Intel Corp. today gave the world the first in-depth look at its upcoming 10-nanometer Tiger Lake laptop processors, which the chipmaker says will provide the biggest performance jump inside an existing chip architecture in its history.

It also shared details about its upcoming graphics cards for data centers.

The Tiger Lake processor previews comes at an important time for Intel, which is working to modernize its chip production facilitates amid increased competition from rivals such as Advanced Micro Devices Inc. The company is retooling its plants from making 14-nanometer chips to more sophisticated 10-nanometer products like Tiger Lake. Intel’s production of 10-nanometer silicon has suffered multiple delays, but those supply issues are now resolved, company officials disclosed during the processor reveal.

Tiger Lake is a refinement of Intel’s previous mass-produced 10-nanometer chips for laptops. Those chips offered higher 10% to 20% more instructions per clock cycle than its 14-nanometer chips. However, they had lower core frequencies, which offset much of the performance gains from the extra instructions per clock cycle.

In the new Tiger Lake line, Intel is increasing frequencies back up to unlock those performance gains. It also switched the processors to a new transistor design, or stack, that it says is 17% to 18% faster. 

It’s this “intranode” performance jump that Intel touts as the biggest in its history. The company says the speed increase is comparable to improvements it previously only achieved in much larger “full node” overhauls of its chip design. 

Tiger Lake “appears to deliver on what Intel has been saying it wants to do by leveraging its 6 Pillar Strategy,” said Patrick Moorhead of Moor Insights & Strategy. “While it’s unclear how Tiger Lake competes with AMD at this point as this wasn’t the product launch, the company did seem to scale Tiger Lake pretty well and add new technologies where it needed it.”

One of those new technologies is Tiger Lake’s integrated graphics processing unit, which is based on the Xe-LP architecture that was also detailed in-depth for the first time this week. Intel is promising 50% more raw graphics processing than in its earlier laptop processors.

The Xe-LP integrated chip is part of the Xe product line, a new GPU line that includes parts not only for consumer devices but also data centers. The data center variant is called the Xe-HP. It’s based what Intel describes an industry first “multitile” design, which essentially combines up to four individual chip dies into a single GPU, and has cores optimized to run artificial intelligence workloads. 

“Xe-HP is the industry’s first multitiled, highly scalable, high-performance GPU architecture, providing petaflop-scale AI performance and rack-level media performance in a single package,” said Intel Senior Vice President Raja Koduri (pictured).

The Xe-HP graphics cards use the same new transistor designs as Tiger Lake. The design, dubbed SuperFin by Intel, features lower electrical resistance that reduces heat buildup in circuits and allows them to run at higher speeds as a result. 

“While likely behind in density, if you look at Intel’s new entire transistor ‘stack,’ it’s likely the highest performance in the industry,” Moorhead said. But, he added, “obviously, execution is key. This wasn’t a product launch, rather a technology disclosure, without any discussion on ability to execute.”

Intel recently said that it expects a six-month delay in the execution of its planned shift to a seven-nanometer chip fabrication process, the next technological leap after the current 10-nanometer processor generation. That announcement led to one of the company’s biggest one-day stock price dips in recent years. Moorhead sees Tiger Lake as an encouraging sign for the company. 

“Overall, I feel better about its technological future as it didn’t ignore the obvious fab issues and at the same time gave reasons to believe that architecturally, it could be back on top in the future,” he said. 

Photo: Intel