RAM and VEGAS Pro 21?

Matthias-Claflin wrote on 2/1/2024, 11:53 AM

I have been considering returning to video editing, previously used VEGAS Pro as a videographer and the last time I upgraded was VEGAS Pro 16. I left videography to pursue another career path but want to get back into it as a hobby. After going over the google sheet, Vegas Pro Benchmarking Response, I still have a few questions I was hoping people could answer around here.

First, as of now my plan is to do a Ryzen 7 7700X build with an Radeon 6700XT. I do not currently have a PC so I'm starting from scratch. I intend to upgrade to VEGAS Pro 21 and use both the VEGAS Pro NLE and the included effects software (for some light motion graphics). I will be editing mostly HD footage (I have no intention of upgrading my camera, Canon C100). My previous build (which I sold a few years ago) was a Ryzen 7 1700, RX580 and 32GB of RAM.

My big question is, how much RAM will VEGAS actually make use of? I intended to buy 2x32GB but those who report their RAM on the spreadsheet have 32GB total. Will I benefit from 64GB of RAM when using VEGAS?

I'm happy to take any other advice you guys have on PC building while you're here!

Comments

RogerS wrote on 2/1/2024, 7:17 PM

It will make use of the ram you buy. My laptop has 32GB and VEGAS often uses 20+ not leaving a ton for the browser and other purposes. My desktop is 64GB and always has a comfortable amount of ram. Personally I'd buy 64 as the price isn't so high these days and you won't have to think about it for years.

You can see the PC I built in my signature and here: https://pcpartpicker.com/b/rZ9NnQ

I went with Intel mainly for the benefits of its decoding in VEGAS. Though for C100 HD files it may not really matter at all.

Matthias-Claflin wrote on 2/1/2024, 8:01 PM

Thanks! I was worried it might be overkill. Good to know it isn't. I was thinking about the Intel/AMD difference. I haven't followed computer tech in a while, would there be significant benefit to going intel? The spreadsheet didn't seem to show much difference between them when accounting for similar GPUs so I planned to go with AMD because that's what I know. Thoughts?

RogerS wrote on 2/1/2024, 8:17 PM

The benefit with Intel CPUs is that many models include integrated GPUs with QSV decoding which supports more file types and is more stable in VEGAS than AMD/NVIDIA.

Especially if others give you footage like HEVC it works the best. (newer cameras have formats like log footage encoded as HEVC 10 bit 4:2:2 which only decodes in hardware with Intel ARC or iGPUs. 10-bit 4:2:0 does decode with NVIDIA or AMD hardware though in VEGAS you can't beat Intel QSV decoding at present).

The two benchmarks in my signature use easy to decode AVC media and are loaded with GPU-intensive Fx so the bottleneck is not the decoding.

Up to you what you'll use this system for over its life but I build my computer based on what I saw on this forum and in the benchmarks and would build it the same way today.

john_dennis wrote on 2/1/2024, 8:20 PM

Executive Summary

https://www.vegascreativesoftware.info/us/forum/system-upgrade-2023--137456/?page=3#ca892625

All The Minutia

https://www.vegascreativesoftware.info/us/forum/system-upgrade-2023--137456/?page=1

Matthias-Claflin wrote on 2/2/2024, 3:51 PM

After all your help here guys, I made the purchase today.

CPU: Intel i7 14700k
GPU: MSI Ventus GeForce RTX 4060 Ti 16GB
RAM: G.SKILL Trident Z5 RGB Series 64GB (2 x 32GB) 288-Pin PC RAM DDR5 5600
MB: ASUS ROG STRIX Z790-H Gaming 
SSD 1: Corsair MP600 PRO NM 500GB
SSD 2: Corsair MP600 PRO XT 2TB

I'm really looking forward to seeing what kind of performance I can get given the specs on my previous PC and my current laptop. Thanks for all the help!

RogerS wrote on 2/2/2024, 9:12 PM

Looks like a great system and should serve you very well in VEGAS. When it comes try the benchmarks in my signature and see what times you get. I also did render tests (Mainconcept then NVENC) to watch for throttling with HWInfo64 and used the info I saw to reduce voltages/wattages in bios as the MB defaults were not reasonable.

Matthias-Claflin wrote on 2/3/2024, 8:58 AM

My last box is due to arrive Thursday. Should have it up and ready for benchmarks by Friday night. Definitely plan on fiddling with it before benchmarking it. Thanks for all the help!

Howard-Vigorita wrote on 2/3/2024, 10:23 AM

@Matthias-Claflin Here's a Vegas slide show I did from a facebook post during my last build a couple years ago that you might find helpful. Hope you keep us posted because I expect to be building a new system similar to yours later this summer.

fr0sty wrote on 2/3/2024, 7:44 PM

VEGAS Pro, you won't notice a huge difference from 32 to 64GB. VEGAS Effects is a different story.

