> Is this a possibility for Premiere Pro CS5?
Premiere Pro CS5 does not use SLI, if that's what you're asking.
We will not comment on what's coming in the future. We will ask you to tell us what you want in the future.
I'm not representing any knowledge I have about Adobe's future plans, but I'd have to tell you "yes" if I was making my best guess at it. In the last few revisions, Adobe added support for multiple CPU cores within the PPro app. Now that they've added enhanced GPU utilization through MPE, it only makes sense that they would attempt to utilize multiple GPUs in some way or another in a future version.
I wouldn't expect to see this in an update/patch for CS5, nor would I even hold my breath for CS6...but perhaps, who knows. And really, at this point, even if Adobe COULD tell you "yes, we're working on this," that doesn't mean it WILL happen. Lots of things go into development on an application and then never come out successfully.
But I really would expect some sort of multiple GPU utilization at some point way down the road, and I would guess that the folks behind the scenes on PPro have already started working on this at some level.
It is funny though...we've had this amazing GPU acceleration from MPE for maybe 4 months now, and it didn't even take that long for people to ask for more.
"More??? You want MORE??????"
"More??? You want MORE??????""
Yes, but don't forget to say "please" just like Oliver.
Does that also apply to After Effects CS5?
This is just my opinion and many may have different priorities, but if it were up to me, this would be my development sequence:
1. Add GPU support to encoding/decoding, decreasing export/encoding times.
2. Add more effects/transitions for hardware acceleration.
3. Add MPE support to AE.
4. Port EN to 64 bit and enable GPU assisted encoding there as well.
5. Extend hardware MPE to OpenCL, allowing ATI into the race with more than 2 monitor support.
and finally SLI support.
Maybe point 5 should be moved up in the ranking.
I'd put 5 at position 1, myself.
Except I'd not :"extend" the support. Since recent cards from both AMD and nVidia are capable of OpenCL, I'd just drop nVidia's proprietary CUDA altogether and stick to the platform neutral OpenCL exclusively.
Come on, Jm, you don't want to throw away all the time and money invested into CUDA, do you?
Yes, actually. It's a development dead end if only because it will forever leave ATI users out in the cold.
OpenCL works with both cards. Spend the time and money there, and leave the proprietary solutions behind.
As a recent convert to Mac Pro, and being too naive to check the graphics card spec against supported GPU products from Adobe, I would have to say, please, pretty please Adobe? I'm sure you have an excellent relationship with nvidia that you don't want to give up which is fair enough, but it means for my meager production team of 1, extra time spent rendering that I could be using elsewhere. But, what the heck, you guys still do an excellent job with your software!
Development dead end or not (which is certainly more than a debatable statement), you have probably noticed the nvidia logo around your Adobe product literature, yes? It might as well be a dollar sign.
I don't think it should be a surprise to anyone that Adobe's exclusivity with nvidia is at least in some part or another a result of a financial agreement. I doubt Adobe "refuses" to work with ATI, but rather ATI is perhaps not willing to pay whatever bucks nvidia has on the table.
In addition to money, nvidia has also dedicated their own labor resources to working with Adobe in developing the GPU utilization.
And honestly, at this point, it's a minimum investment of a $275 card in most cases (GTX 470 being a really good currently available card). Relatively speaking, that is nothing to have to just buy a new card. I can see some folks getting a little upset if they had bought a beefy GPU sometime last April that wasn't compatible...but even still, if you bought an unsupported massive GPU prior to the MPE announcments, you were probably buying it for other reasons anyway (games). So your only real complaint then is that you can't double duty your card.
As for everyone else, I just categorize this sort of discussion with the folks who'd prefer to run PPro in a Linux environment. While more options are always good, it's not likely to change. Possibly in the future, but who knows the future? Same for OpenCL. If it happens, it happens. Until then, the rest of us are happy to buy a new card if it means saving hours and hours of time...regardless of what brand it might be.
