Sony Vegas Pro 12 Benchmark
Welcome to my Sony Vegas Pro 12 Benchmark page. The purpose of this page is to analyze the preview and rendering performance of Sony Vegas 12 Pro for the following reasons:
For information about the test and how to run it please scroll to the area below the results table.
March 13, 2014 -
November 30, 2013 -
Investigation of GT2 GPU performance with Sony Vegas Pro 12
You can download this investigation here.
Results for the
original test using the Sony Press Project veg file
Systems are ranked based on the rendering rate of the XDCAM EX score
If you are considering buy an AMD 7xxx
series graphics card or higher please realize that OpenCL using the Mainconcept AVC/AAC
format does not work. If
you select "OpenCL" the render will occur but it won't be GPU accelerated.
Sony has been notified of this issue but claims they can't reproduce it. It happens with my 7770 card and with Norman's 7950 as you can see below. Perhaps if we can document the issue with more users Sony will take notice.
Due to the sheer size and buying power of the gaming community, extremely powerful GPU's have become available at very reasonable prices. With the release of version 12 Sony added extensive GPU support in the form of OpenCL (Open Computing Language) to leverage the power of these modern GPU's. Currently only video cards (GPU's) with a certain minimum of computational ability (compute power) are supported by Vegas. Sony's GPU acceleration page that includes Vegas capable video cards is located here. Here's what Sony has to say about the new GPU acceleration in Vegas Pro 12.
"Significant portions of the application were entirely reworked, resulting in an enhanced and more creative editing experience. Over 45 effects, transitions, generators and compositors are GPU-accelerated in Vegas Pro 12, as well as a substantial amount of built-in video processing such as crossfades, fades, alpha compositing, framerate resampling, interlace processing, pan/crop, track motion, opacity, fade-to-color, and multicamera display."
In order to properly analyze these Vegas benchmark results it's necessary to understand a little bit about how Vegas works when it comes to previewing and rendering the timeline. The process of creating a final output file from the Vegas timeline can be thought of as a two step process. During the first step all of the video clips (text and/or other graphic elements) along with all of the transitions, effects, transforms, etc must be "assembled" into a single video stream. I don't know if assembly is the proper word for this operation but it seems to be a fitting term so lthat's what I'm going to call this process. With Vegas Pro 12 the Sony engineers have reworked the code of Vegas so that many of these timeline operations (see above quote from Sony) are accelerated by the GPU during this timeline assembly phase. If your CPU and GPU are powerful enough to keep up with this assembly of the timeline in realtime then you will see a perfect preview at full playback frames per second (fps). If not, then frames will be dropped and the video preview will stutter and lag. This is why various preview qualities are available in the preview window. Lower preview qualities and resolutions make it easier for your hardware to keep up with the assembly of the timeline. The CPU and GPU work hand-in-hand in this assembly process. One of the purposes of this website is to find the correct balance of CPU and GPU power so money is not wasted on one or the other unnecessarily. For example, the GPU is constanting waiting on the CPU or vice-versa. This optimum balance of CPU and GPU may well be different for different users depending on their workflow.
When you press the play button in Vegas the timeline is assembled and you view the project. Note that GPU acceleration must be turned on in the "Options>Preferences>Video" tab for Vegas to use the GPU when assemblying the timeline. Also note that if you turn on or off the GPU acceleration in Preferences you must restart the computer for the setting to take effect. If you select "Render As" then the assembled timeline video stream is then sent to the transcoding codec of your choice. As far as I know only two of the available video output formats are GPU accelerated. The first of those is the Mainconcept AVC/AAC codec. In order to enable GPU acceleration for Mainconcept rendering the "encode mode" must be selected in the Mainconcept Custom Settings dialog box. The choices are CPU only, Cuda (specifically for nVidia video cards), and OpenCL (for AMD video cards). The other GPU supported transcoding codec is the Sony AVC/MVC format. And again GPU acceleration must be selected in the "Custom Settings" dialog box. Choices include "CPU Only," GPU if available," and two Intel Quick sync options, "Quality" and "Speed." All other render to formats are rendered using only the CPU. Remember that assembly of the timeline is still GPU accelerated but all "Render As" options, but only the two mentioned above are Assembly and Transcoding GPU accelerated.
By monitoring CPU and GPU usage during preview (assembly of the timeline) and rendering we can analyze what is actually going on "under the hood" of Vegas in terms of how it is using available hardware resources. I suggest using GPUz to monitor GPU load. GPUz can be run from the executable and therefor does not have to be installed. CPU loading can be viewed using Windows task manager. The test script is the Sony press project which can be downloaded from a link about 2/3's of the way down this page. This is an assembly heavy project so it will make good use of GPU accelerated systems.
Running the Test
The easiest way to report your scores is to use the excel based reporting
form I created that you can download here:
Benchmark Reporting Form.zip
Just download it, extract, fill it in and e-mail to me at email@example.com - Remember to restart computer when changing GPU acceleration setting in preferences.
Also look closely at the Excel sheet title columns to make sure your are running the tests correctly.
Also, can you send me an assessment of the stability of your system? Rock solid (never crashes), Very stable (crashes occassionally on 1+ hour project with third party plugs), Stable (crashes on 1+ hour projects without third party plugs), Moderately stable (usable, but crashes a few times a day), Unstable (crashes all the time for no apparent reason)
Explanation of "Eff" in the results table
Eff or efficiency shows how many MHz it takes from one of your cores (only physical cores count, not logical) to render at 1fps. So my 2500k requires 2173MHz (out of a total of 4x4200=16,800MHz) to render the Mainconcept part of the test at 1fps. My older Core2Duo laptop is less efficient requiring 2764MHz to render at 1fps.
Downloading the raw excel data