In August, microprocessor giant Intel (NASDAQ:INTC) is expected to launch a new processor family called Coffee Lake-S for the enthusiast desktop computer market. The initial enthusiast-focused Coffee Lake-S parts will, per a recent leak, come in four- and six-core variants, up from the two- and four-core configurations of the current Kaby Lake-S processors.
The Coffee Lake-S parts are also expected to be manufactured using the company's new 14nm++ manufacturing technology, which promises a roughly 10% performance improvement over its currently shipping 14nm+ technology.
In this column, I'd like to explain what I hope to see from this product.
Pushing ahead on per-core performance
In going from the top Skylake-S part (Core i7-6700K) to the top Kaby Lake-S part (Core i7-7700K), Intel kept the basic processor architecture the same but used the improved manufacturing technology to increase the operating frequencies of the chip.
Intel's efforts bore fruit, as the table below illustrates:
|Frequency (GHz)||Core i7-6700K||Core i7-7700K|
Now, it was relatively easy for Intel to convert the manufacturing technology gains into straightforward frequency increases because the 6700K and 7700K are based on the same architecture and have the same number of cores.
With the top Coffee Lake-S part -- likely the direct successor to the Core i7-7700K in terms of product positioning -- Intel will will be increasing the processor core count by 50%. At the same time, Intel isn't expected to increase the rated thermal design power of that hex-core chip (so power draw shouldn't, in theory, go up much) and a 10% improvement in manufacturing technology performance isn't going to completely offset a large core count increase.
So, I expect that when all six cores on the upcoming Coffee Lake-S parts are loaded, they will run at less than the 4.4 GHz that all the cores on the 7700K do when they're being used.
However, I think that the gains from 14nm++ should allow Intel to increase the frequencies at which the Coffee Lake-S processor cores run at when between one and four cores is utilized. This allows Intel to have its cake and eat it, too, so to speak.
What can Intel realistically achieve?
If Intel wants to build some really impressive products for the enthusiast market, then a straight 10% increase in one- to four-core frequencies over the 7700K chip would probably do it. However, given the very high frequencies that the 7700K already operates at, 10% might be tough to achieve.
If Intel can deliver at least a 5% frequency boost over the 7700K when one- to four-cores are being used, I'd consider that a worthy achievement.
Then, if when all six cores are loaded up, Intel can achieve frequencies of around 4GHz (a reasonable reduction from the 4.4GHz all-core turbo on the 7700K), the chip would also be an attractive choice for users interested in high performance in scenarios that can use a lot of cores.