Worst CPUs ever, now with poll!

[SOFTengCOMPelec]

Clones didn't make IBM disappear. Their hand was forced though, so they had to adapt. Same could have happened with Apple since they now make significantly more money on iPads and iPhones than Macs.

It gets complicated, because IBM had a range of stuff that they were selling. At that time, I think it was Mainframe computers, which were their main money earner. A bit like Intel, with its highly coveted server end of the market, dominance.

If IBM had ONLY been selling IBM PCs, then the clones could have had a much more dramatic effect on IBM.

Apple had/has a much narrower market segment (they don't even sell Mainframe computers, never did), so it could have been a lot more dramatic/damaging to Apple.

But you are right, they may have been able to adapt/recover. E.g. By marketing a high quality, high end computer, significantly better than the hypothetical clones we are talking about.

The IBM clones probably hurt Apple, anyway, because it gave the competitors a cheap alternative to lure potential Apple customers away, onto the IBM/clone PC side of things.

 

[gdansk]

Without a doubt the i432. It was an absolute was of transistors. At least hand optimized IA-64 could get damn nice performance. Handwritten i432 was still crippled regardless of the optimizations because of the object and safety overhead. It was the epitome of excessive CISC architectures.

Also, NetBurst is a close second.

 

[ElFenix]

Most of the PCs in the early days (even now, I guess), were made up of IBM clone machines. Because IBM was forced to allow clones, due to anti-monopoly decisions (in 1956 ???), when IBM got badly hit by official agency(s).

IBM wasn't forced to allow clones. the original IBM PC was a screwed together box of commodity parts with the exception of 1 thing: the BIOS. some former TI guys reversed engineered that, built a compatible version, and then put together their own boxes of commodity parts (which created compaq).

 

[SOFTengCOMPelec]

IBM wasn't forced to allow clones. the original IBM PC was a screwed together box of commodity parts with the exception of 1 thing: the BIOS. some former TI guys reversed engineered that, built a compatible version, and then put together their own boxes of commodity parts (which created compaq).

IBM would have (in my opinion) liked to have made as much as possible, of the IBM PC, propriety. Which would allow them to keep a decent market share, and/or get nice licencee revenues.

But ..

1956: Consent decree
The United States Justice Department enters a consent decree against IBM in 1956 to prevent the company from becoming a monopoly in the market for punched-card tabulating and, later, electronic data-processing machines. The decree requires IBM to sell its computers as well as lease them and to service and sell parts for computers that IBM no longer owned

introduced rules, which meant that they created/sold the IBM PC, in the way they did.

If that judgement had NOT existed, IBM would have been free, to make it considerably more proprietary.

The BIOS was what little they were able to make proprietary, given the 1956 stuff.

Quoted source

Referred source on judgement, used by article

DON'T necessarily think that I'm the source of this post (I am quoting the details from memory), which comes from an apparently reliable book, I have read, quite some time ago.

 

[ElFenix]

IBM would have (in my opinion) liked to have made as much as possible, of the IBM PC, propriety. Which would allow them to keep a decent market share, and/or get nice licencee revenues.

But ..



introduced rules, which meant that they created/sold the IBM PC, in the way they did.

If that judgement had NOT existed, IBM would have been free, to make it considerably more proprietary.

The BIOS was what little they were able to make proprietary, given the 1956 stuff.

Quoted source

Referred source on judgement, used by article

DON'T necessarily think that I'm the source of this post (I am quoting the details from memory), which comes from an apparently reliable book, I have read, quite some time ago.

if the consent decree really only required that IBM had to sell, not just lease, computers, and it had to provide service and parts for the computers, then IBM didn't need to use commodity parts to accomplish that. it could have been full of proprietary parts just like pretty much every other computer IBM built in its history.

 

[SOFTengCOMPelec]

if the consent decree really only required that IBM had to sell, not just lease, computers, and it had to provide service and parts for the computers, then IBM didn't need to use commodity parts to accomplish that. it could have been full of proprietary parts just like pretty much every other computer IBM built in its history.

tl;dr
They had to allow other manufactures to be able to make the individual parts (hard disks etc). I.e. Public domain interfaces.
I'm NOT sure how far it went, but I think it prevented them from using proprietary parts to make the PC.

The above is my interpretation, and could be WRONG!

The following link, is a very detailed explanation of the judgement.

http://www.cptech.org/at/ibm/ibm1956cd.html

 

[Hi-Fi Man]

Why is the PowerPC 970 on the list??? Sure it couldn't be used in laptops due to energy inefficiency but it still was a decent chip (especially the 90nm 970x shrinks), better than presscott that's for sure.