Matthias-Claflin wrote on 2/3/2024, 9:02 PM

Thanks, @fr0sty! I will definitely be using VEGAS Effects. I used to use Hitfilm Pro 13 (I think it was 13?) after I switched from the Adobe suite to VEGAS. I liked it alright and the integration was pretty good so I'm hoping VEGAS Effects can adequately replace Hitfilm in the workflow completely!

Matthias-Claflin wrote on 2/10/2024, 11:35 AM

So I'm back! I have built the PC (exact specs in signature). I am happy with it so far for the two games I play. I played around my with C100 footage I need to edit with both VEGAS Pro 21 and VEGAS Effects and it works a treat. I did run the 4k benchmark from @RogerS's signature. Was quite disappointed with my results to be honest, in terms of comparing it to others and honestly haven't replicated my best render time of 01m:01s since that test (rendered in QVC rather than NV or MainConcept). Most of the time it renders in UHD at about 01m:27-33s. Honestly this is fine and doesn't bother me too much but I was hoping to be under a minute, though I never really use UHD footage (I'm a hobbyist, not a professional). I tried "parking" my E-cores to see if I saw any improvement but honestly it didn't make a difference, but maybe I'm doing something wrong in that respect (used Affinity setting in task manager to park them, had VEGAS set to "High" priority). I ran userbenchmark on the system and it said my CPU ranks in the bottom 20th percentile of i7 7700K CPUs tested so maybe I need to optimize my bios settings or something.

That said, I am thrilled with my PC overall and don't really have any reason to tinker too much. If anyone has recommendations, or suggestions, I'm happy to test them out and report back. I appreciate all the information from everyone on this forum and look forward to having more fun with VEGAS in the future! I really missed video editing (haven't done much of it since January 2021! Yikes!)

Howard-Vigorita wrote on 2/10/2024, 2:03 PM

@Matthias-Claflin I was just looking at the Sample Project charts and your time was almost the same as others with a similar cpu and gpu. Keep in mind that Sample Project is light on media load but heavy on generated media. So the igpu doesn't impact as much as gpu media generator performance. I notice that someone with the same cpu as yours but without an igpu got the same time with an Nvidia 4070. Amd gpus happen to run that benchmark's media generator faster than Nvidia but are not necessarily faster all around. The RedCar benches in my signature is a little more balanced with different versions for different media. While the other bench in @RogerS signature is better reflective of modern Vegas capabilities since vp20.

Matthias-Claflin wrote on 2/10/2024, 2:11 PM

Thanks for the response! I appreciate it and will check out the bench in your signature. Curious how the machine shakes out. I was just really hoping I could pull a little more performance out (like 2 seconds) but I'm not upset. It certainly is lightning fast with my HD footage anyway and the two games I play are in the hundreds of fps now.

RogerS wrote on 2/10/2024, 10:41 PM

So I'm back! I have built the PC (exact specs in signature). I am happy with it so far for the two games I play. I played around my with C100 footage I need to edit with both VEGAS Pro 21 and VEGAS Effects and it works a treat. I did run the 4k benchmark from @RogerS's signature. Was quite disappointed with my results to be honest, in terms of comparing it to others and honestly haven't replicated my best render time of 01m:01s since that test (rendered in QVC rather than NV or MainConcept). Most of the time it renders in UHD at about 01m:27-33s. Honestly this is fine and doesn't bother me too much but I was hoping to be under a minute, though I never really use UHD footage (I'm a hobbyist, not a professional). I tried "parking" my E-cores to see if I saw any improvement but honestly it didn't make a difference, but maybe I'm doing something wrong in that respect (used Affinity setting in task manager to park them, had VEGAS set to "High" priority). I ran userbenchmark on the system and it said my CPU ranks in the bottom 20th percentile of i7 7700K CPUs tested so maybe I need to optimize my bios settings or something.

That said, I am thrilled with my PC overall and don't really have any reason to tinker too much. If anyone has recommendations, or suggestions, I'm happy to test them out and report back. I appreciate all the information from everyone on this forum and look forward to having more fun with VEGAS in the future! I really missed video editing (haven't done much of it since January 2021! Yikes!)

Hi Matthias, I looked at your bench result and made a comment. Put dynamic ram preview back to the default of 5% and you will likely do somewhat better. I would not park your E-cores, they are used in rendering. I would also not set VEGAS to high priority, just keep it open in the foreground.

Your results aren't so different than what I'm getting with a similar CPU and higher end but older generation GPU, so seems fine. Go to filter views and only look at the UHD renders.

Keep an eye on the cores with Intel Extreme Turning or HWInfo64 and if you see any signs of throttling with extended renders then it would be good to go into bios and tweak voltages or clock speeds. Some motherboards come with unrealistically high wattage amounts that require too high voltage to meet and start throttling under load. Intel Extreme Tuning has a CPU stress test you can use to see if that happens.

Also give the VP 20 benchmark a try if you feel like it too as it uses very different Fx.

