45nm Core 2 Quad launch delayed due to erratum?

Processor bugs, or errata, seem to be an epidemic lately. After AMD’s erratum number 298 found its way onto all quad-core Opterons and Phenoms, word is now going around that Intel has also found a bug in its upcoming 45nm Core 2 Quad processors, causing it to postpone the chips. French enthusiast site Hardware.fr has the skinny:

According to our sources, dual-core Penryn processors (code-named Wolfdale) will indeed launch in mid-January, but quad-core versions (code-named Yorkfield) won’t become available until late February/early March. Behind this delay lies a bug that, in some very rare cases, could cause a system to crash. Although the bug has only been detected in the lab through a new validation procedure, Intel has decided to fix it before the chips’ launch.

Hardware.fr goes on to say that Intel’s recently launched 45nm Core 2 Extreme QX9650—which is still up for sale—is also hit by the bug. However, quad-core 45nm Xeons reportedly aren’t affected.

We asked Intel to comment and received this response from the company’s PR manager, Dan Snyder:

45nm Core 2 Quad launch is planned for Q1’08, and we are still on track for that. We can’t comment on web speculation.

The late February/early March schedule mentioned by Hardware.fr is a far cry from the January 20 date that’s been quoted on other sites, but it still fits within Snyder’s Q1 2008 time frame. If Hardware.fr is right and the chips have indeed been postponed, though, that could give AMD enough time to prep for Intel’s 45nm assault by rolling out bug-free, B3 revision Phenoms and introducing models with higher clock speeds. (Thanks to TR reader Flying Fox for the tip.)

Comments closed
    • cripplecore
    • 12 years ago
    • green
    • 12 years ago

    unless of course the source got it wrong and it was a case that intel is delaying 45nm quads due to an amd bug resulting in no competing product til march

    i’ll be waiting for nehalem performance details to come out before making an upgrade of either amd or intel anyway
    so if current quads are buggy i don’t care as long as they have it fixed it by around ’08q3/4

    • pluscard
    • 12 years ago

    I have to congrat TR for at least posting “the other side of the story”.

    This is not the popular opinion, but it’s Intel that has everything to lose, while AMD has everything to gain. AMD just got an 8% stake from the arabs, and if you’ve been following them, they seem to quickly raise any amount of money it takes to stay in biz thru the design cycles.

    While Intel longs and employees hope like hell that AMD goes bust, so Intel can get back to biz as usual, AMD has continued to gain share virtually every quarter for the last 4 years.

    The cpu biz is extremely high over head, high fixed cost biz. Once you hit enough volume to break even, the profits quickly mount. Conversely, if Intel loses share to the point where it can’t keep it’s fabs full, it’s profits will evaporate overnight.

    While I congratulate Intel on a fine job with the C2D, remember that 45nm is not a new design, it’s a new process. It shouldn’t require debugging like the brand new K10/spider platform will. I’m confident once the K10 and spider mature, people won’t be able to say enough good things about their buddies at AMD.

    Plus

      • orthogonal
      • 12 years ago

      l[

      • Flying Fox
      • 12 years ago

      q[

    • wingless
    • 12 years ago

    Wow! Intel has so much money they can afford to fix the problem BEFORE the release of the chip. AMD bought ATI and ruined things in the short term for themselves. I hope Bulldozer and Fusion and the new 45nm process tech pay off. (Before you chime in about 45nm, AMD is using all new materials for 45nm than they did with the doomed 65nm.

    If Intel had bought ATI, I wonder how well they would be doing now LOL….

    • cegras
    • 12 years ago

    Am I seeing deja vu?

    I think I’ve read boing’s comment at least 5 times by now.

    • Mr Bill
    • 12 years ago

    Will there be a patch? Maybe we could do a Core 2 Quad patch versus Phenom TLB patch runoff?

      • Flying Fox
      • 12 years ago

      The fact that they are delaying the launch (if true) to fix this means no shipping CPUs will be with this erratum. Of course, there will be others that they ship with.

        • eitje
        • 12 years ago

        q[

          • Mr Bill
          • 12 years ago

          Precisely 😉

          • cripplecore
          • 12 years ago

          No, they DON’T say that.