If any PowerPC CPU should be on there it should have been the whole 74xx line. The worst memory interface ever and those chips had a hard time scaling up in frequency. No amount of L3 cache could've saved a 7455 from being tied down to it's 100MHz/133MHz/166MHz SDR FSB. Apple tried it's hardest to conceal the weaknesses of it all but Motorola just gave up in the end leaving IBM to pick up the slack with what was a good chip just too workstation/server oriented to be put in a laptop, typical IBM design really.

 

[SOFTengCOMPelec]

if the consent decree really only required that IBM had to sell, not just lease, computers, and it had to provide service and parts for the computers, then IBM didn't need to use commodity parts to accomplish that. it could have been full of proprietary parts just like pretty much every other computer IBM built in its history.

I know I just answered this, but I thought I would give you the exact line from the link. There could be other relevant parts to it.

(c) IBM is hereby enjoined and restrained from requiring any purchaser of an IBM tabulating or electronic data processing machine to have it repaired or maintained by IBM or to purchase parts and subassemblies from IBM.

 

[Blue_Max]

I seem to recall that IBM chose the 8088 because of how cheap and common it was... "found in many soda machines" was a quote that stuck in my head.

 

[SOFTengCOMPelec]

I seem to recall that IBM chose the 8088 because of how cheap and common it was... "found in many soda machines" was a quote that stuck in my head.

I think they wanted it to be ready, in very tight timescales, and somewhat lowish cost to produce, so the 8088 (with 8 bit externals, rather than 16 bit), made sense to them.
I think they only expected to sell a relatively small (to IBM) number of these "micro computers".
(Speculation and/or slight recollection on my part in this sentence) They may have been worried about losing sales/profit on their other expensive computers (such as Mainframes), if the IBM PC was TOO good.

If I had a time machine, I would try and talk them into using the Motorola 68000 and NOT using Bill Gates Microsoft. Instead go for D.E.C.'s new OS. Which I have heard rumors that Microsoft purchased the original DOS (early pre release version), which was itself an unauthorized copy (rewritten though, to avoid copyright issues or something) of what D.E.C. was about to release at some point (maybe).

 

[soccerballtux]

definitely the original Phenom. Too power hungry, didn't clock as high as desired, and a year or two late.

 

[ElFenix]

I know I just answered this, but I thought I would give you the exact line from the link. There could be other relevant parts to it.

No, that means people could buy used parts or second source parts if they were available - that doesn't require IBM to use commodity parts (and again, IBM mainframes used proprietary parts - PC was an outlier).

Where the 1956 consent may come into play was a documentation requirement. IBM was required to provide documentation (that it created for its own repair and technical reasons) to others at a reasonable cost. The Compaq guys used IBM's PC documentation when they reverse engineered the BIOS.

 

[SOFTengCOMPelec]

definitely the original Phenom. Too power hungry, didn't clock as high as desired, and a year or two late.

I had one of the first (earliest ones physically available to the general public). Which had the TLB bug, and had to run 10% slower, as a result.
Getting an AM2(+) motherboard, was a mini-nightmare as well.

 

[Hi-Fi Man]

I had one of the first (earliest ones physically available to the general public). Which had the TLB bug, and had to run 10% slower, as a result.
Getting an AM2(+) motherboard, was a mini-nightmare as well.

That whole TLB bug was just a major fiasco for AMD. Salt in the wounds after conroe.

 

[Gikaseixas]

P4 and cacheless Celerons

 

[SOFTengCOMPelec]

No, that means people could buy used parts or second source parts if they were available - that doesn't require IBM to use commodity parts (and again, IBM mainframes used proprietary parts - PC was an outlier).

Where the 1956 consent may come into play was a documentation requirement. IBM was required to provide documentation (that it created for its own repair and technical reasons) to others at a reasonable cost. The Compaq guys used IBM's PC documentation when they reverse engineered the BIOS.

I am glad that we are in apparent agreement, about the documentation. (Because of the 1956 stuff).

I'm not knowledgeable enough about the 1956 stuff and its impact on IBM, to usefully agree or disagree with you, as regards the use of commodity parts.

So you could well be right about that.

I'm not good with legal documents, and it seems to be a very long and complicated legal document. (1956 stuff).

I was led to believe that IBM were forced into making the IBM PC clones, relatively possible. But that information source (a book), could have been wrong/mistaken and/or I may be mis-remembering the finer details.

Repeat of link to 1956 stuff

 

[soccerballtux]

That whole TLB bug was just a major fiasco for AMD. Salt in the wounds after conroe.

it didn't affect anybody but a subset of server operators and was patchable with recompiled software. ?

