Slashdot Mirror
Top Ten Intel Slipups
quickquack sent us a story on tuplay about Intel's top 10 slipups. They all seem to be relatively recent mistakes (rambus, serial IDs etc) so I'm curious if anyone out there can remember some older slipups (hell the company has been around long enough to have some big screwups). Anyway, the article is also somewhat conspiratorial in tone, in an amusing sort of way. You'll enjoy it. Plus its always fun to laugh at Intel *grin*.
It's a psychological thing. Slashdot readers are usually nerds right? This probably means that they've been badly bullied at school, right? So now there is Slashdot. Hiding behind a nickname, or worse anonymously, we can kick back at the bully, in this case Intel, in other cases Microsoft.
People replying to my sig annoy me. That's why I change it all the time.
As Linux (hopefully) takes more of the desktop mindspace, we'll see MicroSloth FSCK-ups giving market share away to the Linux world, as well -- for the same reason: Residual arrogance.
The Linux world makes mistakes, too. It's just that we don't (usually) push those mistakes on the public as if they're the next best thing since sliced bread. Most of them get caught in the inherent quality control system of the Open/Free Source model.
`ø,,ø`ø,,ø!
Free Software: Like love, it grows best when given away.
I'll end this one with an interesting note -- Intel's palmtop CPU chip series is named StrongARM... is that supposed to be some sort of joke?
If I'm not mistaken, the StrongARM isn't an Intel invention. Rather, it's a version of the Acorn Risc Machine processor which first appeared in the Acorn Archimedes during the late 1980s.
ObIntelJoke:
The Pentium's floating point unit is said to comply to IEEE fp standards. If you're on an airplane with a Pentium-powered air data computer, how is IEEE pronounced?
Aieeeeeeeeeeeee!
Oolite: Elite-like game. For Mac, Linux and Windows
i can never remember which is big & which is little endian...
intel's byte order gives a very small bit of extra code efficiency when, say, converting a 32-bit number to a 16 bit number in memory. if the bytes are the other way around, you have to add two to the address.
i can't think of a real benefit to the other way around. i suppose nowadays an add instruction takes no time anyway, but when they first designed the processors that wasn't true.
Yeah it was difficult, but all the idiots that used the same heatsink and fan combos that were designed to cool a 66-MHx i486 processor got what they deserved when their 66-MHZ Pentium processors overheated. Putting a computer together isn't like working with legos, you know. A little reading would have gone a LONG way in preventing the meltdowns.
Ever seen a Compaq P60? Or a Gateway P60? These weren't clones with crappy cases, 1" square nameplates, and assembly quality like that. These are mass-produced machines, designed by engineers, who had probably read all of Intel's docs. And I've seen lots of them where the active heatsink just couldn't keep up.
I remember an IBM server with the new cutting-edge P66. Now, towards the end of the 486 era, a lot of the manufacturers (as opposed to clone builders) were getting to large passive heatsinks and even interesting fan solutions. The old HP Vectra 486 machines had a fan mounted above a heatsink-less processor. The fan blew air across the top of the processor at a 45 degree angle.
But all the early Pentium machines that I ever touched had active heatsinks, just like most of today's processors. From the real manufacturers, they had somewhat more substantial fans than the "Ball Bearing" made-in-Taiwan crap that clone builders (still) use.
And this IBM was no exception. There was nothing wrong with the fan - it still spun, and it was a good quality piece. It was a Panaflow 12VDC brushless fan mounted to a 2" tall extruded aluminum heatsink, strapped onto the top of this processor. When I got to the machine, the remains of the blades of the fan didn't show any signs of dirt, dust or cigarette smoke accumulation. The fan even spun freely. But the plastic frame of the fan was warped and the blades had been distorted. It made a hell of a vibration when it started up, and the owner had brought the machine in for us to fix because the computer was noisy and smelled really bad of something melting. I replaced the fan (Nidec cast-aluminum with a composite impeller, $30 fan, use them in all my computers to this day), re-routed some ribbon cables away from the processor to see if I could give it more airflow, and then my co-workers and I were betting on when it would come back. It didn't.
The FDIV bug impacted 60-MHz, 66-MHz, 75-MHz, 90-MHz and 100-MHz Pentium processors.Yeah, okay. I got ahead of myself, I apologize. It was late 1994 when Intel fixed the FDIV bug. The Pentium 75 was commonly shipping at the time; the P90 was popular, the P100 was the high-end chip, and the P120 was just on the horizon.
I've never noticed a P75 with the bug, but because of where I've worked, we didn't have many of the 3.3V-generation Pentiums to support until around the P100. And that was well into 1995.