          They note that it is still for sale, despite being a 45nm Yorkfield.

          And hardware.fr has retracted its “intel confirmation” of the rumor.

            • Mr Bill
            • 12 years ago

            Cyril quotes it above in the article. Here it is cut and pasted from the translated link…
            q[

    • boing
    • 12 years ago

    I stand corrected. 🙂 I was referring to the price-tag of the K6-III and the Netburst-CPU’s being able to replace your stove for cooking.

    • mortifiedPenguin
    • 12 years ago

    Any word on exactly what kind of errata this is? If it was one similar to AMD’s 298, that would actually be kind of funny.

    • flip-mode
    • 12 years ago

    One way or another, this helps AMD’s credibility.

      • Flying Fox
      • 12 years ago

      One way or another, this should make people realize that hardware != no bugs. They have bugs just like software. People just have to learn to accept it or never upgrade from their “perceived” bug-free hardware.

        • Krogoth
        • 12 years ago

        Yep, hardware always had bugs. It is the only the major ones that get all the attention. The reason why the minor bugs are non-issues to end-user is all from errata and microcode patches. They typically come before the hardware is officially released or shortly afterwards.

        QA also plays a large role during designing stages. Silicon typically has to get re-tape a few times in order remove the nasty bugs. It is becoming more different to do as chips become more complex.

          • Mr Bill
          • 12 years ago

          I suppose the major difficulty hardware presents is that any fixes in subsequent tape-outs must fit into the same volume (area & layers) as the original?

        • flip-mode
        • 12 years ago

        I’m fully aware. I don’t know if everyone else is.

    • boing
    • 12 years ago

    I’ve used nothing but AMD since 1999, and don’t regret at all that I bought an Intel C2D e6750 two months ago. I’m neither any Intel nor AMD fanboy, and I think the Phenom is the biggest crap AMD has released since the K6-III.

    The Netburst-archictecture was mostly a piece of crap unless you were into video-editing. Intel gaining back the performance crown _[

      • d2brothe
      • 12 years ago

      I agree with you on pretty much all of those points, but I would have to say the K6-III wasn’t bad, it performed well for an AMD processor of the time. Also…the Northwood based netburst chips weren’t that bad…they competed well with the then current Athlon 64s

        • ludi
        • 12 years ago

        Ditto; the K6-III was, in fact, an excellent processor. It just happened to be a bit ahead of its time and consequently all that cache made it run hotter than a Phoenix manhole cover. When the K6-3+ came around, it proved quite handily that all that processor ever needed was smaller process technology.

          • swaaye
          • 12 years ago

          If it was such a wonderful chip, why did the 180nm K6-III+ only officially do about 500 MHz, even with a much simpler cache than Coppermine? That would be because the core wasn’t designed well enough to scale much higher in clock speed. They don’t overclock much past 600 MHz, and the two I’ve had only do that with extra voltage.

          K6-III was only decent for office apps, really. The FPU was awful, and 3DNow! was a lost cause. As such, its performance was quite poor in 3D games and, as you might imagine, any app needing something beyond integer performance. It wasn’t good competition for P3 Katmai, let alone Coppermine.

          Oh, and the platform was horrible beyond words. Awful chipsets, terrible cheaply-built boards. AGP barely worked, PCI was slow, RAM bandwidth efficiency and latency was hardly better than that of a P5, courtesy of the retro mobo cache architecture and bad chipsets.

          K6-III wasn’t a smart buy, but it was the best AMD could put out on that architecture and platform. The CPU was way too expensive for what you got though.

            • bhtooefr
            • 12 years ago

            You do realize that that design was only meant to do 180 MHz, as the NexGen Nx686, right?

            550 MHz (IIRC, the fastest stock speed for any K6 family chip) is pretty good for that design.

            NetBurst is much more of a POS than that – it was supposed to hit 4 GHz from the start, and the 4 GHz Pentium 4 never happened.

    • flip-mode
    • 12 years ago

    Hello, Captain Obvious, was #4 just joking?

      • provoko
      • 12 years ago

      I don’t think so. Lets try to come up with as many things to defend Intel or attack AMD with. Haha.

      • JJCDAD
      • 12 years ago

      Uh roger that Lieutenant flip-mode.

