Be careful what you wish for. Sony did this and nobody liked it because they recommended professional hardware that no one here was willing to pay for! They were gorgeous systems from BOXX that used workstation class GPU's. It seems that at the time, people here only wanted to buy consumer grade PC's. The same thing happened to Adobe. They had a list of accepted graphics cards which contained mostly workstation GPU's and customer figured out how to hack the list and add their game cards.

Avid does something similar in listing approved hardware configurations. The problem is... you need to be willing to buy them because gamer cards are not tested with workstation class apps and the drivers are not as stable as workstation class cards. I bought an NVIDIA Quadro 4000 and never had a single problem with it and Vegas Pro.

The problem MAGIX now has is that Vegas Pro doesn't support any modern GPU's for rendering because of 3rd party encoders that got abandoned by their manufacturers. What MAGIX needs to do is fix their encoders to support modern GPU's.

It would be nice though to know what cards MAGIX has tested with. This is one of the reasons I moved to the Mac. The GPU's in my Mac are the same GPU's in the Final Cut Pro X developer's Macs. It's nice to not have to hassle with all this GPU nonsense. ;-)

~jr

It would be nice though to know what cards MAGIX has tested with.

~jr

Yep, exactly. And if people dont want to spend the money on recommended cards then they have to accept their system may not be optimal. Thats fair enough. But Magix should really release recommended hardware at least so people can choose. SCS never gave us that information.

Johnny is wright - there has been recommondations for GPUs for Vegas by SCS, and they made sense but where expensive. And I do not know what is expected here today. We know quite well that GPUs like the AMD R9 390 (with or without X) support Vegas in a nice way in the playback behaviour of effects like transitions or color corrections. We know quite well that you can forget the render support. If the development team can update that with Vegas 14 that would be great - but maybe it is not the most pressing point. I would like to see a better preview with 3rd party cards like AJA or Blackmagic before that, and also professional codecs like ProRes are important too. At least for me. The R9 390 X is doing a nice job here. 

SCS never gave us that information.

Not true. That information was provided starting with VP11 and maybe still is but it never changed because Vegas never got updated afterwards.

With respect, I never seen any official data OldSmoke, and i remember there being multiple requests made with the response from SCS as 'we cannot comment on hardware..' or words to that effect. God knows how many threads with GPU findings all because people were confused. The lack of information and direction as staggering. I can only hope Magix is more forthcoming. Which is the point of this post.

It is really a lose, lose situation for a company to go into specifics on hardware spec beyond min requirements. Like CPU speed, core, ram, disk space and such. The internet will tee off on anything you say. If you say X "works", I guarantee the net will find crashes/problems and now you are a liar in their eyes. It's the wild west out there. Too many degrees of freedom to really lock everything down.

Really the Vegas video engine is coded to OpenCL. OpenCL has nothing to do with AMD, Nvidia, Intel or whoever. Those vendors have OpenCL compiler and driver support. It is their business to make sure that is as bug free as possible. If is Vegas developers business to make sure their stuff is as bug free as possible. Maybe they have to work around bugs in vendor implementations (OpenCL or anything else).

One can have perfectly generic OpenCL code but that may not wring the most performance out of a given OpenCL implementation. AMD and Nvidia GPUs are structed differently at various technicaly levels. A certain orginization might be "ideal" for implementation X but not for implementation Y. A valid structure is not necessarily and optimal structure. A good example of a concept like this is the Luxmark benchmark commonly used as an indicator of OpenCL performance. Luxmark 3.0 did not work well on Nvidia platforms even after Nvidia improved their OpenCL drivers. Nvidia informed the developers of some suboptimal things they were doing and Luxmark 3.1 performs a ton better on Nvidia (same GPU and driver). It also means you cannot compare 3.1 to 3.0 numbers. Could the Luxmark situation be similar to some reasons why Vegas likes AMD OpenCL implementations so much better than Nvidia? Things that make you go...Hmmm.

With respect, I never seen any official data OldSmoke, and i remember there being multiple requests made with the response from SCS as 'we cannot comment on hardware..' or words to that effect. God knows how many threads with GPU findings all because people were confused. The lack of information and direction as staggering. I can only hope Magix is more forthcoming. Which is the point of this post.

Open the "Vegas_readme.htm" file in your Vegas Pro 13 "REadme" folder and you will find the following under system requirements:

Supported NVIDIA, AMD/ATI, or Intel GPU with at least 512 MB of memory (for GPU-accelerated AVC rendering and video processing; 1 GB recommended for 4K):

NVIDIA

Requires a CUDA-enabled GPU and driver 270.xx or later.

