Intel: We Aren't Stepping Back to 22nm Haswell; We Never Left

Intel recently posted a product change notice (PCN) to its quality document management system that stated it would reinstate its 22nm Haswell Pentium G3420 processors due to a "new roadmap decision." Intel now tells us that it made the statement in error. Here you can see the original PCN:

(Image credit: Intel)

After reaching out to the company and waiting several days for a response on the matter, it turns out the company never intended to cancel the product. That means the "new" qualifier for its roadmap decision isn't entirely accurate. Per Intel:

"Intel regularly updates customers on product changes via our Product Change Notification (PCN) process. On December 5, a PCN was erroneously posted on product discontinuance of the Intel Pentium Processor G3420. There is no change to the Intel Pentium Processor G3420 product roadmap, and that product continues to be supported by Intel."

Intel says the original notification of the Pentium G3420 discontinuance was made in error, and the Haswell 22nm processor, now a dinosaur by semiconductor standards, lives on. The company issued a new PCN to correct the record because the language in the existing PCN obviously needed to be corrected.

The fact that the G3420 lives on isn't entirely surprising, as processors on trailing-edge manufacturing nodes are often produced long after their debut, typically to satisfy particular needs for OEM systems and IoT applications.

Intel posts documents to its quality document management system to keep its partners apprised of changes to its products. However, this marks the second incident of an incorrect listing in recent months, with a previous entry listing that certain Apollo Lake processors suffered from an LPC bus degradation issue that required a new stepping of the processor. Intel also corrected that PCN to reflect that the chips didn't require a new stepping, after all.

Paul Alcorn
Managing Editor: News and Emerging Tech

Paul Alcorn is the Managing Editor: News and Emerging Tech for Tom's Hardware US. He also writes news and reviews on CPUs, storage, and enterprise hardware.

  • bobalazs
    wait so, am i gonna be able to use socket 1150 again?
    Reply
  • larkspur
    I reported this to you guys last Friday December 6th shortly after your original article was posted. Obviously I don't work for Intel but I linked the evidence and was (apparently?) ignored. In the meantime at least 2 other tech sites picked up the story based on Tom's erroneous report... Thanks for issuing this correction Paul.

    Edit: And the original article claimed that Intel's 22nm is obsolete - which, obviously it is not. Legacy CPUs, chipsets, lower-performance, purpose-built ICs are produced on older nodes. That doesn't make them obsolete - just not cutting-edge. Considering Intel's continued shortages on their overused 14nm node, it's not at all surprising that their 22nm GP node is still getting heavy use...
    Reply
  • bit_user
    larkspur said:
    Edit: And the original article claimed that Intel's 22nm is obsolete - which, obviously it is not. Legacy CPUs, chipsets, lower-performance, purpose-built ICs are produced on older nodes.
    It is obsolete, for building CPUs. I think that was sufficiently obvious, though it wouldn't have hurt for @PaulAlcorn to qualify it as such.

    And, again, thanks for your post in the other thread.
    Reply
  • bit_user
    BTW, @PaulAlcorn I noticed the Pentium G3420 supports ECC memory. That suggests the reason for keeping it alive might've had to do with some specialized machinery using them - not necessarily consumer-oriented desktop PCs. Or at least it might've been a determining factor in deciding which low-end desktop CPU model to keep around.

    https://ark.intel.com/content/www/us/en/ark/products/77775/intel-pentium-processor-g3420-3m-cache-3-20-ghz.html
    Reply
  • clutchc
    Intel is having issues with shortages.
    https://techreport.com/news/3467271/intel-revives-haswell-era-chip-amid-14nm-shortage/
    Reply
  • NightHawkRMX
    AMD has had stock issues with the 3rd gen threadripper (may smooth out soon) and Ryzen 9s being hard to find. However, one look on PCpart picker shows intel is having issues too.

    Recently Intel has had supply issues pretty much year round.