    • JJCDAD
    • 12 years ago

    Intel is doomed! OMGBBQ!!!!111ONE!!!!

      • UberGerbil
      • 12 years ago

      Of course not. Unlike certain other companies, Intel isn’t losing money every month.

        • JJCDAD
        • 12 years ago

        It was a joke. I thought the OMGBBQ would give it away. 😛

          • provoko
          • 12 years ago

          Haha. I thought it was funny. I was going to do that to the seagate article but I thought no one would get it.

          I liked your use of the word “one” instead of the number “1”. Haha.

      • BoBzeBuilder
      • 12 years ago

      What have you been smoking lately? Have you seen intels latest financial results? even if there was an erratum in all of their lineup of cpus, they still wouldn’t break a sweat. Its AMD that’s going down man.

        • provoko
        • 12 years ago

        Have you seen Caterpillar’s or GM’s? When did this site become SR Stock Report?

        Grow a sense of humor.

      • jodiuh
      • 12 years ago

      lol, you forgot chopper

    • Anonymous Coward
    • 12 years ago

    How could it not effect Xeon?

      • d2brothe
      • 12 years ago

      *shrugs*…there are differences between the two processors, it is quite possible.

      I have to laugh at this…where are all the doomsayers that are harping on AMD all the time.

      Regardless this is somewhat of a reprieve for AMD, like the article said, it should give them some time to get the B3 revision out the door and ramped up.

        • UberGerbil
        • 12 years ago

        Intel has had errata before. The difference is, Intel is profitable and not in debt. The doomsayers aren’t just looking at the two companies’ technical positions, they’re also looking at their financial positions. Intel can afford to stumble now and then; AMD cannot.

        • SPOOFE
        • 12 years ago

        “where are all the doomsayers that are harping on AMD all the time.”
        Acknowledging Intel’s rampant profitability, perhaps?

        • provoko
        • 12 years ago

        The AMD doomsayers are just Intel fanboys that were deprived of performance for nearly 10 years. They’re just getting their kicks, punches, catapults, trolls and Altair like assassins in.

          • d2brothe
          • 12 years ago

          I’m more refering to how in general, most comments about the errata for the Phenom have followed the pattern of “OMG…this is terrible…what a piece of crap CPU…why would anyone buy AMD right now…its totally broken and nonfunctional”…where clearly…its no better than intel in that regard…

            • boing
            • 12 years ago

            The difference is that Intel didn’t release a CPU with a bug that they considered serious enough to release a “fix” later on that impacted performance as much as >50%.

            • d2brothe
            • 12 years ago

            Actually…you don’t know that…Intel may yet release a fix. Also, saying that the patch causes up to and over a 50% decrease in performance is…missleading, in general, its around 10%…in one application…which has questionable memory performance to begin with…the patch performs particularily poorly.

            • SPOOFE
            • 12 years ago

            Yes, they may yet. But they haven’t. So the situations are different. Why would one expect the same reactions to different situations?

            • ludi
            • 12 years ago

            If you were expecting anything like consistency out of the people who post THOSE sorts of comments, then I think you should prepare for disappointment 😉

            • smilingcrow
            • 12 years ago

            Unfortunately AMD’s erratum is just the tip of the iceberg. Intel already has decent performing quads available at 65nm which over-clock well and run at a fairly low TDP. AMD on the other hand have their own erratum, a design/process that doesn’t clock very high, a PR nightmare and a high TDP; so not exactly a comparable situation.
            If Yorkfield is delayed by two months I don’t see it significantly helping AMD. Although at this stage I think they’ll be grateful for any break that they can get.

      • DASQ
      • 12 years ago

      The Xeon’s are on a different production line?

      And here I was hoping to pick up a Wolfsdale in February :/

        • Flying Fox
        • 12 years ago

        Wolfdale is fine and on track for the January launch.

          • DASQ
          • 12 years ago

          I meant Yorkfield. Oops.

      • UberGerbil
      • 12 years ago

      Perhaps because the problem is avoided by the circuitry that keeps caches on multiple MPU packages coherent, which is disabled in processors destined for single sockets? Just a guess, but there really are differences between the two.

Pin It on Pinterest

Share This