AMD to crap on customers, Q1 06

[Tannin]

You are grossly incorrect.

We (and countless others like us) shipped a massive volume of Cyrix-equipped systems, and with perfect satisfaction. Through the 100 to 200MHz era - which was a long, long time - we sold more Cyrix-based systems than all other manufacturers combined, and back in those days, remember, I was not semi-retired as I am now: we sold enough systems to keep eight people gainfully employed and paid on time. That is a lot of chips anyway you slice it. Yup, we had dud ones. We also had dud CPUs from Intel and from AMD and from IDT. No particular pattern to the failures, except the usual one that applies to all CPUs as compared to other parts (like mainboards, PSUs, and so on), i.e., they failed a lot less often than the average component.

You are just plain wrong about Cyrix. If you couldn't make them work properly, then you just weren't doing it right. Simple as that. We put hundreds upon hundreds into service. Some of them are still in service. Hell, this very post is brought to you courtesy of the Cyrix 6x86-200 Classic that has been running my Smoothwall 24 hours a day, 7 days a week for ... I'm not sure for how long, two or three years, certainly.

 

[time]

Mercutio touched a small, lonely, nearly forgotten nerve with me when he viciously savaged poor, defenceless Cyrix CPUs.

Below PR233, Cyrix chips were head and shoulders above Intel equivalents - except for floating point operations, where they unequivocally sucked.

Part of the substantial performance improvement was down to their higher FSB (i.e. better memory bandwidth), although I suspect their far bigger L1 cache had something to do with it.

Unfortunately, Intel absolutely ruled the motherboard chipset roost in those days, and Intel certified their chips to just 66MHz FSB, whereas the most desirable Cyrix CPUs demanded 75 or even 83MHz.

Herein lies the problem: only a very few motherboards were designed to run above 66MHz, and almost none at 83MHz. In fact, I don't believe any Intel chipset-based motherboards offered safe operation at 83MHz (including THG's claims about certain Asus motherboards) - over 40MHz (FSB/2), the ATA bus could be relied on to introduce corruption.

This is the most likely explanation for Mercutio's observations. And in those days, no-one who was anyone would even consider a non-Intel motherboard solution.

 

[time]

I also have a complaint with Tannin's account: the Intel HX chipset was an excellent platform for Cyrix CPUs - in fact, the best, provided you had a motherboard that supported 75MHz operation.

Boards that fulfilled this criteria included ones from MSI (the best) and DFI (really quite remarkable sophistication for the time, including automatic powerdown on CPU overheat!).

Later on, as Intel inflicted TX on an unsuspecting, compliant world, Via chipsets rose above their feeble Intel counterparts and swept them away - for those who notice these things.

We actually ran a web server, including Cold Fusion and at one stage even Oracle, on a DFI HX motherboard with a Cyrix/IBM 150. It was fast and it was flawless.

Of course, as Tannin says, Cyrix progressively lost credibility above PR233.

 

[Mercutio]

If I make the not unreasonable assumption that the multiple different white-box vendors did in fact have at least a partial clue, I'd expect that at least one or two of them would build a "proper-according to Tannin" Cyrix system, that worked as well as AMD and Intel or WinChip or NexGen systems of the day.

I've saw no evidence of that. I've never seen one.

So *I* worked on them - and I know I'm a good tech, and I'd been doing it for a while even back then. Some other white-box tech guys built them them, back in the days before hobbyist computer techs (ie people whose idea of technical information is a "graphics card shootout" at THG) were common. Yet the builders were shipping machines that couldn't be made to work properly for any length of time, until the processor was changed.

Given that experience, there is nothing you could say that would convince me I am anything other than absolutely correct. I saw systems that didn't work. I would fiddle with those PCs, they would remain borked. I'd change the CPU and magically they would work again.

 

[time]

If I make the not unreasonable assumption that the multiple different white-box vendors did in fact have at least a partial clue ...

That really is an unreasonable assumption!

Look, the damn things not only worked, they worked well, at least with suitable motherboards (eg Shuttle VX). They needed extra cooling, but not that much when compared to Socket A, for instance.

Bluntly, I've had more problems with Intel CPUs of that vintage in terms of overheating or instability. Ah yes, and AMD CPUs, specifically K6-233, which was the most diabolical CPU I've ever seen.

 

[Mercutio]

So somewhere in the world, probably next to the candy cane trees and gumdrop mountains, there were motherboards that enabled the magical Work Like A CPU Is Supposed To feature?

 

[Tannin]

As you say, Time, HX boards worked just fine with Cyrix CPUs. The problem with the combination was performance: although HX boards were the best thing you could buy for a Pentium, and worked well with K5 or K6 too, the several different ones we benchmarked with Cyrix chips were all very slow. I have no idea why, but the measurements were unmistakable. (I used to waste a ridiculous amount of time doing formal benchmarking in those days. I have no idea why.)

