Slashdot Mirror
Intel Recalls 1.13-GHz P-IIIs Due To Glitch
KuRL writes "C|Net is reporting that Intel has begun to recall their 1.13-GHz chips, which had the best clock speed on the market, due to a glitch that caused the chip "to malfunction in laboratory tests under certain conditions." Yes, it was only that specific. It is quite clear that Intel rushed this chip out upon hearing that AMD would be releasing a 1.1-GHz chip of their own."
According to Tom's Hardware (located withing the last couple of sentences on the page), the 1.13GHz perform fine when underclocked to 850MHz.
If you can read this, then I forgot to check "Post Anonymously".
I think Tom did the exact right thing. Kyle Bennett from HardOCP sent his CPU back, they returned what they said was the same chip, basically saying "we don't see anything wrong". Mysteriously, the returned CPU performed much better than it did before. Tom had made so much noise (and deservedly so), that he HAD to keep his proof or he could have gotten burned so hard.
Let's say that both Kyle and Tom returned their chips, had them switched with working ones, and Intel then fixes the problem quietly without a public recall (which could be easily done considering how few are really out there). Tom then looks like a total idiot who can't test a computer and Intel buys their way out of a huge mistake.
Tom, if you're reading this, you're awesome. Keep up the damn fine work.
-B
Intel has begun to recall their 1.13-GHz chips, which had the best clock speed on
the market due to a glitch that caused the chip "to malfunction in laboratory tests under certain conditions."
Intel executives today cited "ambiguity in language" as the source of the misunderstanding that caused them to release their 1.13 GHz chips a bit too soon.
"We thought that the glitch was what gave us the top speed," said one exec. "Really, that's the impression we had from the report our engineers gave us. And even some Slashdot headlines."
Sources say misuse of commas or failure to see them is becoming a widespread industry problem. "There's so much focus on the dot... some people think that's it, period, as far as punctuation goes. We really also need to look at the comma,"
said an economist somewhere in New York.
In other news, Intel also lost a bundle when they misunderstood some foreign currency prices...
Libertarianism is rich wolves and poor sheep playing gambler's ruin for dinner.
I guess it's time for us to dust off the f00f bug jokes that plagued Intel when the Pentium first came out....
Q: What's that 'Intel Inside' sticker called? A: A warning label!
Comeon folks, use 'em if you got 'em...let's get all this out of our system before Jay Leno and Dave Letterman get on the bandwagon tonight and ruin the fun for everyone...
The Tyrrany Begins....
Finding God in a Dog
I am a fan of the Athlon, and have been more pro-AMD since the overhyped and under-performing Pentium II was released.
So I shed no tears about this recall. Intel's position as a 2000lb gorilla needs to be shaken; they have dealt with issues like the 486DX50, and Pentium FPU bug in a less-than-graceful way.
This product was a crude attempt to FUD AMD out of the market with what looks like an overclocked part. But AMD has been guilty of faulty components as well.
An earlier poster mentioned some bad K5 CPU samples. I would go so far as to claim the entire K5 line with its "PR" rating was a joke. The performance was poor and the failure rate high.
At that time AMD was willing to do anything to stay afloat...this continued into the earky K-6 (less than 300Mhz) era. But since then AMD has been producing an enterprise-level CPU, even if some of the motherboards for the Athlon were not up to snuff.
My point is that both Intel and AMD have at times traded reliability for release dates. We should not let them forget it, but nor should we rule out one manufacturer's product on the basis of rumor or a recalled product.
Wow. Congratulations to Tom from Tom's Hardware and Kyle from Hard|OCP. Somehow I have a feeling that this recall is more due to your findings than to "tests in laboratory". At least Intel reacted the good way realising they made a big mistake and did the right thing by recalling. What annoys me is that they had to wait until some chips had been shipped before realizing their CPU still had issues.
I wonder how those who flamed Tom for his negative review feel now...
"If liberty means anything at all, it means the right to tell people what they do not want to hear"
Recall? What are they recalling? They have to actually have sold some of the chips before they can recall them. Even Dell, Intel's lapdog, has been unable to get these chips in any kind of volume. This "product" was pure vaporware from day one.
I am not an idiot. Please use my name to email me.
"That's right, I'm quoting myself."
-Upsilon
Well, I still feel good about not liking Tom and his reviews. It is not so much that he found the bug but the fact that "I'm going to keep this as evidence and not help Intel" and his general approach to this situation...
And his attitude shouldn't be what it is considering what he is doing
Speaking as someone who has been following hardware sites since Tom's page didn't even have its own domain...
Tom used to be noticably pro-Intel. This was true until somewhere in the AMD-K6/Pentium Pro days, when he started noting that the K6 would run at 100 Mhz bus, and the socket 7 Pentiums wouldn't. He then gradually started becoming more pro-AMD and less pro-Intel.
I see two possibilities here.
1. He judges things purely on technical advantages, and really thinks AMD is better.
2. Someone at Intel pissed him off somewhere along the line, and he has a personal bias.
It's also possible that both of the above are true.
There is one thing I can't stand about his site - since he added the daily "technical news", I have seen news bits that are obviously press releases, and contain obvious technical errors that Tom should notice. I don't think he actually reads some of the stuff before it gets posted as "news".