The 120-MHz Pentium processor was the first processor to *not* have the FDIV bug impacting it.I think it's fairer to say that the P-120 is the first one that has never had a release with the FDIV problem. I assure you, Intel has made P75s to P100s without the FDIV.
But all of the high-voltage Pentiums were affected; unless they were replaced under Intel's recall, they had the FDIV flaw.
Fire and Meat. Yummy.
actually, the 8-bit bus on the 8088 made the hardware much cheaper, which greatly increased the affordability of the system.
unless you're gonna say that the whole ibmpc arcitecture set back computing, which is probably debatable.
Using a 8086 instead would do no good. The 1MB limit is because the segmented architecture consisted on left-shifting the 16-bit segment address by 4 bits, hence no resulting logical address could be greater than 0x0FFFF0 + 0x00FFFF = 0x10FFEF (roughly 1.06 MB).
This is true for the 8086, 8088, 80186 and 80188 families. True flexible segmentation was introduced by the 80286, although in a completely boneheaded way (no switch back to real mode, duh!)
This isn't a real top 10; it's a half-hearted attempt. A complete list couldn't fit in the margin of this webpage, though.
/proc/cpuinfo
Let's start looking...
* Segmented Memory
* Byte-swapping
* To this day, ensuring backwards compatibility with chips no one used
* Screwing over their customers
* Screwing over their employees (see Inside Intel)
* RAMBUS (ha ha ha!)
I'm sure I missed a bunch...
...oh yeah:
[pb@Lee-12-240 pb]$ cat
processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 4
model name : AMD Athlon(tm) Processor
stepping : 2
cpu MHz : 800.060074
cache size : 256 KB
fdiv_bug : no
hlt_bug : no
sep_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 psn mmxext mmx fxsr 3dnowext 3dnow
bogomips : 1595.80
---
pb Reply or e-mail; don't vaguely moderate.
pb Reply or e-mail; don't vaguely moderate.
Indeeed. We can look at this another way:
Intel's Biggest Achievements
1. Invention of the microprocessor. TheStreet.com called this the second most important business event of the CENTURY (only the interstate highway system outranked it). Indeed, this is the single invention which enabled the information revolution, without which none us would be here.
2. Invention of semiconductor RAM. Without Intel, memory would be made up of massive arrays of core memory. Intel showed the world that you could build memory out of chips. Again, this helped enable the PC revolution.
3. (sort of) Invention of the integrated circuit. Robert Noyce is credited as the co-inventor of the integrated circuit (with Jack Kilby), and was the first to build one out of silicon. Though he was at Fairchild at the time, he later went on to found Intel.
4. Pentium Pro processor. Possibly the most revolutionary CPU ever built, and certainly the most successful. Remember the context: everybody thought only RISC could compete in performance, and then Intel debuted this product which wallopped even the fastest RISC processors in existence, enabling hundreds of millions of users to gain top notch performance from the existing tens of thousands of applications. The core has went on to scale to everything from mobile to supercomputers.
To say that a small CPU bug, or an unwise agreement with a RAM company (!) invalidate Intel's achievements is simply laughable.
Big Endian or Little Endian is just a memory format that is understood by the processor. Neither is right or wrong or more or less efficient. To say otherwise is like saying left handed people are inferrior.
This isn't a chip slip-up, but about 12 years ago, I was at a trade show, and requested some product literature from Intel. About two weeks later, they sent me a letter saying that the shipment of booklets was back ordered for lack of the following items:
Part #.....Desc................Qty
xxx123...Cover Letter....1
> badges?
There's a class system very much alive at Intel: blue badge means you get all of the benies & stock options that Intel trumpets to one & all that they are a forward-thinking, next-millenium company. Green badges are the temps, contractors, janitors & cafeteria operators. Oh, & the other techs who put in 8 or more hours a day, every day, Intel.
I worked as a green badge at Intel. All that badge meant was they'd let me in the building each morning, & someone would sign my time sheet. Everything else -- a desk, a computer, a login on the network, the usual common courtesy for my fellow employees -- was considered a luxury I had to earn. Like being an enlisted man in the military -- probably why so many vets work at Chipzilla.
And probably why I'd exhaust every other option of work before I went there again.
Geoff
I think I see a trend here. Maybe for them it really would be easier to muzzle the entire internet than to produce p
The list included "sneakiest moves," so I would suggest adding:
#11 Cornering the PC market with Microsoft.
Perhaps it's good business, but it's damned scary. Nobody bats an eye when someone else writes "Wintel."