    The I3 9350kf, is still often out of stock or sold at inflated prices like amazon at $225. Even though this CPU has GPU removed to increase yields, it is still hard to come by. In addition, the locked i3 9300 and i3 9320 are experiencing stock issues.
    https://pcpartpicker.com/product/DDtQzy/intel-core-i3-9350kf-4-ghz-quad-core-processor-bx80684i39350kfI5 9500 and 9600 are hard to find and carry inflated prices sometimes reaching $350 due to low stock.
    https://pcpartpicker.com/product/bKH8TW/intel-core-i5-9600-31-ghz-6-core-processor-bx80684i59600Intel Pentium Gold CPUs are hard to find and carry vastly inflated prices compared to even the superior core I3 CPUs. For example, the Pentium G5400, G5500, and G5600 all costs close to or above $100
    https://pcpartpicker.com/product/9PgzK8/intel-pentium-gold-g5600-39ghz-dual-core-oemtray-processor-cm8068403377513
    So Intel seems to be in worse shape than AMD in some ways, but i suppose this is since much of Intel's manufacturing capacity is going towards OEM systems and mobile devices, markets where AMD moves far less product. AMD is focusing on enthusiast sales. I also noticed many of the above CPUs experiencing supply issues are locked CPUs that would be more tailored to a prebuilt.
    Reply
  • TerryLaze
    NightHawkRMX said:
    So Intel seems to be in worse shape than AMD in some ways,
    Both companies just had their best economic quarters in years,neither one is able to produce enough chips to satisfy demand,bad or worse shape is a very wrong way of putting this.
    Reply
  • spdragoo
    larkspur said:
    I reported this to you guys last Friday December 6th shortly after your original article was posted. Obviously I don't work for Intel but I linked the evidence and was (apparently?) ignored. In the meantime at least 2 other tech sites picked up the story based on Tom's erroneous report... Thanks for issuing this correction Paul.
    (snip)

    I remember checking on Intel's ARK site last week when the original post came out... & (at least that day) the G3420 was listed as "Discontinued". Today, it's back to "Launched"...so apparently it wasn't just the PCN that was erroneously changed by Intel, but also the ARK entry.

    So I wouldn't classify it as TH "ignoring" you. Rather, I'd consider it, "despite 3rd-party information supposedly contradicting Intel's official announcement, we'll wait until Intel officially clarifies the issue"...which they now have.
    Reply
  • larkspur
    bit_user said:
    It is obsolete, for building CPUs. I think that was sufficiently obvious, though it wouldn't have hurt for @PaulAlcorn to qualify it as such.
    I think we have two different perceptions of what a CPU constitutes. I assume you mean cutting edge-high performance PC CPUs. On that, sure I can agree with you. I am referring to the term CPU which refers to any central processing unit. I already told you the definition of obsolete according to Oxford in a previous post... We can disagree on our nomenclature and therefore be in agreement but I will continue to disagree that Intel's 22nm node is obsolete - it is very much in use lol! If you are only thinking about retail PC hardware then sure, ok I'm not arguing that these have reached an out-of-date status. But 22nm CPUs are very much in production at Intel - they aren't the center of a PC like you think - they are the center of a DEVICE - think of a device you use and it might be powered by one... :)
    Reply
  • larkspur
    spdragoo said:
    I remember checking on Intel's ARK site last week when the original post came out... & (at least that day) the G3420 was listed as "Discontinued". Today, it's back to "Launched"...so apparently it wasn't just the PCN that was erroneously changed by Intel, but also the ARK entry.

    So I wouldn't classify it as TH "ignoring" you. Rather, I'd consider it, "despite 3rd-party information supposedly contradicting Intel's official announcement, we'll wait until Intel officially clarifies the issue"...which they now have.
    Nah, I reported it to the mods. It had nothing to do with ARK and everything to do with Intel QDMS (Quality Document Managment System) - if you know how to read it you could have easily seen it was an "about face" by Intel - either a mistake or a quick reaction to a big client (which is what I still think it was). The original author missed that... And then the original smear article stayed up for days while other tech publications picked it up straight from Tom's... heh heh - That's how misinformation spreads... Anyway don't worry about it - happens in journalism all the time. Like I said - I'm glad Paul took care of it and now Intel managed to play it off like it was just a "mistake"! LOL!
    Reply