Trade in your new gtx 1080 for a 6 year old card, then everything should be fine😂

JN & relaxvideo... appreciate your response but really does not help me with my issue... I am new to vegas pro / about 8 weeks now... can you explain why nvidia not a good with vegas pro and suggest what card would be best? Am still a newbie with vegas pro and still in the learning/struggling curve

Thanks

Hi Mike, hope this FAQ can help:

https://www.vegascreativesoftware.info/us/forum/faq-how-can-i-stop-vegas-pro-hanging-during-rendering--104786/

and this too:

https://www.vegascreativesoftware.info/us/forum/faq-graphics-cards-gpu-acceleration-for-vegas-pro--104614/

Maybe just wait, something on the horizon!? .. https://www.vegascreativesoftware.info/us/forum/came-across-this-today-magix-vegas-and-mainconcept--104891/

Then again, many SVP users have faded away and perished while waiting and waiting😩

It's possible they may come back later...

Thanks for the links M... I read all of the content and definitely helped me understand some of the challenges with video cards. Luckily, I had my new pc custom built locally and have 14 days to make hardware changes as needed. After reviewing info from the links, seems that AMD may be the best card for my needs. Going to contact S1P in the morning to get more info and will make card changes based on their suggestions...

Thanks---Mike

..will make card changes based on their suggestions...

If Vegas is your priority then I would currently suggest that they suggest "Nvidia GTX 1080 out, AMD RX 480 in". They may be resistant to that, as it's a cheaper card.

Thanks Nick... Yes Vegas is my priority for now... And spoke with them today and they said no issues at all as long as I make a decision within the 14 days...Just curious but was reading a post about using 2 video cards? Would that be an advantage for render time? and if so, complicated to set it up properly?

...2 video cards? Would that be an advantage for render time? and if so, complicated to set it up properly?

Only an advantage if the 1st card is great at OpenCL for GPU accleration of video processing (e.g. AMD RX480) and the 2nd card is an old card that's specifically suited to GPU-accelerated rendering with the MainConcept AVC codec (i.e. Nvidia GTX 580 (or earlier) or perhaps AMD HD 6970 (or earlier)). These are old cards you'll probably be looking for on eBay, not supplied new by your system builder. It shouldn't be too complicated to set up as long as you have a motherboard slot(s) available and sufficient wattage in your power supply. I've had both GTX 580 and AMD HD 6970 happily in my machine at the same time. Having said that, while RX480 + GTX580 in the same machine is probably just about the fastest current way of churning out AVC videos from Vegas, it's probably not worth the bother unless you're in a production environment where that's important. Think of the extra heat, power requirement, potential hassle, noise, slight loss of quality with GPU-rendering etc..

Nick said ...

"Only an advantage if the 1st card is great at OpenCL for GPU accleration of video processing (e.g. AMD RX480)"

I'd be surprised if leaving in the 1080 wouldn't be at least as good for playback on the timeline as the rx480, if not better, given it's a much more powerful card. OpenCL advantage of AMD vs "modern" Nvidia cards is marginal, and card performance needs to be taken into account also.

If anyone has similar to Motherboard .. MAXIMUS VII RANGER. CPU .. Haswell Core i7-4790K. Memory .. Corsair DDR3 16GB GTX 1080, my spec. I can compare with an AMD RX480 that anyone has with an i7 -4790K i.e. timeline playback times.

John Rofrano supplied a sample test file that attempted to seperate out playback and render times by virtue of the project nature, of his .veg file. I have the file but can't find the origional reference/post.

All joking aside Mike, if it was me i'd leave the PC as is with the 1080. If Magix don't provide improved gpu accelerated template(s) for modern cards in the short term you can still add in a cheap gtx 580 for rendering with gpu acceleration. If there's anything to the link I previously copied, we all might get lucky😊. Maybe just sit tight for a short while anyway.

These are the tests I did using the J.Rofranos test file, using gtx 580, SVP 13, note there's only 1 reference to the "NEW" gtx 1080, i.e. its playback time, which is maxed out at 29.97 fps.

The GPU used is the GTX 580, the new GPU is MSI GTX 1080 using J. Rofrano test file ...

P.C. was used (laptop in brackets) ...

Timeline GPU Acceleration OFF ... Playback 0.72fps (Laptop 0.370fps)

Timeline GPU Acceleration ON ...Playback 15.5fps (Laptop 1.36fps)

Timeline GPU Acceleration ON ... New GPU, MSI GTX 1080 ... playback 29.97fps

I haven't included the render test times, for clarity, but they are obviously much faster using the gtx 580.

This is the John Rofrano post and test file reference ... https://www.vegascreativesoftware.info/us/forum/i-found-the-best-graphics-card-for-vegas-pro-12-0--97029/

Nvidia GTX 1080

NeatVideo will be happier than last Christmas. Probably "happier than any other plugin out there". With Re:vision's RSMB and Twixtor sharing the second step of a "happiness podium".

...Regarding Vegas... As far as I could see, Vegas would never eat up that much resources.
(Maybe that's my bad luck with Vegas, ha-ha-ha.)
Seriously, there are other (than Vegas) bottlenecks to be expected.

I updated the John Rofrano post with the results of the Sapphire Nitro RX 480 8GB card. While render times improved from my original scores and timeline performance is good for Full HD, the whole process is underwhelming for the Canon XC15 4K media (16-20 fps on Draft Auto with CPUs only working at half of capacity). I expect to be underwhelmed with the i7-6850 upgrade, also.

Mike PM'd me about his dilema between the GTX 1080 and the RX 480 in light of JN__'s comments above, and specifically how he can test the peformance.