 

Former user wrote on 2/10/2024, 10:58 PM

I did run the 4k benchmark from @RogerS's signature. Was quite disappointed with my results to be honest, in terms of comparing it to others and honestly haven't replicated my best render time of 01m:01s since that test (rendered in QVC rather than NV or MainConcept). Most of the time it renders in UHD at about 01m:27-33s.

I ran userbenchmark on the system and it said my CPU ranks in the bottom 20th percentile of i7 7700K CPUs tested so maybe I need to optimize my bios settings or something.

Hopefully that means your CPU is limiting performance and not the 4060ti . I did a test with a 4090 for the 4K NVENC benchmark and I can turn it down to 30% GPU power and hardly anything changes

51.12  100%  (voukoder)
53.42  100%

54.45  50%

54.51 30%

Your CPU does much better with NLE's compared to 7000 series Ryzen and my 5000 series. I don't know much about Intel, but check on power limits and throttling. What is your cooler?

 

 

 

Matthias-Claflin wrote on 2/11/2024, 4:02 PM

Hi Matthias, I looked at your bench result and made a comment. Put dynamic ram preview back to the default of 5% and you will likely do somewhat better. I would not park your E-cores, they are used in rendering. I would also not set VEGAS to high priority, just keep it open in the foreground.

Just did this with Main Concept I got 01m:23s. With NV I got 00m:58s. With QSV I got 00m:52s. Shocking how changing the RAM preview made everything genuinely so much faster. Wild, but also great!

Former user wrote on 2/11/2024, 4:47 PM

Just did this with Main Concept I got 01m:23s. With NV I got 00m:58s. With QSV I got 00m:52s. Shocking how changing the RAM preview made everything genuinely so much faster. Wild, but also great!

Yeah should never have it at 0 percent, that was mostly for a black frame issue that existed in previous's vegas's with the negative that is caused a slowdown with both playback and encoding. The interesting part is why you get 58s with NVENC, and 52s with Quicksync. I'd imagine you don't care and will simply use QSV.

Both my testing of running my GPU at 30% and running your GPU with the QSV encoder proves the 4060ti isn't a bottleneck in this benchmark, it's curious about the NVENC encode being 6 seconds slower, unless QSV doesn't have the pulse of encoding that NVENC does via Vegas every 50 frames or so, and instead QSV is writing much more often

Only other mystery is how I get 53.42s for NVENC, while you get 58s even though you have the much better CPU for Vegas. If anyone has ideas I'd like to hear them.

RogerS wrote on 2/11/2024, 8:10 PM

I added your other times to the spreadsheet as I think it's informative to other users the difference dynamic ram preview makes. Congrats, your system is now tied for 2nd place for UHD renders!

QSV does seem to write continuously though I don't know if that's the difference or rather an advantage of being on the CPU bus? In this benchmark I also have a gap of 52s for QSV and 59 for NVENC.

Former user wrote on 2/11/2024, 8:56 PM

QSV does seem to write continuously though I don't know if that's the difference or rather an advantage of being on the CPU bus? In this benchmark I also have a gap of 52s for QSV and 59 for NVENC.

@RogerS that's a great idea, and the problem with Vegas using discreet GPU encoder/decoders is that Vegas is using kludged code meant for Qucksync, and not taking into account of extra latency and different timings, I think that is much more realistic than Howard's idea that you're spreading the load causing less stress? for the discreet GPU. That never made sense to me. And also I think Howard founds his QSV better on CPU rather then on his Intel GPU,,, though that's a distant foggy memory

 

Matthias-Claflin wrote on 2/11/2024, 10:23 PM

I wanted to quickly add that I had a minute to play with the RAM preview and going a little above 5%, using 10% and then using 13% (which happened to be about 8GB of RAM) was slower in NV than previous renders. Adding about 5-10 seconds. It added about the same to my QSV render times (putting it around 1 minute each time). It seems like 5% dynamic RAM preview is the sweet spot, but maybe you guys already knew that. I just had to test myself.

Also Thanks Roger, I wonder if the other user who sent in i7 14700KF (?) and RTX 4070 results had his dynamic RAM preview too low or too high?

Last changed by Matthias-Claflin on 2/11/2024, 10:24 PM, changed a total of 1 times.

PC:
i7 14700K
64GB DDR5 5600mhz
NVIDIA GeForce RTX 4060Ti
500GB SSD (OS)
2TB SSD (VEGAS Pro 16)

Laptop:
AMD Ryzen 7 5800H
16 GB RAM
NVIDIA GeForce RTX 3050
500GB SSD

Cameras:
Canon C100
Canon 77D
Canon T3i

RogerS wrote on 2/11/2024, 10:35 PM

Column T is dynamic ram preview (it's to the right as I realized its importance and added it to the form afterwards). The other user also had it set to 0.

I think 5% is the default for a reason and it's good to do your own tests to confirm. I often keep it at 10% as I actually do ram previews (shift + b).