“Recently, there have been a couple of articles on errata making the rounds, the first on Intel, the second on AMD. They both make my head hurt. A lot. No, not the errata, but the explanations, and lack of understanding of some simple concepts. When the first one, Intel, came out, I was teetering on the verge of breaking out the cluestick, but thought the better of it. When the conspiracy theory second article came out, well time to spring into action, two days late as usual.”
…is the amount of errata on the new Core Duos, something like 30 in 20 days?
And the P4 only had 65 in it’s entire lifetime?
Is this begining a new trend in even lower quality processors from Intel?
Will Mac users have to face exploits based upon these unknown errata?
What about the hyperthreading exploit?
Is Microsoft totally to blame for exploits or is it partially Intels and Microsoft’s inability to move fast enough to fix them?
Edited 2006-01-29 00:26
The Core Duo has been out for a long time. Processors are available to the manufacturer’s testers and partners long before they are launched to the public. Several iterations of the Core Duo have likely gone through the hands of testers over the last several months. Since the bulk of processor errata are usually discovered during this phase, the idea that “34 errata within 20 days of launch” means that many more will pile up in the coming months is inaccurate.
Wait… what?
Did you just find a way to blame Microsoft for something that is Intels fault (and possibly Apple as well?) and they have NOTHING to do with?
Edited 2006-01-29 01:12
Good grief. *Please* read the article. The author basically says that all of this is standard, and both Intel and AMD have been doing it for a long time, and will continue to do it long into the future.
Neither AMD’s or Intel’s processor is low quality, these are bugs that no user will see, because the people writing the Operating System, or the BIOS, or the motherboard makers *use* these documents that everyone is flinging around as fodder for conspiracy to make sure that everything works smoothly.
Ok cool, just in the early release.
*crosses fingers*
Two thumbs up for the daily shreed of reason!
What’s got me shocked is that this needed to be explained. I’m no CPU guru, nor do I claim to be (tho I do love reading about the technical stuff) but its the same procedure in almost any product based field. They all have revisions, they all have erata sheets, and most of the time its standard practice not to remove the original stuff.
What, never read a patch update or readme document?
Some of my mac using friends (I am a mac user as well) want to hate on Intel for some reason, and are severely against Apple’s recent addition of Intel based machines to their lineup.
They posted the article about the OMG BUGS IN CORE DUO in our IRC channel and went on and on about how lazy Intel’s designers must be. I went out to Freescale’s website and grabbed the latest errata sheet for the processor in my Powerbook, and then went to IBM’s site and grabbed the latest errata sheets for the G5s in their G5 machines. The latest PowerPC 970 had bugs in the high 20s, and the Freescale 7455 chip has, I think, 42 bugs currently, even though it has been out for years.
The long and short of the story matches up with the original article. Processors have bugs, and us they generally won’t affect anyone except low-level coders such as kernel developers. This is much ado about nothing.
The long and short of the story matches up with the original article. Processors have bugs, and us they generally won’t affect anyone except low-level coders such as kernel developers. This is much ado about nothing.
And even then, there are ways to work around the problems, using microcode updates etc. so it doesn’t mean its all doom and gloom, and given the high performance of todays processors, any work arounds would have negiable effect on the over all performance.
Now, as for the ‘whinefeast’ by those who complain about the bugs in PowerPC, PPC970 and Intel – sure, we *COULD* have relatively bug free cpu’s, but there will be a massive price tag associated with them – hence tteh reason why it costs more to purchase a SPARC processor or a POWER one – increased time spending testing and debugging results in more costs that have to be recooped via the product line up.
In the end, I think it is best to get everything into perspective in regards to some of these so-called bugs; like software bugs, it isn’t the issue of bugs, but how they’re handled
“hence tteh reason why it costs more to purchase a SPARC processor or a POWER one”
You are right, if they were to fix every single bug (i.e. find a bug, start producing new revision) the chips would cost massive amounts of money.
But you are also wrong, a SPARC or a POWER cost more because they are manufactured in smaller quantities. They both have as many bugs as an AMD or Intel processor (one can even argue that AMD and Intel chips get *a lot* more testing since they end up in motherboards and using BIOS software from a wide variety of manufacturers).
The bottom line is… *all* processors have bugs, that’s just the way it is.
#1 64 isn’t the number the I recall regarding the PIV. AFAIR the PIV was about a year when it speeded past the PIII, that had crossed the 100 error limit. I’m puzzled, where have all the errors gone ?
It depends on how you count it:
The “Mobile Intel Pentium 4 supporting hyper-threading on 90 nm process technology” has 56 errata listed.
The “Intel Pentium 4 processor on 90 nm process technology” has 38 errata listed.
The “Mobile Intel Pentium 4 processor-M” has 92 errata listed.
These are the only specification updates I have on my hard drive (there’s probably more I haven’t got yet, for newer CPUs).
Now, if you add them all up you’d have more than 100 errata items, but some of them would be duplicates.
If you pick the worst, then you’re looking at 92 errata items. If you read the fine print (the summary of changes) you’ll notice that this specification covers 8 different steppings, and there isn’t one single stepping that is effected by all errata. The worst is the B2 stepping which is effected by 71 of them. The best is the D1 which is only effected by 40 of them.
-Brendan