I do like the reviews and articles that are actually written by his staff, though. I think keeping proof of the situation was important in the case of the 1133 Mhz chip. I think if he would have sent it back the problem may have dissapeared, much like it did with Kyle's chip. I think when he sent it back they did a microcode load or something, then returned it saying there was nothing wrong with it.
None of the four chips that this group of reviewers got would compile the Linux kernel. When Kyle slowed them down to 850 Mhz they worked. This points to the chips being flaky at their rated speed.
As far as Tom's attitude, it is the sum of all his experiences, just like anyone elses is. Though he could make a concious effort to change it, I think his honesty is important in an industry where too many reviewers simply regurgitate praise.
--- "So THAT's what an invisible barrier looks like!" - Time Bandits
NEW YORK-- Intel Stock (INTC) tumbled a quarter of a point when the corporation released news that it was recalling ALL of its new 1.13GHz CPU's today.
The 1.13GHz CPU, or Central Processing Unit, is the chipmaker's fastest consumer product to date. It is unclear as to how many of the units have shipped over the past few weeks, but based on the company's current yields, industry experts predict it may be as many as a dozen.
-CausticPuppy "Of all the people I know, you're certainly one of them." -Somebody I don't know
Tom blew the whistle, but HardOCP supported his findings. Older readers of both websites know that these guys were quite "antagonistic" before they run into each other at some (Asian?) computer show a few months ago.
In this instance however Kyle from HardOCP supported Tom, as he had bad experience with the chip as well. And eventually HardOCP hosted the testing session with 3 CPUs (one from AnandTech, credit to them too) and an Intel's engineer, where the problems were confirmed.
I'm just pointing this out since HardOCP is relatively low-profile site, and they deserve some credit here. They are not as thorough as some other review sites but they sure are fun to read.
This is an interesting examples of one of the pitfalls of being a market driven company instead of an engineering driven one. Intel's marketing department tried to score a win over AMD and succeeded in jeopardizing their own reputation. Surely Intel's engineers knew the chip wasn't ready for prime-time but here business needs won out over engineering. AMD returned to their engineering roots out of necessitiy after their difficulty in the early-mid 90s and it is clear that they know set the tone. This also underscores the strength of sites such as Tom's and the Open Source community when it comes to keeping big corporations honest. While Intel may be able to pull the wool over the public's eyes (much as M$ does) they can't pull it over everyone's and now we are able to shout load enough for them to listen.
Ah, there's lots of gloating going on here. Much of it includes "rah, rah, AMD" sentiment, I see. I can understand it, but some caution is well deserved here. The x86 chips that are being released these days are immensely complicated. Windows is buggy, yes, but if someone wrote a less buggy clone of Windows, it is still going to be an overly complex piece of software that I would not stake my life on. AMD is prone to exactly the same issues. It's not like AMD chips are orders of magnitude simpler than what Intel is working on. AMD could easily stumble at any moment, as could anyone working on something so complex. There's a bit of a blind eye toward and AMD problems at the moment, too, just as patches released right after new Linux kernel get spun into "Look at the quick turnaround!" instead of "Critical bugs in Linux kernel!."
The key difference being that Ford underestimated the magnitude of the lawsuits AFTER they'd put the cars on the market, not knowing that they were prone to blow up on people.
Incorrect. Ford's engineers were fully aware of the fault before the Pinto's release. The problem was a poorly shielded gas tank that tended to get crushed in rear-impact collisions, causing an explosion. The engineers brought the information to management, along with a proposed (and inexpensive) fix. Management did the math on the number of deaths, how much each would cost them in lawsuits and PR, weighed it against the per-vehicle cost of the fix and decided to produce the vehicle unmodified.
The Pinto is now a text-book example in engineering ethics. The really chilling part of the story is that the primary motivation for not including the fix was that it would push the car's cost up enough to force a reprice. Marketing had been advertising the Pinto's low low target price like mad and weren't about to back down. Thus, the deadly defect remained largely due to marketing promises.
While it's easy to villify Ford or any other company for making decisions like these, it must be remembered that this is the extreme case of a common phenomenon. It isn't surprising that by spending more money on a car, you can design it to be safer. You expect a Volvo to be safer than a Kia, don't you? If the engineers so liked, they could keep lopping life-saving features onto a car until it 1) looks like a tank, and 2) can only be afforded by the very rich. Clearly, this is taking things too far. The line has to be drawn somewhere, and equations based on cost and fatalities are the tool used to make these decisions.
In the Pinto case, however, Ford went too far. The gas tank problem was a severe one, and it lowered the safety of the Pinto below consumers' reasonable expectations of safety. That is the key. Consumers may have expected the inexpensive Pinto to be less safe than a luxury car, but no one expected it to be as unsafe as it was.
Anyway, this Intel mess is really a whole different ball game. Failing on speed promises like this isn't life threatening to anyone, so the stakes from an ethical viewpoint drop markedly. (Don't tell me about air-traffic control, none of those people would touch these chips) This is really just another case of Intel throwing vapor around. It surprises no one.
--Lenny