New technique improves computer processor performance

New technique improves computer processor performance

Unique method takes advantage of individual CPU and GPU strengths

BY JEFFREY BAUSCH

Engineers at North Carolina State University have found a way to significantly boost the performance of fused-together processing units.

Example of processor fusion: Phenom II CPU combined with DirectX 11 GPU (Via: pcgameshardware.com)

Preliminary tests show a boost in performance of 21.4%. Here’s some background on the project:

Concept

The team fused a graphic processing unit (GPU) and central processing unit (CPU) on the same die to take advantage of each other’s strengths. Typically speaking, a GPU is good for executing computational functions while a CPU is better equipped to perform tasks like data retrieval and decoding.

“Chip manufacturers are now creating processors that have a ‘fused architecture,’ meaning that they include CPUs and GPUs on a single chip,” says Dr. Huiyang Zhou, an associate professor of electrical and computer engineering who co-authored a research paper on this project. “This approach decreases manufacturing costs and makes computers more energy efficient.”

“However, the CPU cores and GPU cores still work almost exclusively on separate functions,” Zhou adds. “They rarely collaborate to execute any given program, so they aren’t as efficient as they could be. That’s the issue we’re trying to resolve.”

Testing

Conducting their preliminary tests on a simulated AMD CPU with L3 cache, the NC State team focused on the CPU’s core strength: fetching/decoding data. In this case, though, they had the CPU do this for the GPU. The GPU’s job, in turn, was focused primarily only executing assigned tasks.

“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,” Zhou explains.

So, the CPU is focused on making sure the GPU has the data it needs, and the GPU is focused on executing functions. The result: An average improvement of 21.4% in computer processor performance.

Outlook

This process was not possible in the past because GPUs and CPUs were on separate chips; the two simply never “spoke” to one another. The reason why this is exciting is because it’s a step toward us better understanding how heterogeneous system architecture (HAS) will come to influence tomorrow’s computer processor market.

The Paper

How, exactly, the team shifted the focus of the two processors will be presented in a paper entitled CPU-Assisted GPGPU on Fused CPU-GPU Architectures on February 27 at the 18th International Symposium on High Performance Computer Architecture, in New Orleans. The paper was co-authored by North Carolina State University Ph.D. students Yi Yang and Ping Xiang, and by Mike Mantor of Advanced Micro Devices (AMD).

Research for the project was co-funded by the National Science Foundation and AMD. ■

Press release via : news.ncsu.edu/releases/wmszhougpucpu