CPU’s and GPU’s what is doing what at what stage…?

[Sonic 67]

[Stephen Mann] “There’s nothing in the GPU API that tells it how to do simple computing, like 1+1.”
You are horrible wrong. The present day GPU’s are actually full feature CPU’s, with massively parallel cores. They can be programmed in C++ like any CPU, do floating point calculations single or dual precision, used in heavy scientific work…
They are even the base of several of the top supercomputers (nVidia Tesla, AMD).

[Stephen Mann]

I said there’s nothing in the GPU API. Not the GPU. The supercomputers using GPU’s massive parallel processing don’t use the same drivers that are used in video processing.

Steve Mann
MannMade Digital Video
http://www.mmdv.com

[Sonic 67]

This is getting to offtopic.
Your information is dated, probably pre-2012. Read more about OpenCL and CUDA API’s. They are full blown computing environments, no different that x86 CPU ones, just running on a different processor.

[Norman Black]

[Stephen Mann] “There’s nothing in the GPU API that tells it how to do simple computing, like 1+1. But point it to the data for a series of frames and ask it to compress them into a GOP, then it’s right at home. Theoretically faster than the complex CPU.”

Wow. Educate yourself please.

[Matt Dunphy]

In 2013, I upgraded to Vegas 12, got a new computer & a video card specifically to utilize GPU-assisted rendering.

In my experience, enabling the GPU actually resulted in slower render times. Whenever effects were being rendered, my CPU’s usage would drop from 100% to a fraction of that, presumably having the computer send all the work to the GPU. Maybe it’s helpful if you’ve got an older processor, but ultimately, I felt like I’d blown a couple hundred bucks on an unnecessary video upgrade. Live n’ learn.