Do NOT buy a Pentium-4.
December 28, 2000 9:51 AM Subscribe
Do NOT buy a Pentium-4. Wait until the second version (different pinout) next summer, or buy an Athlon or a P3.
Wait. I know the P4 is a bogus upgrade, but isn't the article just about problems with PCI, video cards and the P4?
posted by capt.crackpipe at 4:29 PM on December 28, 2000
posted by capt.crackpipe at 4:29 PM on December 28, 2000
Um, no. They found a second manifestation of that same bug, only this one's worse. It was exposed using a hardware DVD decoder which tried to use a sound card on the PCI bus to do surround-sound decoding. The result was that the bus got wedged and there were lots of sound dropouts. What's weird about it is that it worked with a previous sample board, and it works with slower PIII boards (same sound and DVD board). The problem appears to have been introduced at the last minutes somehow. Intel has duplicated and acknowledged it.
But the real killer is that second article I posted; the guy does a really decent job of showing that Intel basically blew it on the P4 architecture. It's long but it's fun to read if you're a geek (like me).
posted by Steven Den Beste at 7:01 PM on December 28, 2000
But the real killer is that second article I posted; the guy does a really decent job of showing that Intel basically blew it on the P4 architecture. It's long but it's fun to read if you're a geek (like me).
posted by Steven Den Beste at 7:01 PM on December 28, 2000
Interesting. I always thought AMD made second-class chips that weren't 100% Intel-compatible (I've only paid close attention to Macs), and could never figure out why the vast majority of the PCs I see advertised these days have AMD chips. I guess this is why.
posted by aaron at 10:40 PM on December 28, 2000
posted by aaron at 10:40 PM on December 28, 2000
aaron - incompatibilty was basically a myth. AMD's real problem was that through the K6 generation of chips they had processors which were cheaper and slower than the Intel chips; they tended to be used most for bargain system. It's been amusing to watch the reversal in the last year - currently Intel's products are more expensive, [often significantly] slower and less reliable.
Incidentally, this is also why the Apple's comeback is floundering. Intel's chip lineup is weak and it's only as good as it is because of AMD. Intel released a 1GHz chip to keep pace with AMD's announced chip and then failed to ship any actual chips for months. When they did ship 1Ghz PIIIs, they had a ton of QC problems, largely because they were shipping what is basically an overclocked design to match AMD in the clock-rate wars. Without AMD, Intel probably would have either shipped the 1Ghz PIII 6 months later or skipped it entirely for the P4.
Ironically, had AMD remained a niche player, Intel & Motorola would have stumbled around the same time and little would have changed. Instead, AMD's offers something like a 3:1 speed advantage for your dollar vs. the G4 (by comparison, a PIII at the same clock rate is ~2.5 times more expensive and slower).
posted by adamsc at 1:18 AM on December 29, 2000
Incidentally, this is also why the Apple's comeback is floundering. Intel's chip lineup is weak and it's only as good as it is because of AMD. Intel released a 1GHz chip to keep pace with AMD's announced chip and then failed to ship any actual chips for months. When they did ship 1Ghz PIIIs, they had a ton of QC problems, largely because they were shipping what is basically an overclocked design to match AMD in the clock-rate wars. Without AMD, Intel probably would have either shipped the 1Ghz PIII 6 months later or skipped it entirely for the P4.
Ironically, had AMD remained a niche player, Intel & Motorola would have stumbled around the same time and little would have changed. Instead, AMD's offers something like a 3:1 speed advantage for your dollar vs. the G4 (by comparison, a PIII at the same clock rate is ~2.5 times more expensive and slower).
posted by adamsc at 1:18 AM on December 29, 2000
Since I don't run DVDs on my machine I guess it seems sort of inconqsquential. Don't use a P4 with DVD hardware and a sound card trying to send out surround sound. The audio might be choppy. Sort of like the old 2.01 - 2.00 = 0 bug.
How many bugs would it take for Intel to send out a recall? Or would machines have to start catching on fire?
A quick google search for "‘pentium 4’ bug" returned ~45,000 documents, while "‘pentium III’ bug" returned ~25,000.
posted by capt.crackpipe at 3:41 AM on December 29, 2000
How many bugs would it take for Intel to send out a recall? Or would machines have to start catching on fire?
A quick google search for "‘pentium 4’ bug" returned ~45,000 documents, while "‘pentium III’ bug" returned ~25,000.
posted by capt.crackpipe at 3:41 AM on December 29, 2000
Both of the known manifestations of this problem show that the Southbridge of the i850 is deeply flawed. It's only a matter of time until yet another case of it shows up. For instance, if you have two SCSI cards and are doing a mass copy from one to the other, will it work? Intel says "yes" but I don't know if anyone's actually tested it. There might be massive data corruption instead. Another example is TV cards (e.g. Hauppage), which use the PCI bus to send the TV graphics to the display card. I think there's a chance that when someone starts trying those they also end up having problems.
Intel recalled a million cards last year because of the MTH problem with the i820. I think Intel's between a rock and a hard place on this one, because a recall will seriously erode confidence in their products, but a refusal in the face of known bugs will equally erode confidence in them as being trustworthy.
This is just the latest in a long line of amazing screwups Intel has made in the last three years or so. Watching Intel melt down has been one of the most surprising experiences I've had while observing the industry. How could a company which had performed so well for so long suddenly start screwing up so badly?
posted by Steven Den Beste at 7:42 AM on December 29, 2000
Intel recalled a million cards last year because of the MTH problem with the i820. I think Intel's between a rock and a hard place on this one, because a recall will seriously erode confidence in their products, but a refusal in the face of known bugs will equally erode confidence in them as being trustworthy.
This is just the latest in a long line of amazing screwups Intel has made in the last three years or so. Watching Intel melt down has been one of the most surprising experiences I've had while observing the industry. How could a company which had performed so well for so long suddenly start screwing up so badly?
posted by Steven Den Beste at 7:42 AM on December 29, 2000
« Older I think they got a bargain. | Newer »
This thread has been archived and is closed to new comments
posted by Steven Den Beste at 10:21 AM on December 28, 2000