not an apples to apples comparison - but a dual processor AMD vs my Quad 6700 wasn't any faster, slower infact - tihnk you will have to wait till 9 is released to be sure - there are several dual processor machines in the 'Render Test' post if you want to sort through them and compare - again, doesn't mean 9 won't better utilize 2 than 1 core, will have to wait till its out to get data on it.

I have a i7 and although its fast if money is not a problem get the Xeon it screams !

Dinos

I never had a i7, but I do have a dual cpu machine. I only spent a little over 2,000 on it. It has dual quad core Xeons and an intel board. I am very satisfied with it. My machine is on the render test thread.

j razz

I've got dual quad Xeons and the machine does scream with Vegas, especially once you set the internal preferences to utilize the cores above the default setting.

That being said, it does seem like the technology was ahead of its time when I bought the machine a tad over a year ago (Dell Precision 690). Today's software still seems a bit slow to really take advantage of all the cores. I was placing a bet on the future that I'm still hoping pays off.

From what I've read about the i7, I don't think you'll be disappointed, especially since they brought back hyper-threading with that chip. The dual Quad Xeons might be faster than the i7, but I don't know if the difference is worth $3,000. I think you'd be better off putting that towards some really high end graphics card, or some AV equipment.

Or...considering where our debt-based, slave economy is headed, you might want to save for the rougher times ahead. Then again, if the economy kicks into hyper-infaltion, your money may soon be worth a lot less than it is right now...so spend and be merry!

It has been reported that the Vegas engineers at NAB said specifically that the multi-core routines in Vegas have been optimized in v9.

I've got a now fairly old Supermicro system. Even now it's pretty darn fast. However I bought if for working with 10bit 4:2:2 (Digibetacam) so things may be different with the modern high compression codecs.

Bob.

What exactly does "faster" mean with an NLE?

I suppose it could be broken down into rendering and previewing right? If all you are worried about is rendering then there is a good chance that Vegas Pro 9 will utilize all 16 cores (8 physical + 8 logical) in a dual i7 system. Based on single i7 performance I think it's a good guess that such a system would be a screamer for rendering in VP9.

But I think an important question is whether or not VP9 will use multiple cores for previewing. With VP8 only one core appears to be used for previewing. Even with my dual C2D processor usage generally stays below 50% during previews.

I personally think that optimized multicore previewing performance is more important than rendering performance. In order to edit quickly and efficiently I think it is important to be able to see the timeline playback in realtime or close to it. Rendering can generally (if there isn't a "today" deadline) be accomplished overnight when the computer isn't being used.

So my advice is single cpu system if VP9 doesn't use all cores for previewing.

Dual cpu system if VP9 has been optimized for multicore previewing.

This is what someone in another forum said about Vegas v9:

"I've seen requests for preview improvements mentioned a number of times in this thread. And I certainly agree that's needed.

Well, that was a big thing actually mentioned at NAB. I even talked with one of the SCS engineers for more details about it. They are still using Video for Windows, but they now have optimized multi-core/multi-CPU rendering for the preview display. He said it would improve preview a lot on new quad core machines, and he even felt it would make a noticable difference on my ancient dual-Xeon hyperthreaded machine.

On the dual-Quad machine I played with there, it was playing 4K RED footage in a 2K project perfectly smooth for me at Best (full)."

Thanks to all for your replies, especially to Hulk and John Cline.

My real question interest is in editing productivity - not just raw rendering speed. I completely agree that renders can be left to lunchtime/overnight/breaks for Champions League soccer <grin>. Besides, if preview speed is improved enough then intermediate renders and ram-previews may not even be necessary.

I don't have to order the hardware for a couple more weeks. Even if none of the Beta users want to risk breaking the NDA and speak up, 9 should be released in 11 days. I will be VERY interested to hear first hand if preview really is multi-threaded in the new version. Also interested to hear if the 64bit version of 9 is complete and not as bug-ridden as 8.1 was/is. Without working 64bit support the best parts of the dual-CPU Xeon are probably worthless anyway.

Oh, yeah: also interested to hear opinions about Windows-7 & Vegas. I will have access to the Release Candidate...

Multicore preview seems like the big white elephant standing in the corner of the room when talking about preview performance.

There must be some pretty serious re-working of the Vegas code required to get this working correctly or else I think we've have seen it happen when we saw multicore rendering, which was a few years ago.

Although I'm not a programmer I have done some programming in the past and I can see how multicore preview would be more complicated than multicore rendering. You don't skip frames when rendering each one has to be rendered. So in a very basic method, in a quad core system you could have each core render a frame. Core 1 renders frame one, core 2 frame 2, etc. Then core 1 jumps to frame 5, etc.. Of course this is a relatively stupid way of doing things as this is a very course way to parallel operations. It's only meant as an example. If all frames required the exact same amount of cycles then it would work pretty well and cpu usage for all cores would stay around 100%.

But of course all frames aren't the same in complexity. Plus you have other background process taking cpu cyles from cores, or really throwing a wrench into things hyperthreading, which doesn't have the full power of a physical core.

But what I'm getting at is this stupid method could kind of work for rendering. But I don't think it would for preview where frames are skipped. A core could be working on a frame which isn't ready when it needs to be viewed and is then skipped! CPU cycles were used (wasted) but the frame was never viewed.

Of course a finer type of parallelism is needed. Perhaps breaking each frame into four horzontal slices so that time wasted would only be the time for the "slice" requiring the most processing cycles. Or the screen could be broken down into 16 slices. But as you make parallelism higher you also increase overhead of keeping track of things. And again it gets more complicated for preview since frames have to be displayed in real time. If they're not ready they aren't displayed.

A very interesting problem I know the gurus at Sony will work out for us sooner or later!

- Mark