It's surely happened to you. A software update intended to make an application run faster inadvertently ends up doing just the opposite. Researchers have designed a tool to identify the source of the problem.
Credit: National Science Foundation/Karson Productions
Get with the program.
I'm Bob Karson with the Discovery Files, from the National Science Foundation.
How many times do you feel things get all glitchy after a software update? Often bugs are introduced. Errors that cause programs to (Sound effect: slow motion voice) slo-o-o-w d-o-w-w-n, rather than run faster. Computer scientists call it: "Performance Regression" -- "P.R."
To get to the root of the "P.R.-oblem," debuggers often laboriously check performance counters within the CPU. When the software runs, these counters track how often it accesses certain memory locations, how long it's there and when it leaves. Info that usually reveals telltale signs of aberrations -- anomalies! Uh, something whack. But with hundreds of performance counters in newer desktops and servers, it's nearly impossible to track every status manually and then look for abnormal patterns.
Researchers at Texas A&M, together with computer scientists at Intel Labs, have developed an automated way to find performance bugs super-fast, based on a form of AI called 'Deep Learning.' Their algorithm is taught what normal counter data looked like on an older, glitch-free version of the software, and then looks at the updated version. In tests it was able to pinpoint bugs in a few hours instead of days.
(Sound effect: spaghetti western riff) A slowdown showdown at the CPU Corral.
"The discovery files" covers projects funded by the government's National Science Foundation. Federally sponsored research -- brought to you, by you! Learn more at nsf.gov or on our podcast.
Images and other media in the National Science Foundation Multimedia Gallery are available for use in print and electronic material by NSF employees, members of the media, university staff, teachers and the general public. All media in the gallery are intended for personal, educational and nonprofit/non-commercial use only.
Images credited to the National Science Foundation, a federal agency, are in the public domain. The images were created by employees of the United States Government as part of their official duties or prepared by contractors as "works for hire" for NSF. You may freely use NSF-credited images and, at your discretion, credit NSF with a "Courtesy: National Science Foundation" notation.
Additional information about general usage can be found in Conditions.