GeForce GPUs: GeForce GTX 4xx Series or higher (or GeForce GT 2xx Series or higher with driver 297.03 or later).

Quadro GPUs: Quadro 600 or higher (or Quadro FX 1700 or higher with driver 297.03 or later).

NVIDIA recommends NVIDIA Quadro for professional applications and recommends use of the latest boards based on the Fermi architecture.

AMD/ATI

Requires an OpenCL-enabled GPU and Catalyst driver 11.7 or later with a Radeon HD 57xx or higher GPU. If using a FirePro GPU, FirePro unified driver 8.85 or later is required. Radeon HD 7xxx or higher recommended for native 4K editing.

Intel

Requires an OpenCL-enabled GPU (such as HD Graphics 4000 or higher).

For the latest information about supported hardware, please see our Web site.
For more information about troubleshooting GPU acceleration issues, we've prepared an article on our knowledge base.

This information hasnt changed since VP11 and was removed from the SCS website just before VP13 came out, I believe. I assume SCS didn't want to fully publish that information anymore as it is dated but it is still valid.

GeForce GPUs: GeForce GTX 4xx Series or higher (or GeForce GT 2xx Series or higher with driver 297.03 or later).

I'd like to see anyone get any real GPU performance gain out of one of those..  Bad information and more like a 'minimum spec' sales pitch than a recommended and tested card. I used a GTX570 which did give a performance gain but lower cards were simply not worth the effort. That was the problem, and you had people on the forum wondering why their 210 card didnt do much in some cases worse than the CPU.. They could have been much more clear about it and save people time and money instead of saying 'we cannot comment on specific hardware'.

Nice of SCS to release dud information and not comment on the forum to clear it up. I recently had a similar incident with Catalyst. Because it didnt work on my laptop. Their minimum spec gave a value for Open CL required which my laptop has.. But after wasting time back and forth, it actually turned out to be the CPU with lacking SSE3 (i think it was). A simple mistake, they had said OpenCL 2.1 OR SSE3.. Not AND.. That wasted about 2 days and i had to figure it out myself. Just shoddy if you ask me.

However, saying all that, thanks for posting the information anyway. I had not seen that list before now.

Moving on in a more positive way, i repeat my post to Magix, Will we see any recommened hardware that has actually been tested by you so we can be sure the GPU card we buy will actually work. Saving us time and money in the process.

Thanks again.

Here is the problem: It's not about one component like the GPU. When I bought my NVIDIA Quadro 4000 it said on the box that it required a Core i7 Quad. Nowhere on the internet on any site (including NVIDIA) did it say that it required a Core i7 Quad but there it was right on the box! At the time, I had a 1st generation Core 2 Quad but I put the card in anyway and it worked pretty good. Then I finally upgraded to a Core i7 Quad and I put my Quadro 4000 in that box and it took off! Man did it fly. All that time I'm thinking that the Quadro 4000 didn't give me the boost I was looking for but it was my CPU that was holding it back. So you need to have a balanced system not just one killer component. That's why it's important to recommended systems and not just a component like the GPU.

The other thing wrong with the information that Sony gave on GPUs is that it was specification driven. What people wanted to know is which model and manufacturer to buy, not just the specification. So saying, "Any card with OpenCL 1.0 or higher" is a nearly useless statement because lots of cards support this poorly. We wanted to know which ones Sony used for development and testing and what do they recommend. That was something Sony was never willing to say. Maybe MAGIX will be more forthcoming. Maybe the lawyers won't let them? 

~jr

We wanted to know which ones Sony used for development and testing and what do they recommend.

jr

I remember clearly that when VP11 came out and the SCS website had the GPU acceleration test published that SCS very well stated which cards they used for their tests, not brands but models. If I remember it correctly it was a GTX580, HD57xx, a FirePro and a Quadro. There was also a chart and in most tests, the GTX580 was on top but for preview the HD57xx was on top. I am not at my system but I believe there is also a PDF file included in the SCS Benchmark project when you downloaded it and it had the same information.

I already had a GTX460 with VP10, when VP11 came out and it finally got me better render times and preview. I then studied their tests and charts and bought a GTX570, the 580 was too espensive.

I doubt you will ever see a software vendor suggesting a specific brand, always only a model or architecture. Also keep in mand that such a suggestion can potentially drive away customers, those that say "that NLE can only work with such expensive hardware". 

I would rather say that the information at the time it was released was more targeted for those that could read it correctly without having to mention a specific brand or model and also not limiting it to the most expensive hardware.

I remember clearly that when VP11 came out and the SCS website had the GPU acceleration test published that SCS very well stated which cards they used for their tests, not brands but models.