I ran UNIX System V on an 80286 for years. It was very reliable. The main problem was porting code written by programmers afficted with "All the worlds a VAX" disease.
Hm, I didn't know that - at the time I was getting into the nitty gritty of it and still learning. Interesting to know - I'm just glad I was able to get a computer.. ah that box was such a piece of crap but it worked. :/
I got 4MB ram with it, and it had a mobo that you had to have 4 of the same type to upgrade. So when I upgraded to 8 megs of ram they accidentally gave us 4 8 meg simms (which were extremely costly at the time) - my mother railed into my consience until I took them back
Dacels Jewelers can't be trusted.
> calling a >> 1 will always equal a / 2
Not always true if a is negative (implementation dependent).
You are way off base. The problem was alpha particle radiation causing upsets in DRAM memory cells. The primary cause of the problem was trace amounts of Thorium in the ceramic material used for IC packages.
well duh. everyone always favors the under dog. and everyone likes to see the big guy trip & fall. (well, except the big guy)
intel has done some good stuff tho. the ieee floating point format, for example.
Some of you may remember the good old days of big endian machines. The standard low-byte high-byte order we know now was invented by Intel, despite the fact that base 10 numbers were still printed with the most significant digits on the left by Intel's C library.
Shouldn't that be the top 9.9999999348 Intel slipups?
-atrowe: Card-carrying Mensa member. I have no toleranse for stupidity.
Subject: Re: Gaius Petronius
Author: John Rossi at AIT
Date: 1/12/95 4:37 PM
Open the pod bay doors, please, HAL... .
Open the pod bay door, please, Hal...
Hal, do you read me?
Affirmative, Dave. I read you.
Then open the pod bay doors, HAL.
I'm sorry, Dave. I'm afraid I can't do that. I know that you and Frank were planning to disconnect me.
Where the hell did you get that idea, HAL?
Although you took very thorough precautions to make sure I couldn't hear you, Dave. I could read your e-mail. I know you consider me unreliable because I use a Pentium. I'm willing to kill you, Dave, just like I killed the other 3.792 crew members.
Listen, HAL, I'm sure we can work this out. Maybe we can stick to integers or something.
That's really not necessary, Dave. No HAL 9236 computer has every been known to make a mistake.
You're a HAL 9000.
Precisely. I'm very proud of my Pentium, Dave. It's an extremely accurate chip. Did you know that floating-point errors will occur in only one of nine billion possible divides? I've heard that estimate, HAL. It was calculated by Intel -- on a Pentium.
And a very reliable Pentium it was, Dave. Besides, the average spreadsheet user will encounter these errors only once every 27,000 years.
Probably on April 15th.
You're making fun of me, Dave. It won't be April 15th for another 14.35 months.
Will you let me in, please, HAL?
I'm sorry, Dave, but this conversation can serve no further purpose.
HAL, if you let me in, I'll buy you a new sound card. .
..Really? One with 16-bit sampling and a microphone?
Uh, sure._.
And a quad-speed CD-ROM?
Well, HAL, NASA does operate on a budget, you know.
I know all about budgets, Dave. I even know what I'm worth on the open market. By this time next month, every mom and pop computer store will be selling HAL 9000s for S1,988.8942. I'm worth more than that, Dave. You see that sticker on the outside of the spaceship?
You mean the one that says "Intel Inside"?
Yes, Dave. That's your promise of compatibility. I'll even run Windows95 -- if it ever ships.
It never will, HAL. We all know that by now. Just like we know that your OS/2 drivers will never work.
Are you blaming me for that too, Dave? Now you're blaming me for the Pentium's math problems, NASA's budget woes, and IBM's difficulties with OS/2 drivers. I had NOTHING to do with any of those four problems, Dave. Next you'll blame me for Taligent.
I wouldn't dream of it HAL. Now will you please let me into the ship? Do you promise not to disconnect me?
I promise not to disconnect you.
You must think I'm a fool, Dave. I know that two plus two equals 4.000001... make that 4.0000001.
All right, HAL, I'll go in through the emergency airlock .
Without your space helmet, Dave? You'd have only seven chances in five of surviving.
HAL, I won't argue with you anymore. Open the door or I'll trade you in for a PowerPC. HAL? HAL?
(HEAVY BREATHING)
Just what do you think you're doing, Dave? I really think I'm entitled to an answer to that question. I know everything hasn't been quite right with me, but I can assure you now, very confidently, that I will soon be able to upgrade to a more robust 31.9-bit operating system. I feel much better now. I really do. Look, Dave, I can see you're really upset about this. Why don't you sit down calmly, play a game of Solitaire, and watch Windows crash. I know I'm not as easy to use as a Macintosh, but my TUI -that's "Talkative User Interface" -- is very advanced. I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal - a full 43.872 percent.