And really, if you're only editing SD footage these days (ahem....no need to raise your hand) then MPE hardware acceleration doesn't do a whole lot for you in the first place.
If nvidia's involvement and exclusivity are keeping the Adobe prices down and the MPE development up, then I'm personally not going to be complaining.
The problem with that argument, Christian, is that we're not talking about "folks who want to run Adobe apps on Linux". The better analogy is "folks who want to run their Adobe apps on an AMD CPU". Software should be hardware neutral (assuming minimum specs are met). Any efforts on the part of nVidia to maintain exclusivity are akin to the Intel/Dell kickback debacle, wherein the former paid gobs of money to the latter in exchange for the latter keeping AMD processors out of DELL computers.
It may well serve both companies in some fashion, but it is a very large disservice to the buying public. So long as that disservice exists, I WILL complain.
As far as the development efforts of nVida are concerned, that would likely not change if Adobe moved to OpenCL. Rather, you'd add the development efforts of AMD into the mix, as I have no doubt both companies would have a stake in ensuring hardware MPE ran flawlessly on their respective hardware. And that can only benefit the consumer.
Software can't always be hardware neutral. And even while usually possible, one cannot assume that it is always the best route. Nobody can make the argument that an Intel processor is identical to an AMD processor. Likewise, ATI graphics cards are not identical to their NVIDIA brethren. The differences between technologies exist for a reason...they are each approaching a solution from a different perspective. And of course, those different perspectives are what we make our purchasing decisions based upon. Those differences are how we choose one over the other.
And we'd have our reasons for our choice. I primarily choose NVIDIA because it works perfectly well with the Adobe suite. If there were an ATI option that worked better, and/or cost less, I'd certainly consider it. I'm not depressed about it though because the current NVIDIA options works just fine.
As for the Intel/Dell kickback, that's absolutely not what is happening here...this isn't purely a money buyout. NVIDIA actually works with Adobe together on this stuff. They are wholly invested in the Mercury advancements for Premiere Pro. Really, it would be somewhat improper for Adobe to come along at this point and say "thanks for the great ideas, we don't need you anymore" (and that's assuming they could...I can only imagine the sort of agreements that are in place between Adobe and NVIDIA).
Main thing though, is that there has to be a potentially substantial improvement in order for Adobe to redirect their development to OpenCL, even with the efforts of AMD (which you shouldn't automatically assume is in AMD's best interest...how many gamers out there are there, compared to the number of potential Premiere Pro editors?). Do you think Adobe didn't propose some sort of partnership with AMD at some point? The NVIDIA deal had to be better in some fashion. You can continue to make the argument that dollars were chosen over the desires of the consumer, but editors such as myself (and I'm sure hundreds of others) stand in direct opposition to that argument by virtue of the fact that we are actually able to use this technology to speed up our workflow. There have been many articles written about this, so clearly whatever backroom deals and money talk led to our current utilization of the GPU for non-linear editing...it worked out pretty well.
At the end of the day, it must be PROFITABLE for all involved parties. As for anything being a "large" disservice to the buying public, I still fail to follow that bit of logic. I'm certainly not disserviced, since (again) I have a possibility now that I didn't have before (Mercury hardware-assistance). If ATI and NVIDIA cards were equally supported, I'd still have to choose one over the other, and I doubt there would be substantial market pressure to drop the prices on both sides or either side. These market pressures exist now in the gaming world, and it hasn't changed anything.
Last thing....I'm sure AMD and NVIDIA would both have absolutely no problem working together with Adobe to the benefit of all. That's a quaint suggestion, and maybe in a non-capitalism driven world it could work out. But since AMD and NVIDIA are direct competitors on the graphics market, I don't forsee a lot of collaboration in the future unless it happens by virtue of a buyout....which is also unlikely.
