Sometimes Canadian Customs lets things sit on the shelf for about 6 weeks if a tag is required, and if there is no tag you may never get the stuff.
JJK

Hey thanks for the stats/info, guys - I'll have to look at buying one of those quad core systems myself next month or so... render times are a major bottleneck to productivity, so if I can get a 20-30% improvement in renders over my asus p4 2.8 win2k system, that would be great... sounds like the new systems will?

anyone have an updated benchmark page, eg render times for say 30-60 minute NTSC video avi clips, using V7 and P4 2.8 etc up through these latest cpu setups?

there was a benchmark page I saw awhile back, it would be good to see something specific w/V7...


ken

Ken,

http://www.hyperactivemusic.com/msprofiles/sony%20vegas%207/sony_vegas_7_audio_benchmark.htm

As posted above I setup a small Vegas 7 benchmark page.

One person I know tested their overclocked to 3.2GHz Quad system and it ran the benchmark in 122 seconds, vs 146 seconds for my Dual core at the same frequency.

Although that's an improvement of about 20% I was hoping to see more like 50% or more improvement since there is 100% more cores when going from the dual to the Quad.

When I run this benchmark the my processor usage is slammed at 100%, full out for both cores. The fellow that ran the test with him Quad was only getting about 70% or so CPU usage. Seeming to indicate Vegas 7 (update c) not taking good advantage of Quad cores. Until a few other people with Quads run the SAME test as someone with duals we won't have an answer.

I moved from a P4 3.06 to a Core 2 Duo at 3.2GHz and the difference in performance is like night and day. It was the best $220 (for the processor) that I ever spent on a hardware upgrade.

If anyone runs my test I will post the results in the table right away. Then we'll have some quantitative analysis of the performance various processors.

Just one person needs to run my test and we'll know for sure how much of an improvement the quad is over the dual.

The test uses DV rendered to MPEG-2, some moving text, a moving PIP, and a few transitions. Just the basics one might use in assembling a video. I guess I should have included some color correction but I can always add to the script.

- Mark

> The fellow that ran the test with him Quad was only getting about 70% or so CPU usage. Seeming to indicate Vegas 7 (update c) not taking good advantage of Quad cores. Until a few other people with Quads run the SAME test as someone with duals we won't have an answer.

I just ran your test and there is something VERY STRANGE about it. For that particular benchmark, Vegas DOES NOT make full use of the QuadCore for compositing. The test took 2:38 and the 4 cores NEVER got above 78% and MOST of the time during the tough rendering they stayed around 30%. That's right; as the rendering got tough, CPU utilization went DOWN!

I tried to reproduce this from scratch and could not. I created a moving PIP, moving Text, 3D transition, etc. and Vegas used 100% of my 4 CPU's in every case. But for your version of the render test it did not. This is very strange. there is something about your render test that makes Vegas go single-threaded!

My initial tests with Vegas 7 were done with the standard render test on the VASST web site. In that test all 4 cores run at 99%. My Pentium 4 3.0Ghz did that test in 1:35, the AMD X2 in 0:39 and the Intel QuadCore in 0:14 which is 2.8x faster than the AMD and 6.7x faster than the P4! Hence my claim of 600% gain.

Doing straight renders from HDV to MPEG-2 is also blazing fast on the QuadCore. This is the real test for me because I shoot HDV. HDV renders were taking 3x longer than real-time. Now they are 1.25x longer. I am very happy with the "real-world" performance of my QuadCore.

> Just one person needs to run my test and we'll know for sure how much of an improvement the quad is over the dual.

I'm afraid not. Your render test seems to have uncovered a condition in Vegas where it forces Vegas to only use a single thread. This is NOT a valid test for Dual Core/Quad Core processors. A P4 single core will win on clock speed alone every time.

~jr

JR,

My test DOES utilize dual cores as my system is utilized 100% and is MUCH, MUCH faster than my old 3.06P4.

But it obviously IS NOT utilizing Quads as indicated by your result and the other fellow that ran the test.

Perhaps the test has to be created in Vegas 7. This test was created in Vegas 5, opened in 7, and then saved in the 7 format. I'm going to rebuild the test in Vegas 7 and then upload it again.

Could you provide me with a link to the VAAST test to which you are referring? I'm not sure exactly where it is located?

Thanks for running the test! I knew something strange was going on and now I have a clue as how to fix it.

- Mark

JR,

New test script here: http://www.surfcitybeachpatrol.com/newvegas7benchmark.zip

Just a DV clip, some transitions, color correction, moving text, and a PIP. Does your system make use of all four cores on this one?

No audio, just render it out to "DVD Architect NTSC Video Stream."

Mark

Mark, That new render test had all 4 cores pegged at 100%! I contacted Sony and they said that the old benchmark project had Film Effects on it which is a single-threaded FX. ;-) Whenever an FX has a yellow icon it will be single-threaded. This is good to know for the future.

My time for that new test was 3:02. What was yours?

The VASST render test is here.

~jr

JohnnyRoy, could you run the Cheroxy's veg at
http://s92274348.onlinehome.us/Video/rubix.veg
with project properties and view preview at good or best setting. I'm interest in your results.

Thanks in advance.

JR,

Okay, now we've got it!

My overclock Core 2 Duo ran my bench 4:32, yours 3:02. Now that Quad is DEFINITELY cooking!

