"Any word on whether there is any chance at all we'll be seeing this functionality in Vegas soon?"SCS has announced its intention to use OpenCL, see: Sony NAB Announcement 10d and more

I did a quick Google search of "Quick Sync" & "OpenCL" and the relation between the two is unclear. For that matter, if you read the reference thread, exactly how SCS is planning to exploit OpenCl is also unclear.

...Jerry

The issue with new tech is that it's unproven & well, it's new. I'm sure intel has no "real world" tests to show how it works. I've seen intel have a dedicated programmer take a closed source program (ie quake 4) & do a complete graphical overhaul with everything 100% real time ray tracing, shadows, etc. just to show off how well their big processor setups work. They haven't taken Vegas, Premiere, Avid, FCP, etc., done a complete playback/encoding overhaul and implement it with their code to show advantages.

They're not interested because, odds are, it won't help much (much like the Nvidia CUDA stuff).

That anandtech report seems to indicate notable reductions in rendering speeds and very little quality loss.

If we're willing to trade off quality for speed let's go back to SD & our modern systems will handle that so fast we wouldn't need to upgrade for a decade.

Want want better looking AND faster. I could transcode a BD to HDV in real time with my video camera if I wanted. That doesn't even cost me anything more then two tapes & my camera.

Take a look at their example shots. The "old" rendering model (CPU, nothing fancy) *IS* noticeably better looking then the rest.

Like he says in the article, if you want to do small stuff for smart phones, ijunk, etc. it's great. Based on their numbers, here's the total rending times of their BD:
Nvidia GTX 420: 50.35 minutes
AMD Radeon 68/6970: 58.07 minutes
Sandybridge: 33.87 minutes
i5 x86 (our render type): 58.07 minutes
Phenom 2 X6 x86 (our render type): 61.84 minutes

That's in the order of lowest render quality to best (and speed for the x86's). You're cutting your transcoding time to ~1/2 but you're taking a noticeable quality hit. As far as we know, no FX/pan/crops/tack motions/etc. can be done in this, those which can make the transcoding time meaningless (if you have a 10 hour render, what's an extra 60 minutes in the long term: not worth the loss imho).

Thing is, some of us are rendering on-site at expos and places like that. IF it IS possible to have it work in tandem with lightweight effects (a video bug here and there, perhaps a short lower third, and open/close transitions to prerendered spots) then it would be of immense help for that type of work. Not all of us need the most pristine quality in our HD renders, just something that suffices and looks notably better than SD (or CUDA for that matter, which is pretty bad I think). Speed IS VERY important in those sort of situations, though some editing is necessary still.

I really, really hope this eventually crosses into the realm of possibilities. ArcSoft's apparently got it going in their MediaImpression editing software, but I don't think that's a very good package. However, the MediaConverter engine (which MediaImpression uses) IS very good, and EXTREMELY fast when paired with a Quick Sync capable machine. This is what I hope we see in Vegas eventually, as Vegas is way better in every other way however! :-)

From what I read Quick Sync will only work if you are using the IGP.
I'm not 100% certain of exactly what all the implications of this are though. Depending on the chipset the IGP is disabled anyway and is also not usable on the 2600K processor.

Bob.

I keep reading that CUDA - OpenCL or other GPU accelerated compressor give bad results in terms of quality... Instead I think that they are just as good as the software they run is...

If the software codec implementation that runs on the GPU is bad, the result will be bad... there may be limiting factors to the implementations.

I think that full software codecs for "standard" cpu have been tuned for best results for years, we just need to wait a bit more...

I'm wondering if the accelerated stuff is worse because of the hardware limits. All GPU's use 8 bits per pixels, there is no limit on the software mode because it just makes things up. So it gets better values by not rounding values until after the math is done.

I just wanted to update regarding this situation now that I have received my new notebook, which is powered by an i7-2620M dual-core Sandy Bridge processor.

Using Sony Vegas Pro 10.0d, a render of a particular project (using 1920x1080 30p source video rendered out to 1280x720 30p) takes approximately 3 minutes (per 1 minute of source video).

Using Adobe Premiere Pro CS5 with Quick Sync support, a nearly identical project using the same media and parameters and very similar editing takes just 1 minute (per 1 minute of source video).

I hate that this is the case, because not only is the editing work easier and quicker in Vegas, but I just love Vegas as a whole (and have long been a huge supporter of it). But considering that Premiere is literally THREE times quicker thanks to Quick Sync video support, it will make our expo trips that much more productive. As a result, I fear I have no choice but to sacrifice the marginally longer editing procedures and reap considerably expedited render times as a result.

If anyone at Sony happens to be reading this, please, PLEASE make Quick Sync a reality. It really IS that much quicker, and quality differences are almost negligible.

-Steve

Edius uses QuickSync, but only if you are using the Intel graphics, not if you are using a third party card.

Edius can use quicksync, even when you have a dedicated card too!

When can we use Quicksync in Vegas? Any plans?