Now the question is will the 5.1ghz capable do big editing project and stable? Day in and day out?. As its stands mine is working computer, for everything day in and day out.
One thing I couldn't help noticing was how much the Radeon7 stood out from the crowd in the LUT test. I think it's probably pretty much the same as the Vega64 but in a smaller dye that should let it cool more efficiently. I've played with a Vega bios a bit and noticed it basically uses temperature limits to dynamically auto tune gpu voltage and clocking so I'd expect more efficient cooling to boost performance. I'm probably a little late to the party but after reading that article I only just now installed the free OFX 3D LUT Creator plugin for the first time and it looks quite interesting... wonder why Magix left it out from all the other OFX effects it supplies with Vegas. Puts Vegas right on par with Da Vinci color tuning. Looking forward to getting my hands on a Radeon7 to go with it. Btw, I get 25 sec on the red-car with an air cooled rx580.
I'm using my 9900K for daily paid work, so I plan to keep it at a safe & quiet 5.0ghz. Even though it manages 5.1ghz without risky voltage or temp increases my radiator fans spin-up & my system is no longer quiet... Since the GPU is also liquid-cooled & I have M.2 drives, this system is dead-quiet at 5.0 ghz...
I keep seeing deals for Vega 56 & 64 since VII is here & Navi is arriving shortly. I got a Vega 64 liquid-cooled (refurbished) from Newegg for $350 USD. It shows no signs of dust or wear. IMO AMD is unloading the $300 - $400 range VEGA 56/64 to make room for NAVI while continuing to charge $700 US for the Radeon VII. So for the short-term, Vega 56/64 are a good bang/buck for us Vegas folks & they can be undervolted to save power...
I have a 1950x threadripper, radeon 7 box I just built up and it's not in the ball park. 33 sec is the fastest red car test I've gotten with every driver and version of vp16 I can try out. It's head and shoulders above my older intel nvidia machine, but nothing like these numbers folks are posting here.
Now the question is will the 5.1ghz capable do big editing project and stable? Day in and day out?. As its stands mine is working computer, for everything day in and day out.
Well it depends, apart from proper BIOS tuning, cooling, etc.., it is also matter of which application you run, and which type of processor you have:
1) the processor: for Intel the newer 9900K has a soldered TIM versus previous generations which helps a lot to get the heat out properly, that is why it is officially (no overclock) can go (single core) to 5GHz. That is why I got 8700K handpicked and de-lidded and improved TIM CPU from der8auer, it is tested by der8auer to 5.1 GHz with upping to 1.38V. Not that I want to overclock, as I do not, it just helps to keep the heat down and the noise. My system is fully air-cooled.
2) AVX code instructions (in all it's variants), Intel is more performant with AVX than AMD, but it's a double bladed knife: if the application does not use it, you are not going to benefit from it. I am not sure if Vegas use it, maybe some external plugins do. I know Magix's sister NLE Video pro X likely make use of it as they advertise this on their product website to improve speed on color correction and image processing. But here is the caveat, it is impossible to run AVX without heat issues when the processor is overclocked. That is why in motherboards suitable for overclocking you have settings to automatically lower the overclock multiplier if AVX instructions are used. Kind of defeats the purpose of improved speed with AVX support, at least when overclocking.
Note: you can test your overclocked CPU if it can cope with AVX if you use the newer versions of Prime95: Prime95 AVX versions such as 29.4 as compared to Prime95 non-AVX version 26.6.
I have a 1950x threadripper, radeon 7 box I just built up and it's not in the ball park. 33 sec is the fastest red car test I've gotten with every driver and version of vp16 I can try out. It's head and shoulders above my older intel nvidia machine, but nothing like these numbers folks are posting here.
Cheers, Pete
I've been building my own custom Intel & AMD systems for over 25 years & video editing workstations for over 15... A lot of factors can affect Vegas performance because it is so resource intensive. For my 9900K I had to re-install the Intel GPU drivers AFTER installing my Radeon Vega drivers so that both GPUs would show-up in Vegas. As soon as I did this my Intel QSV render times dropped significantly.
I also use CCleaner to remove Vegas, clear the registry, restart computer, and then reinstall Vegas after the new CPU, GPU, etc. drivers are added. Once I get my system where I want it, I turn-off Windows updates the best I can, create a full image backup of the OS drive, and then test to make sure it works on a new drive. I also do a full image backup before installing any new apps, codecs, hardware, etc. Work files are on a RAID10 with backups on a server...