Yes, I remember that PDF. It was including in the benchmark ZIP file. That was rather good. The AMD Radeon HD 5870 in my 2008 Mac Pro beat all of the cards in that test and played back the benchmark flawlessly at full frame rates all the way through. I was very happy with it.

I doubt you will ever see a software vendor suggesting a specific brand, always only a model or architecture. Also keep in mand that such a suggestion can potentially drive away customers, those that say "that NLE can only work with such expensive hardware". 

True, but even knowing the model numbers of what they develop and test with would be nice to know.

Like I said before, one of the reasons I moved to the Mac was because the GPU in the Mac Pro that I buy is the same GPU that the Final Cut Pro X developers use. It's really great when you are using the exact same hardware configuration the developers are using to test with. Apparently, lawyers are keeping PC users from getting the same benefit. ;-)

~jr

Apparently, lawyers are keeping PC users from getting the same benefit. ;-)

That could very well be the case. On the other hand, a PC user can always upgrade it's hardware at a reasonable cost, try that with Apple's latest MacPro.

That could very well be the case. On the other hand, a PC user can always upgrade it's hardware at a reasonable cost, try that with Apple's latest MacPro.

I would gladly give that up for stability. (which is what I did) ...and how is that any different than Sony telling you to use an GeForce GTX570 5 years after it has gone end-of-life? I can put any GPU I want in my 2010 Mac Pro but I choose to only put Mac Edition cards that are supported by Apple because I value stability above all else. Speed means nothing if it isn't stable!

~jr

 my 2010 Mac Pro

That's old hardware too.

As for stability, no issue here, not with the GTX570, 580, 290 and now Fury X.

I also remember this presentation that we had with Vegas 11. And yes, one of the cards was an very expensive Quadro (today I have also a Quadro K4200 in my system).

But that was never the problem to come up with a recommendation.

The problem was that the GPU acceration implementation became more and more outdated. I also had a GTX570 in those days - BUT that is a card that nobody would purchase today. And a lot of people purchased more moderne nvidia cards - that were not supported any more.

So a list of recommended hardware is great - but only if it is actualized when new hardware arrives.

That's old hardware too.

Yes, and isn't it a testament to Apple engineering that a 6 year old 12-Core Mac Pro still has more than enough power to get the job done with today's workloads. It is unfortunate that Apple chose to go down the road of consumer planned obsolescence with their new Mac Pro (which is why I didn't buy one). So I agree with you that the new ones have gone in the wrong direction. I can only hope that Apple will correct this in their next Pro model. 

~jr

Yes, and isn't it a testament to Apple engineering that a 6 year old 12-Core Mac Pro still has more than enough power to get the job done

Not really. My 2600K with the GTX570 from that time would not be far behind my current 3930K and I would have been able to put in my current Fury X too. A 12 core Xeon, a GTX 580 and Windows 7 in 2010 would be the same today as your MacPro, but would have cost half. Even HP Z workstation as the one used by SCS would still be ok today, for Full HD anyways. The major difference I see is that those that dont want or dont have the knowledge to build their own PCs can get hardware that has been tested, althought primarily for Apple's own OS.

I wonder how well a current MacPro would run Vegas. But by all means, if I ever switch to a Mac, I would also switch to FCP X, it's easy to learn and it's multicam capabilities have some nice features already build in, like the audio sync feature.

Thanks for the reply Eric. My hope was for guidance when choosing a card. Simply to say if a range of product will actually work or not, not a specific card or a particular PC setup. That would be unrealistic.

Simply to say if a range of product will actually work or not, not a specific card or a particular PC setup.

How would that be different from what SCS provided?

We covered this already, going in circles. The difference would be the data would be accurate and up to date. ie. Tested and working. Unlike the 2xx and 4xx cards in the SCS info which are of no performance gain. Also including more up to date cards which work in the real world of Vegas and not just on paper.

Unlike the 2xx and 4xx cards in the SCS info which are of no performance gain.

Again, not quite true. It really depends on the whole system. Vegas was always a NLE that would run on rather low end hardware. If you have an old Q6600 cpu even a 2xx card will give improvement and my GTX460 did actually pretty well combined with the Q6600 and later with the 2600K cpu.

Where do you draw the line? I think that is pretty much the reason why you see under system requirements a rather open specification and a little bit more specific for 4K editing. It would require extensive testing of combinations of CPU, GPU and RAM to get list of recommendations and it would have to be project related too, meaning depending on the amount of work you do with that system, SD, HD, 4K, lots of FX, compositing and so on.

Yes, it should be more up to date but as far as Nvidia goes, the "old" info is unfortunatelly still valid, same for AMD with the exeption that newer AMD cards have improved timeline performance and much faster render times with MPEG2 codecs.