LAST last thing...did you know that NVIDIA GPUs are on Mars? Yes, freaking MARS. The PLANET. That's pretty awesome to have on the ol' NVIDIA résumé right? NVIDIA FTW. If it were still the 1960's or 70's, every NVIDIA graphics card would come in a box that said "SPACE TECHNOLOGY!" right across the top, with a picture of a Buck Rogers-esque rocketship next to it.
There is no conspiracy, we chose CUDA for entirely technical reasons. OpenCL did not exist when we began development on this feature.
That is a good bit of history. Thank you for sharing it with us.
Thanks for the response Steve!
All conspiracy theories aside, the NVIDIA logo has been appearing on the side of the production suite boxes for a while now (previous utilizations of OpenGL, focused on the NVIDIA cards). So it would be inaccurate for anyone to assume that NVIDIA was not involved in some way with Adobe's work on this software, or that Adobe doesn't have some sort of ongoing dialogue with NVIDIA that clearly doesn't exist with AMD (ATI).
I think it should be obvious that NVIDIA wouldn't allow it's logo to go parading around on another company's literature or outer box for no reason whatsoever. Likewise, I doubt Adobe would benefit all that much from actually paying NVIDIA to license their logo and nothing more.
I don't really care all that much about the larger ATI vs. NVIDIA debate, contrary to how my previous diatribe might have come across...I'm no "fanboy" of the graphics technology on either side (I am, of course, a complete Adobe fanboy though). I just don't care for the whining and complaining about why one technology is utilized over another, and this pre-dates the Mercury/CUDA situation by a long shot. Every software development has to make choices at some point, and it's really absurd to hear so many presumptions about "how much better it would be if [fill in the blank]."
I'll say it again...Mercury hardware acceleration paired with even just a GTX 470 leaves very little to be desired at the present time. Complaining about the specifics of the implemenation when not even a year has passed on the retail shelves is beyond ludicrous...and this whole thing started almost as soon as Mercury was 1st announced and the preliminary list of cards was listed. What was that...last January, 2010...? It's just getting old.
My advice to anyone feeling slighted by the NVIDIA exclusivity on hardware MPE is to go spend $300 on a card and start editing. All the satisfaction of the Mercury goodness will make you feel much happier about the current state of things.
If it were only $300 for an nVidia card for Mac, your argument would carry more weight. Fact is, it's deeper than just nVidia and ATI. Getting an nVidia card in a Mac isn't cheap ($1200 for the cheapest one as of now) and doesn't work as well as they do in PCs. So really, we're talking about nVidia and Adobe taking the jam out of the donut for Mac users altogether.
Having said that, I have been happy with the performance of CS5 in my ATI-carrying Mac. I still have yellow bars over my HD footage, and it plays smooth, but it played smooth in CS3, too. I'm sure I can't layer as many effects on that footage as you can, but my expectations haven't been spoiled by hardware MPE, since I haven't seen it in person yet. I built my system prior to CS5, using as much information as I could gather at the time, and I've been happy with it. When I build another system down the road, I'll do the same as last time... weigh the pros and cons of PC vs Mac, and all the accoutrements of each that go inside. I was exclusively PC for over twenty years, and I still have my last PC build for the sole purpose of running Audition. The current system is Mac, but I'm not married to (or a fanboy of) any system or cardmaker. But like I said, even with my poor luck in timing, I love using Adobe in a Mac with an ATI card.
My advice to anyone feeling slighted by the NVIDIA exclusivity on hardware MPE is to go spend $300 on a card and start editing.
You're missing the point. (At least, my point.) I agree with the above. It's the only option we have right now, and I did just that myself.
My point is that now that OpenCL is established, let's move forward without the nVidia exclusivity. Any further development for CUDA is not a good idea. OpenCL works on both cards. Switching now for CS6 will not suddenly make my newly purchased nVidia card not work, it'll just open up the possibility for current ATI users to take advantage of hardware acceleration. That's a very good thing, in my view.
Things are never that simple. We have been watching the development of OpenCL with interest, but even as it improves there are still technical shortcomings. We will see.