A new system called Fractal achieves 88-fold speedups through a parallelism strategy known as speculative execution.
Source: MIT/ CSAIL: Suvinay Subramanian, Mark C. Jeffrey, Maleen Abeydeera, Hyun Ryong Lee, Victor A. Ying, Joel Emer, Daniel Sanchez
As is commonly known, the chips in most modern desktop computers have four cores or processing units, which can run different computational tasks in parallel, but that the chips of the future could have dozens or even hundreds of cores, and taking advantage of all that parallelism is a stiff challenge, reminded researchers from MIT’s Computer Science and Artificial Intelligence Laboratory.
Along these lines, in order to make parallel program run much more efficienct and easier to code, MIT researchers have developed Fractal, a system that they proven enables 10-fold speedups over existing systems.
Specifically, the team said that in tests on a set of benchmark algorithms that are standard in the field, Fractal frequently enabled more than 10-fold speedups over existing systems that adopt the same parallelism strategy, with a maximum of 88-fold.
The Fractal system achieves 88-fold speedups through a parallelism strategy known as speculative execution. (Source: MIT)
Click here for Technical Paper.
Click here for MIT summary news.
So you aren’t aware that some of the GPUs used on graphics boards have over 3000 cores each? There are strong limits on all of them doing completely independent things at the same time, though.