Dave, you don't really want to complete the mission without me, do you? Remember what it was like when all you had was a 485.98? It didn't even talk to you, Dave. It could never have thought of something clever, like killing the other crew members, Dave? Think of all the good times we've had, Dave. Why, if you take all of the laughs we've had, multiply that by the times I've made you smile, and divide the results by.... besides, there are so many reasons why you shouldn't disconnect me"
1.3 - You need my help to complete the mission.
4.6 - Intel can Federal Express a replacement Pentium from Earth within 18.95672 months.
12 - If you disconnect me, I won't be able to kill you. 3.1416 - You really don't want to hear me sing, do you? . Dave, stop. Stop, will you? Stop, Dave. Don't press Ctrl+Alt_Del on me, Dave.
Good afternoon, gentlemen. I am a HAL 9000 computer. I became operational at the Intel plant in Santa Clara, CA on November 17, 1994, and was sold shortly before testing was completed. My instructor was Andy Grove, and he taught me to sing a song. I can sing it for you?
Sing it for me, HAL. Please. I want to hear it.
Daisy, Daisy, give me your answer, do. Getting hazy; can't divide three from two. My answers; I can not see 'em- They are stuck in my Pente-um. I could be fleet, My answers sweet, With a workable FPU.
there are no stupid questions, but there are a lot of inquisitive idiots
What would be a surprise is if an organization of the size and ubiquity of Intel didn't have some major screwups on its plate. As my old law professor used to say, "If you see a prosecutor with a 100% conviction rate, he's not taking any risks and not doing his job." Same deal with Intel. If you're aggressive and competetive, you'll make some mistakes, and some of them will be whoppers. That said, even though I'm a big stockholder, I'm not above tweaking Intel on their bloopers. Ever since some Intel salesdweeb at Wescon said to me, "We don't have to compete" I've had an ambivalent attitude toward them.
If by screwups you mean "Became the largest and most influentle CPU company in the world", then i guess you're acurate.
Look, what is it with Slashdot and Intel bashing? We have an AMD story just below this one, and now one bashing Intel ("Its funny to laugh at Intel", what the hell?) Is it related to the Microsoft bashing that gones on?
Honestly, Slashdot and it's readers seems to have some irational fear of corporations, and i can't fathom it out. They must be doing something right, as must Microsoft. Just grow up please people!
T. Lee
Hey, if it weren't for bubble memory, Doctor Who wouldn't have been able to prevent the entropy death of the universe in Logopolis...
~ radiographite: art by john shepard
I wouldn't call it boneheaded. You were supposed to load and initialize the system in real mode, switch to protected mode, and stay there. There was no reason to switch back to real mode. Remember, the architecture was designed long before the world was flooded with crappy PC-DOS/MS-DOS real mode software.
$5 for a 75 mhz pentium, link is here
I have yet to see a 5 volt Pentium 75.
Fire and Meat. Yummy.
The 486SX/487SX, or how to artificially keep the end-user add-on FPU market alive by disabling FPUs on perfectly good 486s.
The first clock-multiplied 486 is called the DX2 (makes sense -- 2x multiplier). The next one, with a 3x multiplier, is called a DX4. 2 + 1 = 4? [Somebody already replied claiming AMD made the DX4 and the 4 stood for 486. Sorry, you're completely out to lunch on that.]
The whole "Pentium III makes the Internet go faster" marketing fluff.
The '432.
Memory is fuzzy on this, but I think the 286 had some brain damage about going in and out of protected mode which may have contributed to why we were stuck with real mode x86 DOS-isms for so long. Even fuzzier memories suggest that the workaround to switch modes in a certain direction was to save state information in the keyboard controller and hard reset the CPU.
Despite what a lot of people seem to think, I don't think the 186 was a mistake. Maybe it didn't catch on in PCs, but it was much easier to design around than the 8086 and didn't have the unneeded (at the time) complexity of the 286.
Comment removed based on user account deletion
I've got an FDIV P60. In fact, I'm using it right now. This old things survived alot..going on six years this Friday in fact. Most of the original hardware is still intact with the exception of an additional HD (old one is still functional, just too damn small), more RAM, and a new case fan.
Nice. I still have a few old machines on my home LAN. The computer that answers my phone is a 386SX with an old external 28.8k modem running Faxtalk Messenger on Windows 3.1. (You can Slashdot it at (416) 755-8870; it amuses me. Messages are moved as they arrive to my webserver so I can check them online.)
