Skip to main content

Intel Apollo Lake CPUs May Die Sooner Than Expected (Updated)

(Image credit: By muhammadtoqeer / Shutterstock)

Update 9/11/2019, 8:50am PT: Intel pulled the PCN document from its site, and after further investigation, the company tells us the PCN was posted in error. We've provided more detail here.

Original Article 9/9/2019 12:22am PT:

As detailed in a recent Product Change Notification (PCN) document, Intel is refreshing four of the company's Celeron and Pentium Apollo Lake processors due to degradation concerns. The problem stems from the same issue that initially cropped up with the C2000 Atom family, which had enough of an impact that Intel had to establish a reserve fund to cover the costs associated with replacing the processors.

Intel introduced Apollo Lake in 2016, and the chips are based of Intel's Goldmont microarchitecture. Like many chips before and after it, Intel manufactures Apollo Lake on the 14nm process node. Apollo Lake might consist of paltry Celeron and Pentium models, but these chips are commonly found in many budget products, including desktops, laptops, 2-in-1s and AIO (All-In-One) PCs. The following is Intel's description of the problem from the official PCN 117143-00 document.

Intel identified an issue with the Low Pin Count (LPC), Real Time Clock (RTC), SD Card interfaces on Intel Celeron N3350, J3355, J3455 processors and Intel Pentium N4200 processor resulting in degradation of these signals at a rate higher than Intel's quality goals after multiple years in service.

For those of you that want to skip the mumbo-jumbo, Intel is basically saying there are reliability issues with its Apollo Lake processors where the 14nm chips are likely to die before the warranty is up. Having identified the problems, Intel is proceeding to refresh the entire Apollo Lake lineup. Current Apollo Lake chips with the B1 stepping are moving over to the new F1 stepping. These problems stem from the same LPC bus degradation issue that was present on the Atom C2000 family, and also impacted the company's E3800-series processors.

ModelSegmentCores / ThreadsBase ClockBoost ClockCacheTDP
Pentium N4200Mobile4 / 41.1 GHz2.5 GHz2MB6W
Celeron J3455Desktop4 / 41.5 GHz2.3 GHz2MB10W
Celeron J3355Desktop2 / 22.0 GHz2.5 GHz2MB10W
Celeron N3350Mobile2 / 21.1 GHz2.4 GHz2MB6W

The Celeron J3455, J3355, N3350 and Pentium N4200 are the affected Apollo Lake models. The first two models are desktop parts, while the remaining two are mobile chips.

Intel is urging its customers to transition existing affected products to the new stepping as soon as possible. To simplify the transition and avoid confusion between the new and old stepping, Intel adding the "E" suffix to the processor names. As a result, the processors with the F1 stepping will be marketed as Celeron J3455E, J3355E, N3350E and Pentium N4200E. If you're shopping and come by one of affected Apollo Lake SKUs, make sure the processor have the "E" suffix.

It remains to be seen if this issue is as widespread as the C2000 issue, and given that this also impacted the E3800-series processors, we might see even more Intel processor families impacted. We're reaching out to Intel for comment for further details, and will update as necessary.

  • NightHawkRMX
    And wasn't Intel just bashing AMD with degradation concerns?
    Reply
  • WildCard999
    remixislandmusic said:
    And wasn't AMD just being bashed for degradation concerns?
    Yup.

    https://www.tomshardware.com/news/amd-ryzen-3000-boost-clock-controversy-intel-attack,40231.html
    Reply
  • NightHawkRMX
    Intel is being a hypocritical lier, or in other words, Intel is just being Intel
    Reply
  • patbb
    admin said:
    Intel refreshes Celeron and Pentium Apollo Lake processors due to degradation concerns.

    Intel Apollo Lake CPUs May Die Sooner Than Expected : Read more
    You have totally misunderstood this Intel PCN. The current B-1 stepping parts are N3350, N4200, J3355 and J3455. The new F-1 stepping parts will add an "E". Your article is totally wrong.

    Launched Q3'2016 https://ark.intel.com/content/www/it/it/ark/products/95598/intel-celeron-processor-n3350-2m-cache-up-to-2-4-ghz.html
    Launched Q3'2019 https://ark.intel.com/content/www/us/en/ark/products/195254/intel-celeron-processor-n3350e-2m-cache-up-to-2-40-ghz.html
    Reply
  • Soaptrail
    remixislandmusic said:
    And wasn't Intel just bashing AMD with degradation concerns?
    WildCard999 said:
    Yup.

    https://www.tomshardware.com/news/amd-ryzen-3000-boost-clock-controversy-intel-attack,40231.html
    But this is totally different. Intel alluded without proof that Ryzen will have a shortened life where Intel CPU's have a shortened life that is costing them money! So totally not the same thing. Nothing hypocritical at all.

    /sarcasm
    Reply
  • King_V
    remixislandmusic said:
    And wasn't Intel just bashing AMD with degradation concerns?

    I wonder how much they're grumbling at the irony.

    And, that these are likely to fail within the warranty period.

    This can't have been a surprise internally. What lunatic decided to use an unconfirmed source to bash AMD with the possibility of slightly shortened lifespan, when their own stuff (admittedly the cheap, low end stuff) was something they'd (presumably before today) realized was likely to fail within warranty?
    Reply
  • TerryLaze
    Intel identified an issue with the Low Pin Count (LPC), Real Time Clock (RTC), SD Card interfaces on Intel Celeron N3350, J3355, J3455 processors and Intel Pentium N4200 processor resulting in degradation of these signals at a rate higher than Intel's quality goals after multiple years in service.
    Anybody that could maybe explain what these are :
    Low Pin Count (LPC),
    Real Time Clock (RTC),
    SD Card interfaces
    and how they could cause a CPU to die?
    Intel is saying that only the signal from these degrade would that kill a CPU?
    Reply
  • BFG-9000
    Those are all lines that lead directly into the 14nm CPU die, which simply haven't been designed with enough surge or voltage resistance to last through the warranty period--exactly the same problem as their 65nm Sandy Bridge 60-series chipsets had with the SATA II ports that died. Many of those died within months.

    For decades, I only saw one dead Intel CPU--a 350nm Pentium I which would only run when the L1 cache was disabled. And then I saw many dead 130nm Northwood Pentium IVs. More recently I'm seeing a lot of degrading 22nm Haswell chips with the fully integrated VRM. So Intel suggesting that AMD might have future problems just because they are at 7nm is ridiculous--it's all about design, in this case how well protected the delicate internals are from any high voltage I/O lines.

    At 14nm, even the 3.3v SD cards use is "high voltage."
    Reply
  • alextheblue
    King_V said:
    This can't have been a surprise internally. What lunatic decided to use an unconfirmed source to bash AMD with the possibility of slightly shortened lifespan, when their own stuff (admittedly the cheap, low end stuff) was something they'd (presumably before today) realized was likely to fail within warranty?
    Agreed.

    Side note, although it may be "low end", these chips often end up in non-PC applications like a NAS that are expected to run 24/7 for years. Whoops!
    Reply
  • refillable
    When you accuse others of something untrue: immoral
    When you accuse others of something untrue but it turns out you are the one doing it: idiotic
    Reply