Vegas Pro 10 - GPU Rendering Results

Comments

LSHorwitz wrote on 10/13/2010, 7:15 AM
My 2 cents:

You can not be sure what the exact reasons are for V10 to run slower than V9 if you are not intimately familiar with the software development, the software design, the coding optimizations used (or not used), etc., but a couple simple conclusions might be worth pondering:

1. Adding new features typically adds to the software processing burden. "Bloatware" is a term you will find for programs which try to be all things to all people, and in the process give up their performance efficiency.

2. Getting a product to market and then doing many iterations of bug fixes, updates, etc. is an indirect indication of a rushed development effort. There is a reason why Vegas 9 has versions a,b,c,d, etc. whereas Edius Neo has exactly ONE release with no bugs I have encountered.

3. Beta testing and debugging costs big bucks. Some companies do it with in-house and / or hired testing teams. Sony relies on user complaints, and makes fixes after the fact.

So, the bottom line, in my view, is that 3D video and stuff most of us have absolutely no interest in using create this new V10 product, and the majority of us see a degradation in speed.

Larry


dxdy wrote on 10/13/2010, 8:53 AM
I have a Q6600 with 8GB RAM, and a GEFORCE 8500 video card with 512GB RAM, running Win7 64bit Home Premium, separate hard drives for programs/OS, input, and output.

I took a short .veg file, with some SD, some HD, a large .gif, a huge .jpg (4288 pixels wide), some generated text, some transitions, some Cinescore generated .wav files and ran a little test.

In 2 minutes, this configuration using GPU finished 15% of the job.

In 2 minutes, this configuration CPU-only did 19% of the job.

I replaced the old 8500 with a spanking new GTS 450 with 1GB of DDR5, and in 2 minutes using GPU, it did 23%.

I take it to mean my new card gained me 20% or so over the Q6600. I don't claim this was an ideal test or comparo, but indicative.

Two interesting things as I did all of this:

1. With the 8500 installed, it took up to a minute for the choice of how to render (CPU vs GPU) responded. With the 450, instant response.

2. I looked at system utilization during the tests. Each seemed to be in the 80% range. But memory usage never exceeded 2.2GB. I may not understand what the Win 7 memory usage number means, but I was surprised to see it using only 2.2 GB. (Yes, I am running the 64 bit version of 10.0a).

edit: I used the Palit GTS 450 card, which claims core clock speed of 930MHz, other vendors (as seen on Newegg) run as low as 750MHz.
Fotis_Greece wrote on 10/13/2010, 10:27 AM
I now use Premiere Pro CS5 (first time I am not updating to new Vegas version) and find the MPE unbeatable.
It does not speed up all effects but but it is dramatically faster when compared to it turned off.

It's the first time in my life I can render a 60 minute HDV project in about 27 minutes including the audio ac3 rendering selecting the High Quality presets and options at 25 mb/s CBR. The project contains a few color corrections and about 10 third party transition as well as some non accelerated effects.

I must confess that I like Vega s a lot and the interface and audio option is better that PP but the speed of PP is so great that now I have to learn it
MelvinG wrote on 10/13/2010, 11:29 AM
I too love Vegas especially when I switched over to the 64-bit version but I was tempted enough to try PP CS5 to test out its GPU acceleration. However, though I did the posted basic hack adjustment to get the software to recognize my GTX 480 (instead of having to buy a Quadro) and utilize it in the rendering, it didn't work for some reason on my system though others claimed to have success with this procedure.

The lady at Sony tech support really seemed to acknowledge the lack of proper GPU utilization in VP 10 but she equally tried to reassure me that the engineers are aware of this issue and are dedicated to improving the performance over the next several iterations.