Intel Un-Discontinues Pentium G3420 ‘Haswell’ CPUby Anton Shilov on December 9, 2019 4:00 PM EST
As spotted by ComputerBase.de, in a rare event, Intel has canceled its plan to EOL one of its low-end Haswell-generation processors, resuming processor shipments for the foreseeable future. Curiously, the change in plans comes less than two weeks after Intel first began the process of discontinuing the processor. And while Intel does not publish their the detailed rationale behind their decisions in their product change notifications, given the company's ongoing low-end CPU shortage, it's fair to say that Intel needs all of the low-end CPUs it can get at the moment.
The processor in question is Intel’s Pentium G3420, which offers two Haswell architecture CPU cores (no HyperThreading) running at 3.20 GHz, as well as the company’s HD integrated graphics. The 22nm chip is compatible with the widely available LGA1151 infrastructure that supports 53 W CPUs. As a part of Intel’s Haswell family, the chip was originally meant to be used inside low-end desktops, and it has also found its way into devices such as NAS boxes.
Keeping in mind that most embedded versions of Haswell CPUs have been EOLed, this one could be an answer to demand from that market. Alternatively, ComputerBase believes that the change in plans is a stop-gap for Intel, so that they have some kind of low-end Core-based Pentium processor to offer OEMs who are currently being starved of suitable Skylake chips.
The statement from Intel reads as follows:
This revision supersedes the prior EOL notice and is intended to inform customers that they do not need to do anything more on their end for last orders and should plan on this product being available as usual. Please disregard the notice of the product End of Life as shared in prior communications and note that this product will continue to be available for orders as usual. Intel is not pursuing EOL of this product at this time.
- Thecus Announces Two New Rackmount NAS Servers
- Devil’s Canyon Review: Intel Core i7-4790K and i5-4690K
Sources: Intel, ComputerBase
Post Your CommentPlease log in or sign up to comment.
View All Comments
kpb321 - Monday, December 9, 2019 - linkYeah being a 22nm process for the chip this looks like a play to avoid having to use capacity constrained 14nm production for producing low end/inexpensive chips. Being stuck on 14nm for so long and having to start producing much larger chips due to AMD being competitive again and pushing core counts up across the board has really put a strain on their 14nm capacity so it make sense to not use it for low cost low margin chips.
nandnandnand - Monday, December 9, 2019 - linkI can't really fault Intel for this when AMD is still making 28nm Excavator chips. But it is another footnote in Intel's 10nm failure.
Jorgp2 - Monday, December 9, 2019 - linkHow is this even related to 10nm?
mode_13h - Monday, December 9, 2019 - linkThey had planned for their mainstream production all to be on 10 nm, by now. Had that been the case, they'd have had spare 14 nm capacity to make chips to sell into whatever market is buying those Haswell Pentiums.
So, basically, 10 nm delays and problems backlogged everything else in their product linup.
Korguz - Monday, December 9, 2019 - linkif intel didn't have to make anything other then quad core cpu's still for the main stream, and kept 6 or more cores for the HEDT AMD server markets, their 14nm production wouldn't be so constrained, but, in order to compete with AMD on core count, they HAD to put more cores into their cpus or they would be even more behind AMD...
Flunk - Monday, December 9, 2019 - linkThe compute cores are only half the CPU on a 8-core Coffee Lake part. If Intel offered a version with no giant space-wasting GPU they'd be just fine. Intel is behind in core count purely because they don't produce a real high-end consumer desktop chip and AMD does.
Korguz - Tuesday, December 10, 2019 - linkregardless... they are still using wafer space they weren't planning on using to make higher core count mainstream parts, when they didnt think they needed to. some one here did a nice reply to some one else on this, can't remember which article it was for, but it kinda of explained intels constraints quite nicely.
edzieba - Tuesday, December 10, 2019 - link" Had that been the case, they'd have had spare 14 nm capacity"
Had Intel had 10nm ready to roll out to production scale, then they would have done exactly the same as they had for every past process node: migrate all but one line to the new process and retain one low-volume line for extended-support products. With current demand levels, Intel would have been on 10nm and supply constrained rather than on 14nm and supply constrained.
lightningz71 - Tuesday, December 10, 2019 - linkEven if they did as you say, you aren't counting two things:
The first is that 10nm brings a significant density improvement for Intel, meaning that they will be getting FAR more chips per wafer if they do nothing else to increase the effective transistor count of their processors. They would have almost no demand for their leading performance 14nm node in their volume production products, giving them plenty of capacity for the remaining demand.
Second, given the improved competitiveness of AMD, Intel would have likely retained more of their 14nm capacity for low to mid tier products without sacrificing their ability to meet demand on the high end. Their 14nm process is so well refined that they would have easily been able to produce dirt cheap 4 die with iGPU chips to handily beat AMD's mobile products in any market that they chose to compete in. Intel's 14nm, in its many variants, still outclasses global Foundaries 14/12nm process, and even GloFo's latest announcement on an improved 12nm product for next year still won't fully compete on anything but cost. AMD's cheap 28nm A series APUs are really just providing a price ceiling for the absolute bottom of the market. Intel may not have even wanted any part of that market to begin with.
Korguz - Tuesday, December 10, 2019 - linklightningz71 " The first is that 10nm brings a significant density improvement for Intel, meaning that they will be getting FAR more chips per wafer" doesn't mean squat when they cant make any chips that has more then 4 cores( apparently ) , does it ?
you really have NO idea why intels 14nm process is so constrained do you?? its not the mobile space intel is screwed on. its desktop and server.