Earlier this year, Intel (NASDAQ:INTC) launched a unique pair of products aimed at the desktop known as the Core i5 5675c and the Core i7 5775c. These processors came with four Broadwell CPU cores (minor refresh of the prior generation Haswell) as well as the company's highest-end Iris Pro graphics (desktop chips have historically featured vanilla Intel HD graphics).
While the improved graphics performance of the chip was probably interesting to a small subset of the customers for the chip, the more interesting thing about the processor is that to support that graphics performance was a very large on-package cache memory.
This cache memory, in addition to accelerating graphics workloads, also helped improve general purpose CPU tasks, particularly gaming. Indeed, according to The Tech Report, it was able to best the company's newer Skylake based chips in gaming applications.
Unfortunately, this product doesn't seem to have done all that well in the marketplace, and it would appear that many online retailers are no longer even carrying it.
According to a recent leak, Intel plans to release a successor to this part in late 2016. The question, then, is whether there is really a market for such a chip.
Why did those Broadwell-based chips fail in the marketplace?
The prior generation Broadwell parts were announced in June of 2015. Although the chips could be found in a handful of pre-built systems, it was virtually impossible to buy just the chips themselves at the various online retailers.
These chips had just three months in the marketplace before Intel announced processors based on its next generation Skylake architecture. Not only were these chips based on a newer architecture, but the Skylake motherboards included many new features and capabilities that could not be found on the older boards that the Broadwell chips required.
This meant that in order to build a new system around these Broadwell chips, buyers would have to seek out last-generation motherboards and use last-generation DDR3 memory (Skylake brought support for newer, better DDR4 memory) -- not at all ideal.
Will things be different this time?
According to a leak from the generally reliable BenchLife.info, the new Skylake chip with the large cache memory will launch in the fourth quarter of 2016 alongside the company's next generation mainstream desktop chip (without the large cache) known as Kaby Lake.
The big advantage that I see for the Skylake chip this time around is that at the time of launch, both the Skylake chip with eDRAM as well as the new Kaby Lake chip will work in the same motherboards. This means that buyers won't have to choose between the old platform and the new platform when it comes to making a buying decision.
Another thing to note is that this Skylake chip will be quite large by virtue of the extremely large graphics processor embedded onto the silicon die. It is well known that Intel is still facing some manufacturing yield challenges with the 14-nanometer manufacturing technology that this chip will be built on (though according to Intel, 14-nanometer yields are "maturing").
By the end of 2016, I would be very surprised if Intel's 14-nanometer yields weren't in excellent shape, allowing the company to reliably produce enough units to satisfy demand (something that Intel seems to be struggling with currently) at a good cost structure.
It remains to be seen how popular these new Skylake chips will be among PC enthusiasts, but at least it -- unlike its Broadwell-based predecessor -- stands a chance in the marketplace.