Reinforcement Learning Uncovers Silent Data Errors

For top-performance chips in huge data centers, math will be the enemy. Because of the sheer scale of calculations occurring in hyperscale data centers, working around the clock with thousands and thousands of nodes and huge quantities of silicon, extraordinarily unusual errors seem. It’s merely statistics. These uncommon, “silent” knowledge errors don’t present up throughout typical quality-control screenings—even when corporations spend hours on the lookout for them.

This month on the IEEE International Reliability Physics Symposium in Monterey, Calif., Intel engineers described a way that uses reinforcement learning to uncover extra silent knowledge errors quicker. The corporate is utilizing the machine learning technique to make sure the standard of its Xeon processors.

When an error occurs in a knowledge middle, operators can both take a node down and change it, or use the flawed system for lower-stakes computing, says Manu Shamsa, {an electrical} engineer at Intel’s Chandler, Ariz., campus. However it will be significantly better if errors might be detected earlier on. Ideally they’d be caught earlier than a chip is included in a pc system, when it’s doable to make design or manufacturing corrections to stop errors recurring sooner or later.

“In a laptop computer, you gained’t discover any errors. In knowledge facilities, with actually dense nodes, there are excessive possibilities the celebrities will align and an error will happen.” —Manu Shamsa, Intel

Discovering these flaws is just not really easy. Shamsa says engineers have been so baffled by them they joked that they have to be attributable to spooky motion at a distance, Einstein’s phrase for quantum entanglement. However there’s nothing spooky about them, and Shamsa has spent years characterizing them. In a paper offered on the identical convention final yr, his crew gives a complete catalog of the causes of those errors. Most are attributable to infinitesimal variations in manufacturing.

Even when every of the billions of transistors on every chip is purposeful, they aren’t utterly equivalent to 1 one other. Refined variations in how a given transistor responds to adjustments in temperature, voltage, or frequency, as an example, can result in an error.

These subtleties are more likely to crop up in large knowledge facilities due to the tempo of computing and the huge quantity of silicon concerned. “In a laptop computer, you gained’t discover any errors. In knowledge facilities, with actually dense nodes, there are excessive possibilities the celebrities will align and an error will happen,” Shamsa says.

Some errors may crop up solely after a chip has been put in in a knowledge middle and has been working for months. Small variations within the properties of transistors could cause them to degrade over time. One such silent error Shamsa has discovered is expounded to electrical resistance. A transistor that operates correctly at first, and passes commonplace checks to search for shorts, can, with use, degrade in order that it turns into extra resistant.

“You’re considering all the things is ok, however beneath, an error is inflicting a fallacious resolution,” Shamsa says. Over time, due to a slight weak spot in a single transistor, “one plus one goes to a few, silently, till you see the influence,” Shamsa says.

The brand new method builds on an present set of strategies for detecting silent errors, known as Eigen tests. These checks make the chip do laborious math issues, repeatedly over a time period, within the hopes of creating silent errors obvious. They contain operations on totally different sizes of matrices crammed with random knowledge.

There are a lot of Eigen checks. Operating all of them would take an impractical period of time, so chipmakers use a randomized strategy to generate a manageable set of them. This protects time however leaves errors undetected. “There’s no precept to information the number of inputs,” Shamsa says. He wished to discover a approach to information the choice so {that a} comparatively small variety of checks may flip up extra errors.

The Intel crew used reinforcement learning to develop checks for the a part of its Xeon CPU chip that performs matrix multiplication utilizing what are known as fuse-multiply-add (FMA) directions. Shamsa says they selected the FMA area as a result of it takes up a comparatively massive space of the chip, making it extra weak to potential silent errors—extra silicon, extra issues. What’s extra, flaws on this a part of a chip can generate electromagnetic fields that have an effect on different elements of the system. And since the FMA is turned off to avoid wasting energy when it’s not in use, testing it entails repeatedly powering it up and down, doubtlessly activating hidden defects that in any other case wouldn’t seem in commonplace checks.

Throughout every step of its coaching, the reinforcement-learning program selects totally different checks for the doubtless faulty chip. Every error it detects is handled as a reward, and over time the agent learns to pick which checks maximize the possibilities of detecting errors. After about 500 testing cycles, the algorithm discovered which set of Eigen checks optimized the error-detection fee for the FMA area.

Shamsa says this system is 5 occasions as prone to detect a defect as randomized Eigen testing. Eigen checks are open source, a part of the openDCDiag for knowledge facilities. So different customers ought to have the ability to use reinforcement studying to switch these checks for their very own methods, he says.

To a sure extent, silent, refined flaws are an unavoidable a part of the manufacturing course of—absolute perfection and uniformity stay out of attain. However Shamsa says Intel is attempting to make use of this analysis to be taught to seek out the precursors that result in silent knowledge errors quicker. He’s investigating whether or not there are crimson flags that would present an early warning of future errors, and whether or not it’s doable to vary chip recipes or designs to handle them.