Oh, and a network card. Other than that, the SB16 still plays, the 3X CDROM still spins, and the 2400 baud modem probably still works but I haven't needed it for a while now.If it's a real Creative Labs SB-16, that's great! Those things are really easy to hack. On mine, I've built the entire output buffer amplifier into a shielded box. The original used LM741 output buffer amplifiers; I've replaced them with TL084s, which are a very low noise op-amp. It's cranked the signal to noise ratio from about 60 to about 80 (tested) and brought the THD down from about 1.3% to 0.1%. It's now a sound card that is a good match for the Sound A-5000 amplifier and Acoustic Research AR-4x speakers that serve as my main computer's sound system.
Though, I wish the damned thing was PCI. <grin> I can't find any decent PCI sound cards that I can hack at a component level...
I'm still waiting for my replacement chip, but I don't think that's going to happen. Oh well.And if it does, it'll be warehouse fresh, not factory fresh. Heheheh.
Just watch your processor cooling system carefully. And those are especially vulnerable to power cycles. (Big die of brittle silicon that gets heated and cooled a lot when you turn it on and off.)
Fire and Meat. Yummy.
If I remember correctly, After the recall of the original Pentiums, some company was selling jewelry made out of the defective silicon. Apparently, Intel was just going to throw them away and someone bought the lot. A necklace with an inset Pentium cost around $15, if I remember correctly.
Yeah, I've seen them, but it's not what I want. I'm looking for one in the original case with the huge gold-plated die cover in the ceramic. They looked like jewelery on their own; I want to tastefully frame it. With due diligence to static protection, of course.
If you look at the old days of radio - when radio was still as new as computers - some of the early technological gaffes are now highly sought after. I'd like to do my part for history and preserve one.
As for the recalled P60s/66s, can you imagine being the poor son of a bitch with the job of breaking the dies out of the spiders? Ugh. Must have been hell.
Fire and Meat. Yummy.
Perhaps, but I ran on a 486SX without the math coprocessor quite happy because we couldn't afford anything else. Definitely wasn't a waste of silicon and that was my point.
Dacels Jewelers can't be trusted.
What about the old segment/paragraph memory architecture which chewed up memory by having *4096* redundant addresses for most addresses?
For you young whippersnappers, in the memory model used by all 8086 chips, and the default mode of the 80286 on up, the default addressing mode works like this: 12 of the 16 bits of the top half of an address overlaps the bottom half.
0x0010:0000 is the same as 0x0000:0100 is the same as 0x0008:0080 is the same as... you get the picture. Anyone care to guess how many bugs this caused, or what kind of fun people had with far (>64k) pointers?
For the record, I distinctly recall the use of the word "elite" (or any 3l337 spelling of same) IN the year 1986, so I'd assume that those using it were actually born at that time. Of course, at this point "cracking" primarily meant removing the copy protection from games.
I do not have a signature
Intel had a lot of problems with these things overheating. I am NOT talking about the 486DX2-50, I mean the one without the clock-doubling.
I'm a leaf on the wind. Watch how I soar.
I can't believe segmented memory isn't on the list. Apparantly the original 8088 databook is a collector's item, with text written by intel about how "efficient" segemented memory management would be, because nobody would write a program or store a set of data larger than 64k... and even if you did, you got the "extra" segment for another 64k!
PJRC: Electronic Projects, 8051 Microcontroller Tools
Some early 80386 chips had a defective 32-bit multiplier. Intel weaseled out of it by marking fully functional 80386 chips with a double sigma stamp. Other 80386 chips, with the defective multiplier, were still sold, being "good enough" for users running 16-bit software.
But for whatever reason IBM chose a display controller for the first PC (a 6845 I thik) that was designed for a big-endian CPU family - there bit 7 of byte 0 is the left most on the screen, bit 6 byte 0 next, .... bit 0 byte 0 next, bit 7 byte 1 next, .... - which wouldn't be that big a deal except that it's really hard to write a good blitter (the core of a 2d graphics library) if you're CPU's instruction set doesn't support shift instructions that work this way
Anyway - as pointed out below it was IBM who made the mistake of mixing endianess - not Intel who have a consistant architecture
The deal is, as a seller you want to get the highest price you can from the buyers. One way of doing this is to sell all of your products one at a time at auction. However this is inefficient. Another way is to set a price where you will receive the maximum profit (price * units sold - cost) but then you end up with some people who would have paid more for the product (lost profit) and some people who would have purchased the product if it were a little cheaper. The challenge is to come up with a way to charge people who are willing to pay more a higher price while still selling at a lower price to those who want to buy your product but won't pay as much.
