In an attempt to promote their own products, Intel today boldly announced that nVIDIA GPUs are 14 times faster than their CPUs. This is not normally something you would hear a company say, but it comes in regards to claims that GPUs with CUDA are faster than standard processors.
"GPU computing with CUDA is a new approach to computing where hundreds of on-chip processor cores simultaneously communicate and cooperate to solve complex computing problems up to 100 times faster than traditional approaches," declares nVIDIA.
This juicy bit of news was released at the International Symposium on Computer Architecture (ISCA) in Saint-Malo, France, in the form of a technical paper titled “Debunking the 100x GPU vs CPU Myth."
The paper shows an Intel i7 960 running application kernels slower than a GTX 280, but only by 14x and not 100x. However, nVIDIA is quick to point out the use of a previous generation graphics card, and that the codes were most likely run right out-of-the-box, without any optimization. In-house testing like this is standard in the industry, and most manufacturers use similar methods to yield the most beneficial results.
The fact that Intel would introduce such a paper shows they have to be worried about the direction the industry is going in, and see nVIDIA as a potential threat.