Last May, Sandra Rivera, a top executive at Intel chip giantThere is some alarming news.
Engineers have worked for more than five years to develop a powerful new microprocessor for data center computing, and are confident they’ve finally got the right product. But signs of a potentially serious technical flaw emerged during the regular morning meeting to discuss the project.
The problem was so troubling that Sapphire Rapids, the processor’s codename, was delayed — the latest in a series of failures for one of Intel’s most important products in years.
“We’re pretty depressed,” said Rivera, executive vice president of Intel’s data center and artificial intelligence group. “It was a painful decision.”
The Sapphire Rapids launch was finally pushed from mid-2022 to Tuesday, nearly two years later than expected. The product’s long history — combining four chips in one package — highlights some of the challenges facing Intel’s turnaround efforts as The United States is trying to assert its dominance in fundamental computer technology.
Since the 1970s, Intel has been a leader in the production of tiny silicon wafers that run most electronic devices, most famously for a variety of so-called microprocessors, which act as the electronic brain in the computer. most computers. But the Silicon Valley company has in recent years lost its longtime leadership in manufacturing technology, which helps determine the computational speed of chips.
Patrick Gelsinger, who became Intel’s chief executive in 2021, has vowed to restore its manufacturing edge and build new factories in the United States. He was a leading figure when Congress debated and passed the law in the summer to reduce US dependence on chip production in Taiwan, which China claims as its territory.
Sapphire Rapids’ bumpy growth has implications for whether Intel can recover to deliver future chips on time. It’s an issue that can affect a lot of computer manufacturers and cloud service providers, not to mention the millions of consumers who use potentially technology-powered online services. Intel.
“What we want is a steady pace,” said Kirk Skaugen, executive vice president of server sales at Lenovo, a Chinese company that is planning 25 new systems based on the new processors. can be predicted. “Sapphire Rapids is the starting point of a journey.”
For Intel, the pressure is mounting. Along with falling demand for chips used in personal computers, the company faces stiff competition in server chips, its most profitable business. That issue has worried Wall Street, with Intel’s market value falling by more than $120 billion since Gelsinger took office.
At an online event on Tuesday to discuss Sapphire Rapids, named for a part of the Colorado River, Intel customers described plans to use processors that they believe will bring benefits. particularly beneficial for artificial intelligence tasks. The product, officially known as the 4th Gen Intel Xeon Scalable processors, was introduced alongside another delayed addition to the Xeon chip family. That product, formerly codenamed Ponte Vecchio, is designed to accelerate special-purpose tasks and is used in conjunction with Sapphire Rapids in high-performance computers.
In an interview, Mr. Gelsinger said Sapphire Rapids made a hit, despite the delay. He chose Rivera in 2021 to take over its development unit, where she is using lessons from the experience to change the way Intel designs and tests its products. He said Intel has conducted several internal reviews of what happened to Sapphire Rapids and “we’re not done yet.”
Sapphire Rapids began in 2015, with discussions among a small group of Intel engineers. The product is the company’s first attempt at a new approach to chip design. Companies now often pack tens of billions of tiny transistors on each piece of silicon, but competitors like Advanced micro-devices and others have begun manufacturing microprocessors from multiple chips packed together in plastic packages.
Intel engineers came up with a four-chip design, each with 15 processor “cores” that act like individual computers for general-purpose computing. The company also decided to add circuit blocks for special tasks — including artificial intelligence and encryption — and to communicate with other components, such as data storage chips.
Shlomit Weiss, who co-leads Intel’s design engineering team, says the interaction between so many factors is “very complex”. “Complexity often brings problems.”
The Sapphire Rapids team has been grappling with bugs, design flaws, or manufacturing glitches that can cause chips to miscalculate, slow down, or stop working. They are also affected by delays in product production.
But by December 2019, engineers had reached an important milestone called “starting from the ice”. That’s when the electronic files containing the finished design are shipped to the factory for sample chip creation.
Sample chips appeared in early 2020, when Covid-19 forced a lockdown. The engineers soon got the computing cores on Sapphire Rapids to communicate with each other, said Nevine Nassif, the project’s lead engineer. But there is still more work than expected.
One important job is “validation,” a testing process in which Intel and its customers run software on sample chips to simulate computing jobs and catch errors. Once the bug is found and fixed, the designs can go back to the factory to create a new test chip, a process that usually takes more than a month.
Repeating that process leads to late deadlines. Ms. Nassif said Sapphire Rapids was designed to counter AMD’s Milan processors, which were introduced in March 2021. But it wasn’t ready yet in June of that year, when Intel announced a delay to next year to allow further confirmation.
That’s when Mrs. Rivera walked in. The longtime Intel CEO built a successful networking products business before being appointed chief human resources officer in 2019.
“We have to get our execution mojo back,” Mr. Gelsinger said. “I need someone who will jump into the fire and take care of this business for me.”
In October 2021, Ms. Rivera and a top design executive set up weekly Sapphire Rapids health meetings, held at 7am every Monday. Those meetings showed steady progress in finding and fixing bugs, she said, bolstering confidence about starting production in the second quarter of 2022.
Then there was the discovery of the vulnerability in May last year. Ms. Rivera did not describe the details but said it affected processor performance. In June, she used an investor event to announce a delay of at least a quarter, which pushed the Sapphire Rapids launch later than the competing AMD chip launch in June. 11.
“We are ready to deliver,” Ms. Nassif said. The final delay was “sad with all the effort put in.”
Ms. Rivera saw a series of lessons from the failures. One reason is simply that Intel packed too many innovations into Sapphire Rapids, rather than delivering a less ambitious product sooner.
She also concluded that the team should spend more time perfecting and testing their design using computer simulations. Ms. Rivera said it is less expensive to find bugs before they are in the sample chip and can remove features to simplify the product. Since then, she’s moved on to enhancing Intel’s emulation and validation capabilities.
“We used to have a lot of these muscles and we let them atrophy,” Ms. Rivera said. “Now we are rebuilding.”
She also determined that Intel was scheduling more products than its engineers and customers could easily handle. So she’s streamlined that product roadmap, including pushing back the Sapphire Rapids successor to 2024 from 2023.
More broadly, Rivera and other Intel executives pushed the organization to develop better processes for documenting technical issues and sharing that information within and outside the company.
Some Intel customers say that communication has gotten better.
“Is everything going well? No,” said Lenovo’s Skaugen, who used to run Intel’s server chip business. “But we were a lot less surprised than before.”
