When Intel unveiled Tiger Lake in August, it answered a lot of questions about the CPU’s frequency improvements (large), IPC gains (minimal), and next-generation GPU performance gain (lots). One aspect of the chip it wasn’t willing to talk about, however, was whether the upcoming 11th Generation Intel family would be comprised solely of 10nm chips or if we’d see another “blended” launch with a mixture of 14nm and 10nm parts. With 10th Gen, Intel used a mixture of quad-core and six-core parts to blanket the market. This created confusion over which CPU was the better choice — a confusion that wasn’t helped by the fact that Ice Lake and Whiskey Lake excelled at different workloads.
Now, we know the answer to that question, courtesy of a low-key blog post update posted by Boyd Phelps, Intel’s Corporate VP of Client Computing. For the most part, the post walks through the information on 10nm that we’ve previously discussed at ET, but there are a few interesting additional bits. Phelps writes:
The Willow Cove core increases the mid-level cache to 1.25MB — up from 512KB. We also added a 3MB non-inclusive last-level-cache (LLC) per core slice. A single core workload has access to 12MB of LLC in the 4-core die or up to 24MB in the 8-core die configuration (more detail on 8-core products at a later date).
And with that, we have confirmation. Intel will launch Tiger Lake CPUs with up to eight cores. The stage is set for a knockdown-drag-out battle with AMD across the mobile spectrum. In his post, Phelps notes that Tiger Lake is expected to hit frequency improvements of 1.2x and turbo frequencies of 4.8GHz, with “more to come.”
Intel’s Tiger Lake
It would be more surprising — and alarming — for Intel to fail to offer competition in mobile this year. Mobile has been Intel’s greatest historic strength compared with AMD, and Intel has a great deal of experience optimizing mobile systems for minimum power consumption. AMD has been playing catch-up in this regard; winning the Surface Laptop 3 was a big deal for the company a year ago because it represented AMD’s highest-profile OEM win to-date. The unprecedented delays to Intel’s 10nm allowed other competitors to catch up or even surpass the company, but Intel clearly feels they’ve dealt with the issues effectively enough to retake the driver’s seat. Soon, AMD will launch its Zen 3 architecture, and we’ll finally get to see the two manufacturers slug it out with brand-new mobile architectures at the same time.
We’re on the cusp of an exciting period in the CPU industry. Over the next 24 months, Intel will ramp up 10nm and attempt to get its 7nm into market, AMD will continue to evolve its existing x86 designs, and new ARM products from companies like Ampere, Apple, and Nuvia are all expected in-market. Manufacturing trends like chiplet adoption and 3D chip stacking will both be adopted more widely than they are today. Twenty-four months from now, we may see an entirely different set of companies effectively defining what PC performance and power consumption can look like — or, conversely, x86 may have staved off supposedly robust competition through a reliance on features like software compatibility and consumer familiarity in addition to improved performance.
Intel’s blog post does not state if the company will launch 15W eight-core CPUs the way AMD has, or if it will limit its eight-core chips to higher-end parts with correspondingly higher power consumption. Knowing the TDP ranges of the parts would tell us a lot about how Intel expects them to compete with AMD, which is probably why the company is sitting on that news for now.
Now Read:
- Intel Unveils 10nm ‘SuperFin’ Process Node, Tiger Lake Details
- Intel Announces Tiger Lake Coming Mid-Year, 10nm Ramp Accelerating
- Intel Video Showcases How a 10nm CPU Gets Built
No comments:
Post a Comment