uses the one

Vegas

Resolve

Thanks for the input everyone. Sorry for the late night frustration-tainted post yesterday.
I admit I've not been up to date as far as the different and new render issues and procedures go, as up to now, it has never really been a problem, as my jobs never were as time sensitive as the ones I'm working on atm.
I've now caught up a bit and as I write , I'm rendering using a Magix NVENC template. CPU 19%, GPU up to 30% at times. That feels better.

Btw I should mention I did not use mocha to track and mask in the example shown above, which is why it seems so fast compared to your results. I redid same project with tracking masking and see a ungodly slow down in both editors.it is the filter to blame as I said in your last example of sapphire working poorly in vegas, , neither editor is barely using any resources

S_sharpen tracked and masked mocha VP18 3:56

S_sharpen tracked and masked mocha Resolve 3:45

Close enough timings to call it a draw, and unacceptable performance from both. I blame the filter. BCC Noise reduction I noticed doesn't even use GPU. The Boris OFX filters are great for their effects but not so great in design and use.

What am I missing here?

The search function.
This topic has already been dealt with very often and in detail.

@walter-i. Not helpful, why answer so sarcastically & without adding something useful or a link to a relevant post?.

That was not meant sarcastically, but informatively.
Several other insiders have confirmed this.

@fr0sty said:

Some tas@Former userks get slowed down by things other than CPU and GPU utilization... computers are more complex than that, bottlenecks can occur anywhere along the chain.

We've beaten that horse to death, as have we the "VEGAS vs. Resolve" debate. If you have questions about which codecs render fastest, or ways to improve render speeds, we'd be glad to help.

 

@fr0sty I don't think you should be intimidating users into not talking about Vegas being slow, or slower then Resolve, deciding the Vegas user with one of the most expensive and powerful systems on this forum shouldn't be posting about his problems.(Not that he's more worthy of a response, but I can see how he would be more annoyed if he built it for Vegas)

In the end there was resolution, it wasn't a beaten horse, I found Sapphire plugins are very slow when used with masking and are equally slow on both editors , it's not a Vegas fault, even though there is a slow masking fault with vegas, but this is a plugin slowdown. At least these are my finding at 1080P with my system, @Former user 's results are different, he's working in 4K his system is more powerful, maybe that's the reason for the difference, but looking at the pictures he posted, he appears to have masked almost the entire screen with Vegas, but a very small area with Resolve. Does that smaller area take shorter to compute then the larger area?, no idea

TLDR: Boris plugins are slow with masking applied, not the fault of editors unless Gid would like to refute that. If he did everything identically and Vegas rendered in 1hour while Resolve took 8mins that's worth getting to the bottom of. In my testing with Sapphire plugins on Vegas and Resolve behave almost identically in render speed, and Premiere is faster

but looking at the pictures he posted, he appears to have masked almost the entire screen with Vegas, but a very small area with Resolve. Does that smaller area take shorter to compute then the larger area?, no idea

@Former user Hi & thanks, in the first 4 pics, the video i was using was a screen capture of Resolve in the preview,

In both Resolve & Vegas it was exactly the same, I masked just the preview screen part of the capture, a bit confusing i know because it's a capture of Vegas & Resolve with the same capture of Resolve on the timeline in both 🙃

In Mocha it looked like this, it was S_EdgeRays, in that fx you have the option to track the centre of the rays, that's the red spline around the bolt holding the circular blade on, then turn on 'Center Uses Mocha' in the controls in Vegas, the rays then project from that point, the yellow spline is what i masked so only the rays came from the capture preview screen, so twice as much tracking info as maybe you did.

& thanks, yep i mentioned i knew Mocha uses power (it slows the PC i meant),

& ref to Resolve when i said 'although it didn't use a huge amount of CPU, RAM etc. It exported the clip in 8.32mins' surely Frosty is twisting things & being argumentative with the reply 'So you admit...',

The fact it didn't max my PC is good, it used a similar amount of the potential to Vegas yet delivered the export in a much quicker time is something worth noting,

I like Vegas & would like it to be quicker, you ma have beaten that horse to death but everything I've done & read hasn't made it quicker, a beaten to death horse doesn't move faster, if anybody can suggest an improvement i would be more than happy to try it out,

@Former user Not sure if it did it correctly as it appears I ended up with 3 trackers instead of 2, however the identical process was used in each editor. 1m of 1080P29.97 AVC video took this long to render

S_EdgeRays tracked and masked mocha VP18 12:20

S_EdgeRays tracked and masked mocha Premiere 7:38

S_EdgeRays tracked and masked mocha Resolve 5:14

So with additional trackers there is a difference, a single tracker VP18 & Resolve the same, and Premiere maybe 20% faster

@Former user Hi, interesting, you have 12.20, i'm not sure my PC is set up right .

I created & rendered out 2 of the same 1min clip 29.97 AVC as you wrote above,

1 @ 1080p 29.97 & 1 @ 2160 x 29.97.

I was going to compare the two but have only done the 1080 because -

I imported the 1080 one first, I tracked the centre of EdgeRays & created a random mask

rendering out 14mins it estimates, but i notice on your pic further up labelled Vegas you have Cuda showing, mine just says Copy,

I know very little about Cuda? just that it says it's turned on in the Nvidia control panel ?🤷‍♂️

I found the Cuda but nothing to show 🤷‍♂️

I think I have mislead you. I'm not very familiar with mocha within BCC filters, I added 2 tracking points, but there were 3 trackers show, although there was a link between 1st and 3rd tracker, as each of these trackers has their gear icon activated I thought I was making 3x tracks, but probably I was just doing 2, the 3rd tracker being how far mocha should search around target not actually tracking?

So nothing wrong with your setup if that was the situation. The cuda activity is most likely so low on a 3090 when when you use multiple trackers due to how long it takes nothing will show. The cuda activity shown on my 3070 was not using a tracker and using S_sharpen, it was before I realized the tracker component made a big difference especially 2 or more. If you want to see the most cuda activity add Vegas Denoise and render