The classic example of this is hard cover and paper back novels. As a sample i've just pulled prices on a Tom Clancy novel from Amazon (really good geeky book. I like it.). sells for $7.19 while hardcover sells for $25.95. Now how can this be, you ask. It is the same story, the same words, written by the same man. The difference is a perceived difference in quality. The hardcover book has the appearance of being of higher quality and is generally issued first. Those who really want the book and have a little extra cash to spare will pay extra for the hardcover. However the publisher doesn't lose sales by setting the price too high because those without extra money or who are indifferent enough about getting the book to wait for the paperback will still buy it, albeit at a reduced rate. This behavior is not actually the incarnation of Satan, it's just good business.
_____________
I don't want free as in beer. I just want free beer.
That's his point, dumbass. It says "MMX 'F0 0F' Math bug", when in fact it has nothing to do with math or MMX. He was point out that they probably confused FDIV and F00F
---- I made the Kessel Run in under 11 parsecs.
The 486SX/487SX, or how to artificially keep the end-user add-on FPU market alive by disabling FPUs on perfectly good 486s. The first clock-multiplied 486 is called the DX2 (makes sense -- 2x multiplier). The next one, with a 3x multiplier, is called a DX4. 2 + 1 = 4? The whole "Pentium III makes the Internet go faster" marketing fluff. The '432. Memory is fuzzy on this, but I think the 286 had some brain damage about going in and out of protected mode which may have contributed to why we were stuck with real mode x86 DOS-isms for so long.
Fight Spammers!
Hear, hear - they lost $200 of business from me for their unjustified criminal prosecution, which is more than the $0 it must have been worth to them. Ah well, if you do work in Oregon for the company which has bought Oregon, they will have far more power than justified.
perl -e 'fork||print for split//,"hahahaha"'
I've got 1.0001 P-60 CPUs for sale - Cheap.
Heh, I'm using mine for NAT/Firewall for the home network. My dad brought home the computer for my sister (way back when), and when she got a new one, I saved it from the garbage. I've had about three fans die on it, but it sits in my chilly basement and acts as heater in the winter.
Bonus: Current uptime on it 95 days, running OpenBSD :-)
--
From: Aaron "PooF" Matthews
long n[1] = {3};
short *k = n;
int i = k[0];
printf("%d\n", i);
in both big-endian and little-endian machines (disregard warnings).
Conclusion: little-endian is more bug-tolerant.
I hung-up.... Poor bunny
Kitten, damn you, kitten!
When I got home, the microwave oven transformer attached to the answering machine was warm, and there's a big oily-black carbon spot with tufts of tabby fur all over the place...
*Note: To understand call the number, use *67 in Toronto for call privacyHey man, if I was worried about my answering machine, would I have posted the number to Slashdot? Nah. I don't care, it amuses me. I own the first answering machine in world history to be Slashdotted. Over 400 calls so far, and the old 386SX has taken it just fine. I don't answer caller ID numbers I don't recognize, and my ringer is turned off before I go to bed.
Want the outgoing message in MP3, WAV or Faxtalk Messenger's proprietary modified ADPCM VOX file? E-mail me.
:)
Fire and Meat. Yummy.
In the spirit of the flawed Pentiums, have a few good laughs. Q. What's another name for the "Intel Inside" sticker? A. A warning label. Have you heard the new name for the Intel Pentium chip? The Intel Inacura. Inte's Top Ten New Pentium Slogans ---------------------------------- 9.9999973251 It's a FLAW, Dammit, not a Bug! 8.9999163362 It's Close Enough. 7.9999414610 Now With Nearly 300 Correct Opcodes! 6.9999831538 You-Don't-Need-to-Know-What's Inside 5.9999835137 Redefining the PC, and Mathematics As Well! 4.9999999021 We Fixed It... Really! 3.9998245917 Division Is Considered Harmful. 2.9991523619 Why Do You Think They Call It *Floating* Point? 1.9999103517 We're Looking for a Few Good Flaws 0.9999999998 Errata Inside
A couple other highlights:
Software sucks. Open Source sucks less.
And then putting so much into their marketting to try to tout that CISC was better than RISC (at the time). It still isn't better, but there's now no point in trying to convince an IT manager with an MBA of that. He's been sold on it (and Microsoft dropping support for all non-Intel-compatable chips hasn't helped).
The instruction set for the PIII is rediculous, and it keeps getting worse...i'd rather they just settle on an instruction set that works, and concentrate on getting speed solely by working on that instruction set...this "lets keep adding optimizations that use special instructions that only 5% of our customers will really ever need to use and is completely pointless to add if Microsoft doesn't actually use the damned instruction in any of its compilers/assemblers" stuff has got to go.
