Chips with collaborating CPU and GPU lead to faster processors
By Ben Coxworth
February 7, 2012
Want to get your computer to run faster? Well, consider its graphics processing unit (GPU) and central processing unit (CPU). The two work away at their own tasks, each one rarely helping the other shoulder its workload. Researchers from North Carolina State University, however, are in the process of changing that. They have already developed a technique that allows GPUs and CPUs located on a single chip to collaborate on tasks, and it has resulted in a processing speed increase of over 20 percent.
Placing GPUs and CPUs on the same chip is not entirely new - chip manufacturers have already been doing this for some time, in order to lower manufacturing costs and increase computers' energy efficiency. Their being together like that, however, has made the new technique possible. Basically, the process allows the two systems to share computing tasks, each one concentrating on what it does best.
"Our approach is to allow the GPU cores to execute computational functions, and have CPU cores pre-fetch the data the GPUs will need from off-chip main memory," said Dr. Huiyang Zhou, an associate professor of electrical and computer engineering. "This is more efficient because it allows CPUs and GPUs to do what they are good at. GPUs are good at performing computations. CPUs are good at making decisions and flexible data retrieval."
In tests performed so far, the technique has been shown to improve processor performance by an average of 21.4 percent.
Source: North Carolina State University