In a shockingly-limited release to reviewers last week (only two major sites received review samples), Intel launched its latest products targeting the consumer desktop PC market, the Broadwell LGA1150-compatible line of CPUs. This was most definitely a “paper” launch, as the products are not (and may never be) available for sale, and in fact, no images of the CPUs or product packaging has yet been made public. But there’s more to this story than meets the eye (not surprising given how little there is to “see”). In fact, Intel’s Broadwell release had a bit of an unstated purpose. Let us explain…To understand the Broadwell desktop launch, you have to look to the past. You see, back in June 2013, Intel launched its Haswell platform, which included the new LGA1150 CPU socket and “4000” line of CPUs, headlined by the Core i7-4770K. Consumers were a bit frustrated that the LGA1155 socket, introduced in January 2011 with the Sandy Bridge line of processors, had only lasted 2.5 years before being relegated to the dustbin of history, but at least it had seen a small upgrade in performance with the release of Ivy Bridge CPUs in April 2012.
Intel couldn’t very well launch a brand-new platform without some promise of future upgrades, so of course it made clear at the time that socket 1150 would have an upgrade path, to what we now know as Broadwell. Sadly for everyone, including Intel, Broadwell was seriously delayed: it was supposed to launch with the H97/Z97 platform refresh in the summer of 2014, but the CPUs weren’t ready. So consumers instead got slightly massaged Haswell-based CPUs, the Core i5-4690K and Core i7-4790K, which brought back Intel’s more effective heat transfer material (a cheaper, less effective material was used on Ivy Bridge and earlier Haswell models). Intel also gave the 4790K a substantial boost in clockrates, making it essentially a “factory-overclocked” CPU. This placated consumers, but it was really just papering over a major problem: socket 1150 would never receive its promised upgrade path.
The Broadwell chips that Intel actually launched on June 2, 2015, the 3.1GHz Core i5-5675C and 3.3Ghz Core i7-5775C(neither of which can actually be purchased a week after release) are not only a year late, but are actually slower than the chips Intel launched last year. Yes, they offer slightly higher instructions per clock cycle (IPC), and yes, they are impressively efficient, but they don’t have nearly the same clockrates as current Haswell models. They could be considered an “upgrade” option for current H97/Z97 motherboard owners, but in reality, they are just laptop processors with a higher power cap (compare specs of the desktop 5775C to the laptop 5850Q launched at the same time and you’ll see what we mean). And that’s not the worst part. As just about the entire world knows, the Intel Skylake platform, which will use the new Socket 1151, is launching In August of 2015, just two months after the release of Broadwell for the desktop.
Clearly then, you don’t want Broadwell, even if you’ve been waiting a long time to upgrade your Haswell CPU (and keep in mind, you can’t even use Broadwell unless you have the newer H97/Z97 platforms that it was supposed to launch with; H81, H87, and Z87 motherboard owners need not apply). But there’s a silver lining to all of this: the state of today’s CPU market is such that you really don’t need to upgrade, at least if you have an Intel quad-core processor of reasonably recent vintage. Intel’s current CPUS are incredibly fast, and moreover, even the massive, upgrade-denying platform jumps that come every two years (e.g., from socket 1155 to socket 1150 to socket 1151) only bring on average a 15% boost in performance. Yes, that means we’re now seeing less than a 10% increase in speed each year, a far cry from what we’d expect given the doubling in transistors every 2 years promised by Moore’s Law.
All this is not to say that Intel’s new releases don’t offer other improvements, whether feature sets, efficiency, or new specialized instruction routines. But the future is clear: if consumers want significant jumps in processing power, it’s going to have to come through more cores, not just enhancements of current dual- or quad-core designs. In the long run, that’s probably fine, but it will require software developers to find innovative ways to code parallel instruction sets to run on all these cores. In the meantime, we can all feel lucky that software demands haven’t been advancing any faster than the processing power of Intel’s CPUs!