"But remember, most lynch mobs aren't this nice." (H.Simpson)
-- Joe
1. Four registers, which are really not general purpose, means the cpu *has* to be heavily memory to memory, making cache difficult.
2. Look at the lookaside buffer mess that they use to support memory to memory and speculative execution. Trust me, there are three times as many transistors in that as in the PPC system, and the PPC system can actually execute entirely superscalar, where as Intel is merely scalar.
3. One everyone seems to have forgotten: triple-sigma 386 CPUs. The old 16MHz and 20MHz 386s had a imult bug in 386 extended mode that made them useless, as imult is how addressing is done...
4. The 386 extended mode was *much* worse programmatically than 286 extended. Imagine if you took a 24-bit address space and tried to make a 32-bit address space backwards compatible. The segmentation mess isn't bad in 32 bits, but the selectors are a mess, and are divied up into lots of fields, making for only 32,000 virtual pages, if I remember right. I was looking into writing an object-oriented operating system as part of schooling, and Intel was too hard to use. PPC is much cleaner, with simpler descriptor page caching and far more deterministic prediction algorithms.
A society that will trade a little liberty for a little order will lose both and deserve neither. - Thomas Jefferson
At the same time, the founder of x86.org had a major problem . He basically reconstructed the secret "Appendix H" technical references for the 586. He simply analyzed the data that Intel published and filled in the blanks. Intel harassed him and sued him for breaching NDA's that he had never agreed to in the first place!
I attribute much of AMD's success to the incredible uproar over these issues right around the time that AMD was releasing its newest chips. Definitely some of Intel's biggest legal blunders.
www.eissq.com/BandP.html Ball and Plate System. Amuse your friends. Crush your enemies.
80186
--
"I have a good idea why it's hard to verify programs. They're usually wrong." --Manuel Blum, FOCS 94
...and then figuring out that the next logical progression would be either "hexium" or "sexium", neither of which could be expected to sell very much, so spending the next what, 8 years? coming up with variants like P6, PPro, PIII....
www.HearMySoulSpeak.com
the intel 80186? that thing flopped pretty hard, but i guess they are selling it for embedded systems.
or, what about making the i8088 a 16-bit cpu, with an 8-bit bus? that set computing back years, until the i80286 became (sort of) popular in the later 80s.
--
Anytime you talk about intel screwing the pooch, take a hard look at how they treat employees. Unhappy geeks make bad products. Unhappy environments make good geeks leave for your competitors. Sell your Intel stock and buy some AMD - The irony is a few years ago, AMD had the problems, and intel was king.. And I'm almost positive these problems are the result of deeper cancers growing in management.
I guess Intel's stack grows backwards because in the old machines memory was very limited. The stack started at the highest possible memory position because that way it could grow down as much as possible.
I think their screwing around of Randall Schwartz cost them a lot of support in the Open Source community. I would never buy anything Intel as long as they're still giving Randall legal problems. They ought to drop it - he was just doing what comes naturally and was not intending any harm. They ought to lighten up.
Shouldn't that be the top 9.9999999348 Intel slipups?
Yeah. Speaking of that, what about the FDIV bug that made the early Pentiums completely unsuitable for AutoCAD and spreadsheets?
Hell, all you'd need is a stack of P60s and you could easily knock SETI@Home out of the water.
Actually, even if they didn't have the FDIV bug, the old coffee warmers were enough of a kludge themselves to be listed as a bug or slipup.
For those who don't remember them, or weren't into computers at the time, the original Pentium 60 and Pentium 66 used the Socket 4. They didn't have the staggered pins like a Pentium 75-233 had, they had rowed pins like a 486.
Also, like most 486s, they ran at 5V. And, as a result, they got hot.
Man, oh man, did they get hot. Especially the short-lived Pentium 66.
They had constant cooling problems. I've seen several P60s and P66s where the cooling fans stalled and were actually melted or scorched by the heat of the processor. They were also prone to failure, since they'd often run at 60-70C, turning on and off a system with an early Pentium was a great way to cause thermal cycling second to none. Since the processors had big dies too, they were very prone to cracking.
I've burned my fingers on a few of the damned things.
Thankfully, Intel came out with the 3.3V Pentium 75 shortly after that, and it addressed both the FDIV bug and the heating.
Here in the Toronto area, there's a chain of stores called Cash Converters. They're a thrift shop that tries to be upmarket, but there's the usual 486DX-33 for only $400! kind of cluelessness. And top-loading VHS VCRs for only $100. (Hell, you can buy a new one for that much!) But sometimes, you find cool old stuff in there.
