Top Ad unit 728 × 90

Intel’s ‘New’ 8th Generation Processors are Built on Kaby Lake, Add Additional Cores

Joel Hruska
Ever since Intel introduced the first-generation Core i7, it’s followed a predictable series numbering. First-generation Core processors were codenamed Nehalem, second-generation CPUs were Sandy Bridge, followed by Ivy Bridge (3rd), Haswell, (4th) and so on. In each case, new chips — whether they were die shrinks or new architectures — were assigned a new product number. Today, with its 8th-generation chips, Intel is explicitly changing that policy. Unlike previous product generations, the 8th-generation launching today family will span multiple chip families built on 14nm+ (Kaby Lake), 14nm++ (Coffee Lake), and 10nm (Cannon Lake).
The four U-series chips Intel is launching today are fundamentally based on Kaby Lake with the same architecture, the same GPU, and almost the same capabilities. The one minor update to the GPU side of the equation is support for HDMI 2.0 and HDCP 2.2, without any need for third-party solutions. So what, exactly, is new about these chips? Two things: Larger core counts, and slightly higher clock speeds at maximum Turbo.
8th-Gen
Click to enlarge
The table above shows how the new Core i7 chips compare against their 7th-generation predecessors at the 15W TDP. Intel is trading a significant amount of base clock speed for core counts, but the maximum turbo speed on these chips is still higher, in some cases, than the parts they replace. The new i5 CPUs aren’t shown here, but they compare identically to the Core i7 mobile parts, with higher turbo clocks, quad core configurations with Hyper-Threading, which is historically unusual in the i5 quad-core lineup. Intel doesn’t put much emphasis on its quad-core i5s, but the 7th generation quad-cores didn’t have Hyper-Threading at all. The new chips do.
Intel’s mobile revamp mirrors product and price changes the company has already introduced. For the past six years, Intel has followed the same basic processor philosophy: In mobile, dual-core processors were the norm at every level, with only a few quad-core / eight-thread chips available. These have always occupied the top of the product stack and typically been offered only in higher TDP brackets. These new chips change all that.

Performance Scaling Still Unknown

There’s a bit of mixed messaging over how much performance these new cores will offer. According to Intel, it expects a 40% overall improvement, with 25% coming from the addition of two more cores. ‘Design’ and ‘Manufacturing’ also add to the total, albeit in smaller amounts. But that’s actually less performance than we’d expect to see, given that doubling core counts from a dual CPU to a quad CPU can drive more than 25% improvement on its own in desktops. Most modern applications scale fairly well, up to four cores / eight threads, and while that’s not an absolute, the 25% figure is still lower than expected.
CoreImprovements
There are several potential answers to this. One is that Intel is being conservative and publishing numbers it knows it can back up in virtually all circumstances. The other is that Intel has to pull its Turbo clocks down to stay within a 15W TDP. This would be directly implicated if the boost clocks on the 8th-generation CPUs drop significantly as core loads increase. As the table below shows, however, they don’t:
Clock-Chart
As you can see, this chart suggests that Intel has set fairly aggressive Turbo Mode clocks for these cores. Just how theoretical these core clocks are remains to be seen. Intel began offering OEMs more flexibility to hit their TDP and performance targets several years ago, but doing so created odd performance dips and spikes. In several cases, the lowest-end Core M you could buy actually yielded better performance than the higher-end chips due to thermal issues. Whether or not that will occur here is something we can’t judge until products have shipped.

What’s Next for 8th-Gen Core?

Intel has announced that it will launch its next generation desktop processors “in fall,” but those of you hoping to drop a six-core i5 or 12-thread i7 into an existing system are out of luck. The Coffee Lake refresh will require 300-series chipsets, and will not be backwards compatible with existing products. Given how little time it’s been since Intel introduced Kaby Lake, the quick hop from the 200-series to the 300-series won’t sit well with people who just upgraded to the 7700K, especially if the desktop 8th-generation cores make six cores available for the same price Intel used to charge for four.
All the new chips
Intel may not have great options here. Semiconductor designs and products tend to have longer lead times than the public is aware of; it’s not unusual to see someone on LinkedIn who worked on a given product a year or more before it ships. Intel’s 8th-generation Coffee Lake chips are undoubtedly aimed at plugging the hole chips like the 1600X blew in its product family. Kaby Lake would’ve been completed before much information on Ryzen was available, and Intel’s various product shake-ups this year have been a response to it. AMD’s Ryzen desktop family continues to sell well, with 7 SKUs in the Top 15 CPUs according to Amazon.
Intel’s ‘New’ 8th Generation Processors are Built on Kaby Lake, Add Additional Cores Reviewed by Chidinma C Amadi on 3:48 PM Rating: 5

No comments:

Kogonuso © All Rights Reserved!

Contact Form

Name

Email *

Message *

Powered by Blogger.