Brief Analysis:
Your system requires 234MHz to render at a rate of 1 frame/sec
Mine requires 420Mhz to render at a rate of 1 frame/sec

Taking into account your system is running at 2.66GHz and mine is running at 3.2GHz it appears that going from 2 cores to 4 has gained you an 80% performance increase. That's darn good considering 100% would be theoretical and is not going to happen due to various quad core scheduling overhead, memory bandwidth remaining the same, etc.. i.e. everything in your system did not scale times two, only the cores.

So an 80% boost in Vegas over an already fast Core 2 Duo processor is definitely a more than significant performance increase.

I'm going to rewrite my test script but I want to use some HDV footage to create a more forward looking test. Do you know where I can find a little HDV footage I could use?

Also, it's weird how 1 non-multithreaded effect shuts down the whole project. Definitely something to keep in mind. Hopefully Sony will get them all coded to multithreading soon.

- Mark

Think you'll find some here: http://www.vasst.com/?v=HDV/hdvportalnew.htm
Randy

Johnny speed er Roy , with that kind of power how are you approaching the editing of your HDV stuff ?

I have found that editing HDV on the Quad is as smooth and fast as editing plain DV was on my old 3.4gHz P4.

John

> I have found that editing HDV on the Quad is as smooth and fast as editing plain DV was on my old 3.4gHz P4.

I agree with John Cline. I don't treat HDV any different than I did DV. I just capture, edit, and render. The productivity savings of using a QuadCore for HDV shooters far outweighs the price. In fact, the 2.66Ghz DualCore is $500 so a Quad for $1000 is the same as buying two DualCores except that a dual CPU motherboard will set you back another $500 so the QuadCore is actually cheaper in the long run.

> JohnnyRoy, could you run the Cheroxy's veg

I can Preview the Rubix veg at around 6~8fps but in Good or Best mode I can't even render a single frame. Vegas becomes unresponsive. (very strange)

~jr

Thanks JohnnyRoy for running Cheroxy's veg. The difference in render settings is interesting, especially since your quad-core has the power! The question of Sony's media or 3d space or both causing the freeze has my caught my interest. The v7c speed improve may have a hidden problem with memory/resource management or it could be a simple flag setting in the media buffer, but it is a performance killer.

http://www.hyperactivemusic.com/msprofiles/sony%20vegas%207/sonyvegas7hdvbenchmark.htm

New test script using HDV video, SD video PIP, some transitions, text, hi-res photo w/zoom.

> New test script using HDV video, SD video PIP, some transitions, text, hi-res photo w/zoom.

That test took 2:07 on my QuadCore when rendering to MPEG-2 DVD Architect Widescreen video stream but it mostly only used 58% of the CPU's. It would occasionally jump to 100% but it spent most of the time well below that, sometimes dipping as low as 48%.

So that's not a good QuadCore test either. Which means that Vegas cannot always take advantage of all 4 cores. :(

~jr

back in September 06, Hulk posted a link to an early anandtech test that showed Vegas 7 with the qx6700 and other dual cores.

read it here:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2846&p=2

based on that I started watching the qx6700 and just ordered mine for $1009. I'm moving up from an amd x2 4800 and believe I'll see quite an improvement. I can't wait to see it fly! Guess my render coffee breaks are about to get shorter.

Sean

JR,

Try rendering out to HDV 1080-60i template and see if you get better usage.

Mark

Mark, I'm not sure I got better usage rendering to HDV 1080-60i. There were some good peaks at 100% but it still had a fair amount of usage in the 50% and 60% range. I guess the threading support depends more on the FX and not the render target. It took 2:40 to render to HDV M2T.

~jr

2:40 isn't bad. Your quad is running 2.66GHz, right?
I got 2:45 but my dual is running 20% faster than your quad. So you're still about 21% faster than me on this test, which obviously isn't utilizing all four cores. It DOES utilize both of my cores the entire way through.

Strange since this script is pretty typical of one that might be used to put together a quick HDV video. Of course it's very cpu intensive as I wanted to stress the cpu. But still, some HDV footage, color correction, a SD PIP overlay, a few transitions, and a still with a zoom is pretty basic stuff. A fair test I think. There's nothing really crazy in there.

This is the first version of Vegas I think really take full advantage of dual cores, and *pretty good* use of quad cores. I think Sony will continue to work on the multithreading and get all four of your cores churning and burning soon enough.

Just for laughs I ran the benchmark on my old PIII 1.2 mobile. 29:34!!!

Even taking out the multi-core aspect, core to core at the same clockspeed the Core 2 Duo is twice as fast as the PIII.

To put it another way it would take a single core PIII running at 12.8GHz to equal the performance of my Core 2 Duo. Or a dual core PIII running at 6.4GHz.

I remember writing a little article on cpu power and how much is necessary for video editing about 7 or 8 years ago and I said I'd be pretty satisfied with a 10GHz PIII. I guess that estimate was pretty good for standard definition, then HD came along!

- Mark

This test seems to suggest that the decoder and encoder logic for M2t is limited to 1 or 2 cpus if JohnnyRoy 4core only peaks a few times.

Or that something else is the bottleneck.
On my dual Xenon encoding to mpeg-2 in V6 all 4 cores run balanced at around 70% from a single IDE drive, 50% from a 1394 drive and 90% from RAID 0 drives.

A better test of raw CPU grunt might be 1 track with 4 GB FXs in a chain, coming from m2t, encoding to m2t.

If you want to compare disk I/O impact render from uncomp HD to uncomp HD with no FXs.

Neither of these are very real world tests but they isolate the various system components, I think.

Bob.