 

[SOFTengCOMPelec]

it didn't affect anybody but a subset of server operators and was patchable with recompiled software. ?

My home/gaming computer was affected. If I remember correctly, the bios setting disabled the TLB (as a fix), which slowed it down by around 10%.

It took the shine off my excitement, at getting one of the new range of Phenoms.
I had read a lot about it (I think), and was looking forward to it.

But they (AMD) fixed it in later chips, so it was only an initial blip, for early adopters, such as myself.

 

[ElFenix]

I am glad that we are in apparent agreement, about the documentation. (Because of the 1956 stuff).

I'm not knowledgeable enough about the 1956 stuff and its impact on IBM, to usefully agree or disagree with you, as regards the use of commodity parts.

So you could well be right about that.

I'm not good with legal documents, and it seems to be a very long and complicated legal document. (1956 stuff).

I was led to believe that IBM were forced into making the IBM PC clones, relatively possible. But that information source (a book), could have been wrong/mistaken and/or I may be mis-remembering the finer details.

Repeat of link to 1956 stuff

I read most of the 1956 decree. If it had anything to do with the ability to make clones it was due to documentation being available. IBM used commodity parts because it wanted to get into that market (which it had been missing from) in a hurry.

Even then I'm not sure what role the 1956 decree had - the Apple II was legally reverse-engineered as the Laser 128 computer (though it wasn't fully compatible).

 

[Hi-Fi Man]

it didn't affect anybody but a subset of server operators and was patchable with recompiled software. ?

I seem to recall it reducing performance significantly in some scenarios (not just servers). It obviously was a big enough issue that AMD quickly tried to replace the original bugged chips with fixed xx50 series chips.

This article here http://www.anandtech.com/show/2477/3 shows it effected more than just server type loads (especially winRAR) contrary to what AMD stated in a quote on the first page.

​

 

[SOFTengCOMPelec]

I read most of the 1956 decree. If it had anything to do with the ability to make clones it was due to documentation being available. IBM used commodity parts because it wanted to get into that market (which it had been missing from) in a hurry.

Even then I'm not sure what role the 1956 decree had - the Apple II was legally reverse-engineered as the Laser 128 computer (though it wasn't fully compatible).

That makes a lot of sense.
They did want to get onto the market very quickly, because I read a lot about that, when I looked into how Bill Gates (Microsoft), managed to worm their way into the Dos/Windows Monopoly situation.
Apparently DEC was suppose to get the OS contract order.

I get very suspicious that something "funny" happened, because apparently Bill Gates mother, was an important director at IBM, at the time he managed to get the OS (Dos) contract.

Ironically, (this is speculation on my part), if IBM had succeeded in 100% blocking out all clone options. Then I suspect that the clone manufacturers would simply come up with their own version of a PC, which is NOT compatible with the IBM PC.
A bit like the Windows vs Linux situation we have at the moment.
But in time, the "alternative PC", would probably still become king of the PC world, because of its potentially much lower price.

I had a search for the book, I was referring to earlier, but I can't find it (quickly).

 

[AnandThenMan]

definitely the original Phenom. Too power hungry, didn't clock as high as desired, and a year or two late.

Not even close to the worst CPU. I run one currently as a LAMP server it works great churns through request very well and never crashes.

 

[DrMrLordX]

Worst CPU evar might be any Cyrix 5x86. Some of those things killed themselves @ stock settings due to overheating. Epic fail.

I voted Itanium though, since the Cyrix option just wasn't there and because Itanium represents not just the failure of Merced but the failure of an entire ISA. Surprised anyone is hating on the k5 which was quite fast per clock compared to Pentiums. Their major downside was their limited clockspeed.

honorable mention for worst CPU ever: any Pentium with the fdiv bug. Not that I'm the first to mention that.

 

[Blue_Max]

Worst CPU evar might be any Cyrix 5x86. Some of those things killed themselves @ stock settings due to overheating. Epic fail.

honorable mention for worst CPU ever: any Pentium with the fdiv bug. Not that I'm the first to mention that.

Two real good ones there! :thumbsup: I owned that Cyrix... (or was it the 6x86 after it?) and I'm very glad I live in cold Canada! I rarely had any heat issues thanks to that... if it was a hot, stuffy room it would've cooked!

The Pentium 60 & 66 (with the bug and all) ran circles around my 486DX4/120 and I was so jealous...

 

[Maximilian]

Wheres the pentium 4 wilamette? Wheres the original cacheless celeron? Wheres bulldozer? Phenom I? Atom?

They're all various levels of fail but not in the poll :|