Gpu vs Cpu for rendering

[ExtreamChaozZ]

Iv just started doing a fair amount of rendering in Sony Vegas pro 13, but I have noticed that in the rendering settings I could choose to use my cuda cores if I so wished to.
So my question is what is better for rendering 2d mp4 files averaging 20 to 30 minuets? My cuda cores or using my CPU cores
I have a gtx 970 acx2.0 ftw and a i7 4790k both at stock for now.
My rig is all air cooled as well in the coolermaster had x case for optimized air flow.
I ask this as I have seen mix reviews about it and I am quite confused on the subject, thank you.

 

[Wisecracker]

The GTX 970 should be faster than 'software rendering' on the CPU, but Vegas tends to favor Radeons over nVidia in GPU compute.

So ... it will likely be a bit faster -- but not by a great deal.

 

[ExtreamChaozZ]

Alright then, thanks mate

 

[velo3100]

The gpu is better as I have tried this using premier pro on my laptop. The gpu is able to render the clip much faster than my cpu can. Also the program responds much faster as well as you start doing things that are complex.

 

[spdragoo]

It's not usually an "either-or" situation; the software that can take advantage of CUDA cores does so by using the CUDA cores in the GPU to supplement the main processing in the CPU. So, if you can do so, definitely let it use your GPU.

 

[Wisecracker]

It's not usually an "either-or" situation; the software that can take advantage of CUDA cores does so by using the CUDA cores in the GPU to supplement the main processing in the CPU. So, if you can do so, definitely let it use your GPU.

Actually, 'software rendering' on CPU cores sees minor 'percent' increases, generally scaling with the number of cores, and to a lesser extent, 'hyper-threads'

GPU compute sees scaling in 'multiples' versus software rendering on the CPU, generally at a minimum of 3- to 4X faster --- in Adobe CC (using multiple video cards) scaling is seen at 8- to 10X faster, or more, depending upon the container and/or effect(s).

The OP, at no fault of his own, has run into the Maxwell arch 'conundrum' -- meaning that the cards reduced total CUDA cores versus Kepler & Fermi by 1/3. The end result of this is a card like the Radeon R7 260X is faster than the GTX 970 in Vegas GPU compute, and the older Kepler & Fermi cards are infinitely faster than the 970 using CUDA.

So ... (getting into grey areas, here), GPU compute will be faster on the GTX 970 than software rendering on the CPU ... but, not at the 'highest' levels. For the sake of this example, let's just say it will be 2X faster, depending upon his container (MP4/AVC) and bit-rate.

Now -- to make it even more interesting -- the OP just needs enough CPU to keep the GPU working. Over-clocking the CPU will likely gain little, if any benefit --- certainly as opposed to OC'ing the graphics card. Also (and I'm not sure he can do this), the GTX 970 can do Open CL/GL --- it may be advantageous to him to utilize the O/GL-CL compute of the 970 over the flat-out 'CUDA-compute' of the card (I believe you can do this is Adobe CC -- not sure about Vegas).

You will all be tested later 😀