-
Results of my rendering test: CPU only vs. OpenCL
I had one event on the timeline, 1:41 in duration. Footage was 1920×1080 HD, 29fps. No effects applied to the event- just a straight simple video clip with accompanying audio. My render output settings were MainConcept AVC/AAC, Internet HD 1080p. Project and render settings matched the video properties, but I changed the output settings to variable bit rate of 24,000,000 max and 12,000,000 average. All other render settings were default. This was just a test, and here are the render times:
CPU render only: 5:30
OpenCL render: 5:28So basically I have a tie. Not that it matters, but for those that are curious, my original clip size is 271mb and the final rendered clip size is 156mb.
Obviously my render times are quite long, compared to the clip length, but I do have the VBR cranked up quite high. My hardware specs are listed below, in my e-mail signature.
Dave
Intel i5 3570K Ivy Bridge 3.40GHz quad core
Asus P8Z77V-LK
16gb RAM
ATI Radeon HD7850 2gb
Crucial M4 SSD + Seagate Barricuda 7200rpm
Windows 7 Pro 64 -
5 Replies
-
Thanks Mr.Dave for the information.
-
The Main Concept OpenCL encoder does not support the AMD 7xxx series cards so you will not see any performance difference. The encoder falls back to the CPU only encoder. I pooled others in another forum and got another reporting the same situation.
I had an AMD 5850 and the OpenCL encoder was almost 3x faster. My 7950 card gets the same results as you see.
I reported this to Sony, but for a Month now I get the runaround and no response, which is typical of their “support” from my very brief experience.
Seems like a trivial thing for QA duplicate.
-
what Norman said, AMD 7XXX series = Nvidia 6XX series for Vegas Pro 11 & 12, accelerating some effects and previewing and also helping with rendering of certain Sony AVC templates… that’s all, in my experience OpenCL is not optimized yet for best results, a fast AMD GPU using openCL will render slower than some average 5XX last-gen Nvidia GPU.
It’s alright since Sony engineers starting with CUDA and it shows, OpenCL was added as an after-thought… I wonder with the new AMD and Nvidia rebrands (8XXX – 7XX series) if Sony will take the time to properly support these GPUs through OpenCL / CUDA anytime soon, I know Nvidia is trying with latest 314.22 WHQL drivers but broke support again with 320.00 beta
-
“Sony” uses OpenCL fine, as that is what the Vegas video engine uses and only uses. No CUDA. Even Version 11 works with my new card and version 11 was before the new AMD cards came out.
The Main Concept AVC encoder is not written by Sony. Main Concept OpenCL/CUDA are pure encoders running on GPU written via software in the OpenCL/CUDA language respectively.
Sony AVC with GPU support is OpenCL for AMD and CUDA for Nvidia, and it works fine with OpenCL. However Sony AVC is not a pure OpenCL/CUDA encoder. Meaning only some items are GPU accelerated and does not get much speed boost. I believe only motion searching is accelerated. This is similar to what someone is doing in the x264 project.
Encoders in general are poor matches for GPU as increased parallelism compromises quality.
-
Is there any news on this?
I bought the 7870 under the illusion that it would render extremely quickly in Vegas, but have found the same problem as mentioned here. I even saw benchmarks saying the render times were fantastic with the 7870…
Can’t believe it’s not supporting open CL on this card properly. If I were to send the card back, what card will be guaranteed to improve render times and offer similar graphical prowess?
Reply to this Discussion! Login or Sign Up