It may well be true that the GTX cards have caught up a lot in terms of OpenCL. NormanPCN talked in detail about this here. This seems to be very driver dependent, and in general terms I guess the more recent the driver the better. However megabit was happy with his GTX 1080 but then suddenly had problems with drivers 369.00 - 372.90. But the latest driver is 376.33, so maybe it's better again now???

As to the real answer between these GPUs, it's impossible to know until someone does some controlled testing of both cards in Vegas. I would love to know myself, as I'm soon in the market for one of these. Also, Google "amd rx 490" if you really want to fry your brain ;)

Anyway, for a quick test of the performance, stack up some generated media with FX, Pan/Crop, partial opacity. You could start with JohnnyRoy's rendertestjr.veg as mentioned above from this thread, but you'll have to make it more complex and/or rack up the project frame rate as these GPUs are likely to give full 29.97fps with that project as it stands.

For a more thorough test, download the VP11 "red car" project from here. Again, much of that timeline is probably now a bit simple to provide a useful test with recent CPUs/GPUs. To make it really tough you could add a track of generated media with opacity/FX/pan-crop (like in JR's project) on top of it and/or increase the frame rate, perhaps to a weird number in the 50s that is not a multiple of the media frame rate.

Render tests are useful; Even though these GPUs don't support GPU-accelerated rendering, the timeline still needs to be processed and "GPU acceleration of video processing" should speed that up. But for a quick answer, just watch the fps playback figure in white digits under the preview window.

I reviewed all of these comments and links too... A bit confusing but new to all of this so prob to be expected... I'm trying to do a few speed tests with this nvidia card with different formats but not quite sure how to calculate. I downloaded the red car example and rendered... It was 1948 frames and took 2 minutes and 13 seconds to complete...I'm measuring FPS... so basically I determine the total frames and divide by the total seconds to render? (1948 frames divided by 133 seconds?)

Hi Nick ... I redid the Red Car test on my PC with i7-4790K + gtx 1080. You are correct the Red Car project is also maxed out at a frame rate of 29.97, so a better/tougher test file (for fps) or modified version of JR's and/or Red Car is needed to show proper playback fps limitations on non legacy/modern cards. Maybe this is something that you could look into?😊 We need to slow it down😁.

Redid Red Car test on PC, previously forgot to include GPU on (off) values as other posters have.

i7-4790K + GTX 1080 + 16gb ram. [GTX 580 values in square brackets with GPU ON].

XDCAM .. 0:31s (2:58s) [GTX 580 0:33s]

MC .. 2:15s (3:11s) [GTX 580 1:40s]

Just for comparison the render times on a PC with i7-6700K, GTX 1070, 32 GB RAM

Render time: GPU on (off)

XDCAM: 0:28 min (2:38 min)

MC: 2:06 min (2:51 min)

Just ran the same test... PC with i7-6700K, Nvidia 1080, 32 GB RAM

Render Time: GPU on

XDCAM: 0:26 sec's

MC: 2:05 min's

PC with i7-3770K, RX 480-8GB, 16 GB RAM

Render Time: GPU on

XDCAM: 0:31 sec's

MC: 2:33 min's

Most of what I have watched with regards to Nvidia OpenCL improvements have been with the Luxmark benchmark. Vegas has never liked Nvidia OpenCL. It could very well be that Vegas is doing this "bad" for performance on an Nvidia implementation. OpenCL allows one to be hardware agnostic but for best/max performance you probably still need to consider some tweaks per architecture.

One thing to remember about the "red car" project. It stresses/uses the GPU by using a lot of effects. All Nvidia and AMD GPUs are supported by the video effects and compositing engine. A lot of effects goes to more GPU use. No effects goes to little/no GPU use.

EDIT: The red card project is testing and stressing the video engine. A file encoder test is probably best done with just simple media and no effects. Get everything else out the the way so you are testing the file encoder itself.

The Mainconcept AVC GPU support, for file encoding, is separate and independent of the video engine GPU use. It is Mainconcept that has the limited support for only older GPUs. AMD via OpenCL and Nvidia via CUDA.

The video engine is always used. Playback and file encoding (render as). The video stream must always be generated(rendered) before given the frames to the display for playback or the file encoder for "render as".

The thing about playback is that it's performance is clamped at realtime framerate. If playback cannot make realtime then that is one thing. If it can make realtime, then we cannot know how much faster it can really go since it is clamped. Unclamped playback could be used for benchmarking independent of the file encode operation. About the only way I can see to implement this is to do a ram preview operation with a large ram preview buffer. The ram preview is unclamped and renders uncompressed frames to ram so that is fast. So a benchmark could be to buffer 15 seconds of timeline from time X to X+15.

(note vegas has bugs with Nvidia and ram preview. at least on my installation (GTX 980)).

(vegas seems to have a bug with(related to) Nvidia with a runaway thread that saps one full core on the CPU. Most times the thread stops after 45 seconds or so or fully stopped playback. It goes runaway again once playback starts. Repeated playback before the runaway stops make the runnaway time last MUCH longer after playback stops).

File encoding (render as) is not realtime framerate clamped. The problem with file encoding is that it has a lot of overhead and computation and due to this is almost never makes realtime. Each file encoder has it own performance boundaries. More powerful compression codecs like HEVC and AVC are typically slower than encoding to simpler ones like mpeg-2 (XDCAM) or Cineform/DNxHD/Prores.

Hi Norman,

Do you have any experience or an opinion about the AMD RX 480? I've decided that I "will" replace the Nvidia card but not 100% sure which one to go with.