Microprocessor giant Intel (NASDAQ:INTC) recently reported its financial results for the second quarter of the year. The company's business performed well, helped along by better-than-expected results in its personal computer processor business.
Intel's personal computer processor business sells chips for both desktop computers as well as notebook computers. It was sales into the latter that helped boost Intel's overall personal computer processor business, with platform unit shipments up 14% and average selling prices up 6% year over year.
Desktop revenue was down 3% year over year, due to a combination of year-over-year unit shipment declines compounded by a slight average selling price decline.
In a recent column, I argued that Intel could improve the performance of its desktop chip business by building more compelling, targeted products for the segment.
In this column, I'd like to go over three specific things I'd like to see from the company's Ice Lake architecture-based desktop personal computer chips, which should arrive in late 2018 or early 2019.
Intel's mainstream desktop personal computer processors topped out at four physical processor cores since the launch of the Core 2 Quad processor family back in 2007. Over time, those cores have gotten faster, and Intel has integrated more non-processor components into the chips (e.g., graphics/media), but the per-core performance gains that Intel has been delivering generation over generation for the last half-decade have been rather modest.
Intel is expected to launch a six-core mainstream desktop personal computer processor as part of its Aug. 21 eighth-generation Core processor launch, representing the company's first core count increase in mainstream desktop processors since 2007.
With the Ice Lake family of processors -- which will likely be marketed as the company's ninth-generation Core processors -- I would like to see an increase in processor core counts to at least eight.
Remember: Ice Lake will be manufactured on Intel's 10-nanometer chip manufacturing technology, which Intel says delivers a more than doubling of logic density over its 14-nanometer technology. Given this dramatic increase in technology density, packing in additional processor cores could be a great way to translate improved manufacturing technology into value for consumers.
Although "more cores" is generally a compelling selling point, those cores need to deliver performance improvements over what came before them, particularly as most consumer-grade software is still highly sensitive to per-core performance.
With Ice Lake, Intel is expected to deliver an enhanced processor core relative to the core found in the sixth-, seventh-, and upcoming eighth-generation Core families of chips.
Intel has historically delivered a roughly 5% improvement in performance-per-clock in the generation immediately following a new core architecture (known as a "tick" in Intel's now-defunct "tick-tock" development model), and then a 10% improvement in performance-per-clock in moving to a new core architecture (known as a "tock").
The Ice Lake core represents the next "tock" in Intel's processor core development process, so I would expect to see at least a 15% improvement in performance-per-clock over the processor core used in Intel's currently shipping products.
More graphics, too
One area that Intel has worked to improve on has been the integrated graphics processors inside of its chips. Today, Intel's graphics processors are adequate for non-gaming tasks (e.g., web surfing, video playback, etc.), but they're still not suitable for playing some of the most popular games out there today.
Intel's goal shouldn't, realistically, be to displace graphics processors for the high-end gaming market; that's simply not going to happen with a solution integrated into the processor for cost, power, and form factor reasons.
What Intel's goal should be, though, is to build graphics processors into each one of its chips that can handle the major eSports titles, like Overwatch, at fluid frame rates and reasonable quality settings.
Intel is expected to deliver a new graphics architecture with its Ice Lake family of processors that, like its processor cores, should be two architectural generations over what's shipping in Intel's chips today. However, a new architecture alone isn't enough -- Intel will need to cram more of those updated graphics cores into its mainstream desktop chips than it does today to deliver higher performance.
If Intel can offer credible integrated graphics capabilities for mainstream gamers, that could help bolster the value proposition of its latest chips and help its desktop personal computer chip sales and average selling prices.