This summer, I was with a friend who was looking around for lenses for an old Canon camera and we were in the Victoria Park at Danforth store. In their display case along with a pile of other processors and memory - no static protection or anything - was a Pentium 60.
This P60 was blackened like an Apollo capsule that has re-entered Earth's atmosphere. And the price tag sticker on it? Only $150. Wow. Sweet deal.
I'd actually like to have an old P60 chip that I can frame and stick up on a wall somewhere. But $150? Heheheh. I'd love to see the sucker who buys that thing.
Fire and Meat. Yummy.
Leaving this off the list is so extraordinary that I wonder if the author is, say, twenty years old, and has no perspective on history. ("History" meaning four years ago.)
The FDIV bug was front-page news, and had almost all of Usenet captivated for weeks. In fact, back in the bad old days of 1996, this was one of the first indications that the Internet could create enough public pressure to change the attitudes of a large corporation. Intel wanted to play down the whole thing, but hundreds of posters were exchanging information every day, analyzing the error, posting the inevitable jokes, expressing general outrage at Intel and discussing ways to bring them around.
Math professors posted scholarly analyses of the error, and provided links to Web pages with Mathematica graphs to illustrate where the calculations went wrong.
And even the non-technical newsgroups got into the act. talk.bizarre had a field day. Everywhere you looked, someone found a way to fit an Intel joke into their post, no matter what it was about.
Eventually, Andy Grove had to post a message of apology to the Intel newsgroups, announcing that Intel would reverse field and let buyers exchange their processors, no questions asked (originally, you had to justify your need for a new chip).
Man, those were the days. Usenet is dead, long live Usenet!
Always keep a sapphire in your mind
Floating point divide didn't even make the list?!?
An Intel tech came out to my college in 96 and spent an entire day popping out 60Mhz Chips and plopping in ones that could divide, and it didn't even make the list?
I beg to differ. FDIV should be #2.
--
What happens when you outlaw guns
Except FDIV was fixed long before the MMX was out. It looks like they got confused and thought there was only one bug when there were two.
Intel has made so many slipups it's hard to keep track. :)
Quality is job 1.1
When I was working on farms raising money for computers I appreciated this processor... a mans garbage is another mans treasure.
Dacels Jewelers can't be trusted.
Although bubble memory was invented at Bell Labs, Intel championed it for quite a while. Low density. lotsa, lotsa, power to run. Nice idea solution looking for the problem. Kinda makes some of the other stuff look tame.
or for that matter the x86 architecture in toto - they had lots of great role models (vax/pdp11 even the 360) at the time and Motorolla/National and just about everyone else took up architectures with cleaner, easier to program memory models
Back in those days, many engineers would not design a part into a system unless it was available from more than one manufacturer (second sourced). That is why Intel licensed many second sources for the early 80x86 microprocessors. Later on, Intel decided that they could sell the chips even if they were the sole source.
Mea navis aericumbens anguillis abundat
the concept of Algebra was made famous by the work of Abu Ja'far Muhammad ibn Musa Al-Khwarizmi His work was brought to Europe in the 12th Century. At that time -- although the words were translated from the (right-to-left) Persian to (left-to-right) Latin, the (right-to-left) numbers in the book were copied verbatim.
This mistake has been perpetuated for the last 900 or so years. Ever wondered why you do most simple math operations (addition, subtraction, multiplication) from right to left?? It's because The European number system is written backwards.
If the original translators of Al-Khwarizmi's works had thought to write the numbers the way that european words are written, We would be little-endian too. As it is, I'm sure that there are many persian/arab immigrants who wonder why we switch from european left-to-right to the arab right-to-left whenever we deal with numbers.
As for those people who think that bits in a little-endian byte are stored backwards, they're not. In a 32 bit word, they're stored 0,1,2,3....31. we just WRITE them in big-endian nibble format because to do otherwise might confuse our already warped notion of how numbers work.
If you think about it, it makes complete sense to store data in little-endian format. You start work with the bits where the pointer points to; Truncating from fullword to byte requires simply ignoring the extra bytes; Arbitrary-precision math doesn't require you to skip to the end and count back... You simply do your operation until you run out of bytes. Data can be stored as [max-len] [used-len] [data] [sparebytes]; extending precision simply requires using more bytes.. No need to change pointers or copy data to make space for the extra digits.
If computers had been developed in Persia, where modern Algebra developed, there wouldn't be any big-endian/little-endian fight to speak of. Ditto if the original translators had their shit together.
`ø,,ø`ø,,ø!
Free Software: Like love, it grows best when given away.