With near zero competition right now on the desktop front, why rush? Our sources say that, instead of Broadwell, the 2014 will see just a Haswell refresh first, likely a new CPU stepping combined with a new chipset.
Even with only the incremental CPU performance gain over the Ivy Bridge, Haswell still firmly rules the mainstream desktop CPU world. The substantially higher GPU performance and the improved power bill do help as well. So, since the tick tock model from 2008 is anyway a thing of the past now, why not extend the new CPU’s shelf life too, and maximise the return on the investment? After all, it’s not the case like some eight years ago when the competition was actually having a better performing product, and it won’t be the case for a while – read end 2014 at least.
So, our Computex sources advised us to give more focus on the Haswell, as we’ll be stuck with it pretty much until early 2015 at least; that’s the current Broadwell launch timeframe, too bad.
The refresh will offer two things: one is the mid-term stepping with likely a bit of clock speed jump in both mainstream and unlocked version, something we always had with Intel desktop offerings except in the Ivy Bridge generation. Since the Xeon E3 workstation versions of the new Haswell right now will start at 3.6 GHz base top speed rather than 3.5 GHz on the i7-4770K, you could guess that by next year, the top bin for the unlocked desktops will likely sport the same speed, although it may be immaterial to the overclockers pushing the parts towards 5 GHz and beyond. The ‘locked mainstream’ bins would benefit more for the extra one or two notches on the multiplier, of course.
The other upgrade will be the Z97 chipset, originally planned to coincide with the Broadwell launch, but now allowed to carry the Haswell refresh platform side. Basically, the 1000 MB/s SATA Express interface /just one of these slots this time/ for the mass storage is the key news here, plus the Boot Guard and Intel Device protection against low level malware attacks bypassing the OS.
Now, would it be worth for the Haswell users to swap the board next year to gain these two incremental features? Of course not, however these can be an additional carrot for those users who haven’t migrated for quite a while, in many cases due to low perceived processor performance gains, to make the decision to jump.
When to expect this upgrade? The sources pointed out early second quarter of next year, likely the CeBIT time in March (anyone still remember this show?), for the possible launch, subject to the usual and unpredictable changes. What’s interesting is that, some 6 months after that, the Haswell-E high end Socket 2011 v2 desktop CPU should be available as well, paired up with the X99 high end chipset, the desktop version of Grantley workstation/server platform for the 14-core Haswell-EP Xeon E5 v3 CPU series.
This Haswell life extension and extra delay added to the desktop move to Broadwell has another effect: after a long time, will reduce the delay that high end desktop and server CPUs were suffering compared to the mainstream desktop and mobile ones, as all of these lines could then likely have Broadwell refresh in various quarters of 2015. Also, at the same time, gives Intel more time to perfect its 14 nm process, which Broadwell and the Atom-class mobile CPUs would be the first beneficiaries of. Also, if the Z97 chipset with the related new mainboard lines is there for the Haswell refresh, it makes it more likely that Intel will have socketable Broadwell processors available for those boards too, isn’t it?