What is GPU Hardware Acceleration? And what will you benefit from it?

Before discussing about GPU hardware acceleration, here is a chart indicating how different GPU performs based on diverse of collocation of different encoding / decoding built:

The most distinguishable conclusion from this table is Hardware Decode & Encode works better than Software Decode & Encode. Next let's learn about what is GPU hardware acceleration, how does video hardware accelerator work and what will us benefit from this technology?

Hardware accelerationis the use of hardware to perform some function faster than is possible in software running on the general purpose CPU. And GPU hardware acceleration is one of two major uses by this approach.

While by far NVIDIA CUDAstands in the centre spotlight of GPU hardware acceleration stage. Through distributing CPU memory to GPU, consumers video decoding / encoding speed is gallopingly stirred up thanks to CUDA-enabled GPU's parallel computing architecture.

What will you benefit from GPU Hardware Acceleration?

It is pointless to talk about hardware acceleration without particular testing project. As for most consumer users' concern, video transcoding is everything for CUDA hardware accelerator technology.

Take NVIDIA GeForce 8 as an example. Before Hardware Accelerators getting popular, traditional CPU-based transcoding usually makes the primary processor do all the work. And it seems multi-core CPU really done these things competently. However, comparing with thousands of GPU processing threads, 8 cores or more CPU still has great many limits. GeForce 8800 has 128 calculation units, which means it is possible to process video decoding / encoding through 128 threads (of course usually we do not have that many pixels to coop with).

For better understanding of the advantages GPU Hardware Accelerators brought us, we can compare one video transcoder software's performance before and after utilizing CUDA enable SKT.