Are you getting a case of "tech megalomania?" 😀 Anyway, your results are reflected in the Techgage article - looks like the Vega 64 cards are real winners for Vegas (also popular with some of the more technical members of the DR forums). Too bad the Canadian dollar has sunk so low...for now I can only dream.
No. Perhaps i did not clarify myself better. My statement mainly imply to those insisted nivida that and that.
When you VCE encode are you using the QUALITY setting, which is the only one that should be considered for this purpose, or are you using another VCE setting?
Former user
wrote on 5/3/2019, 10:23 AM
FWIW When I use say Nvenc, VCE or Qsv I have always used the default settings. I wouldn't use that if it was for anything other than testing, I would use the best quality available.
There's a pdf supplied with the red car test, it includes the render settings. "For Vegas Pro 11 the “MainConcept AVC Internet HD 1080p” template was used. If its to remain a valid benchmark then users should continue to use these settings, otherwise it's not a case of comparing like with like.
This is the Magix Mainconcept one I use, in the day it was the MC one that's now in "Legacy" settings. This is probably what most other users have used, I guess.
Ok, I have my answer: you have to be sure that preview for the "red car" project is set at Best (Full) otherwise the test render could be done only at Good depending of your actual project preview setting.
Former user
wrote on 5/3/2019, 5:24 PM
FWIW When I use say Nvenc, VCE or Qsv I have always used the default settings. I wouldn't use that if it was for anything other than testing, I would use the best quality available.
Judging by encoding time alone, it seems Nvenc default is the same as 'high performance' rather than 'high quality' which is slower. I know at least one person here has said they believe the quality of encode is all the same for Vegas Nvenc, but I don't know how thorough they did their comparison. I do some lowbitrate encoding 2.5 mbit (720p) for content delivery at that bitrate & I don't trust any hardware encoding at those bitrates. VCE AVC is by far the worst in this scenario
Former user
wrote on 5/3/2019, 6:48 PM
Hi @Former user yes i’ve seen comments to that effect. When I get a chance i’ll do a comparison ffmpeg ssim/psnr comparing with what i’ve already done. The settings i’ve used for HO render quality metrics are 135/100mbps, preset HQ, RC mode Vbr HQ, so i’ll just do default nvenc which is Preset default, RC mode vbr.
When I do red car test I just use default. When I do the HO and ffmpeg tests I use nvenc HQ, i.e. preset=High quality, RC mode=VBR - high quality
Just tried on laptop, no go, zero file size each time with nvenc default setting. I can still create the preset=High quality, RC mode=VBR - high quality file.
I'll give it a go on my PC tomorrow.
Former user
wrote on 5/4/2019, 5:54 AM
Ok, got it to work no problem on PC. The Nvenc default is better quality, using the ffmpeg test, than the Nvenc High quality vbr. I used the same source I had previously used for the HO and ffmpeg testing, 4K UHD, 25fps, 27s duration, rendered out to same specs, using Magix MC Nvenc.
Higher is better ...
Encode mode: NV Encoder Preset: High quality RC Mode - VBR high quality
With this talk of how VEGAS Pro uses GPU, I thought I'd share this observation from today: My brother wanted to stitch together some video clips, and I suggested to him that there's a free program called "Davinci" that he could try to use for that purpose. Since he didn't know how to use it, I decided to try to help...I downloaded, installed it on my own machine. That program told me my GPU was too WEAK for it! It's a VEGA 64 Frontier Edition!
OC'd 9900K to 5.0 ghz and achieved 15s on Red Car on 6/8 tests. Two were 14s...
Ran stable for 1 week, so keeping system at 5.0ghz. What I learned from tests are that Vega 64 GPU contributes even when Intel GPU/QSV is selected. Also, drivers matter. If I choose very latest AMD & Intel drivers my render speeds slow a bit.... IMO combining Intel 9900k with AMD gpu gives benefit of both for various FX & codecs that favor one over other. If you choose Threadripper you lose QSV...m
@OldSmoke I also have a Radeon Fury-X but I am far from getting 18sec on the red car test. Can you tell me with which exact render setting you are getting such result ?
@OldSmoke I also have a Radeon Fury-X but I am far from getting 18sec on the red car test. Can you tell me with which exact render setting you are getting such result ?