As you suggest, the 66MHz limit on the Intel chipsets of the day was significant. Having built more Cyrix-based systems than the average screwdriver jockey, and having tried a lot of different combinations, I never did discover a VX board that was really trustworthy at 75MHz. Some of the HX and TX boards were OK at that speed, but even they were measurably slower with a Cyrix CPU than the better-suited chipsets (VIA and, depending on implementation, SiS).

For the 66MHz Cyrix chips, you could use an Intel chipset if you liked, but there was no point in doing so: it would be a little slower, and usually cost more than a VIA or SiS board too.

Far and away the best chipsets for Cyrix CPUs were the VIA ones. Boards like the superb old FIC PA-2005 were easily the fastest Cyrix boards, were certified rock-solid at 75MHz, and gave us a wonderful run for a long, long time. There was a Gigabyte SiS based board which was good too, which I could look up but can't remember the name of ... no, it's come back to me: the 586S. The later 586S2 was, for some reason, nothing like as good.

Notice, by the way, that although the PA-2005 was clearly the fastest board for (e.g.) a 6x86-200 Classic, it was not a performance leader with (e.g.) a Pentium MMX. For Intel CPUs, use an Intel chipset: HX was best. Failing that, a well executed SiS board like the 586S was OK too. Avoid VIA: it would work OK, but be sluggish. For Cyrix, the exact opposite. And for AMD, use whatever you liked: it didn't make a lot of difference.

Merc, the fact that you "have never seen one" is not evidence. I have seen hundreds of good, solid, well-performed Cyrix-based systems, some of which are still in service today. (How many systems a week did we sell in those days? Lots! Not to mention systems made by our competitors, which varied, as you might expect, from excellent to complete crap, depending on which competitor. You can probably make that "hundreds" of mine above "thousands".) I know that I'll never convince you: your mind is utterly closed on this matter and no amount of evidence will have the slightest effect. Nevertheless, for the record, I'll continue to refute your unfounded and quite often ridiculous assertions on this matter, which is clearly one you haven't got the slightest idea about. You really should stick to posting on topics that you understand.

 

[Tannin]

So somewhere in the world, probably next to the candy cane trees and gumdrop mountains, there were motherboards that enabled the magical Work Like A CPU Is Supposed To feature?

FIC PA-2005 (VIA, 66 or 75MHz)
FIC VA-502 (VIA, 66 or 75MHz)
Gigabyte 586S (SiS, 66 or 75 MHz)
Chaintech 5VGM1 (VX, 66MHz)
ASUS VX-97 (VX, 66MHz)
ASUS SP-97V (SiS, 66MHz)

Many others, but these were the ones we used the most of and liked the best.

 

[Tannin]

I agree, the K6-233 was a tricky damn bastard to get working right: it was a "factory overclock" in all but name.

The early K5 parts were curly ones too, as were the very first 6x86 Classic parts, and the first Pentiums. Of those three, the Pentium 60/66 was the worst, and the K5 probably the best. Note, however, that in all three cases the production guys improved things quite rapidly (well, they did with the K5 and the 6x86, after the first few horror show Pentium 60 jobs, we steered clear of them until the excellent 90 and 100MHz Intel chips arrrived, so I'm only assuming that they got better - we might have tried a bit harder if they weren't so incredibly expensive).

Other chips we had a lot of trouble with (we already mentioned the K6-233):

Pentium 200 Classic (another factory overclock, and not very fast anyway)
Cyrix 6x86MX-266 (that 83MHz bus was just too hard)
IDT C6-225 (nothing like the fuss-free 200 version)
AMD K6-2/550 (a complete dog unless you underclocked it)

Various others, but they were the ones we hated the most.

 

[time]

You really should stick to posting on topics that you understand.

Ouch!

True enough, though. <runs and hides>

Tannin, the MSI HX boards were rip-snorters that, AFAIK, eclipsed everything else regardless of CPU. Cheap, however, they were not. Neither were the DFI that I remember. In fact, "bloody expensive" would be a better adjective.

I think the discrepancy can be explained by different target markets; I was wasting time with exotica while you were selling millions to the mass market. In this era of CPU-integrated northbridges, you just don't see that kind of variation anymore.

 

[Tannin]

I don't know that I ever met an MSI board until we were selling Durons and P-IIIs, Time. No, wait, we used to get 386 boards from Microstar - which I think is MSI. Possibly 486 boards as well, I'm not sure.

I was wasting time too, probably more so than you. Yes I was selling millions to the mass market, but I didn't make millions: it all went on the wages bill and associated overheads. Later on, I halved my turnover and doubled my income. Later still, I halved the turnover again, more-or-less halved my income, and tripled my time off, which made life 101 times better. If only I'd had the sense to do it ten years earlier!

 

[tazwegion]

So *I* worked on them - and I know I'm a good tech, and I'd been doing it for a while even back then. Some other white-box tech guys built them them, back in the days before hobbyist computer techs (ie people whose idea of technical information is a "graphics card shootout" at THG) were common. Yet the builders were shipping machines that couldn't be made to work properly for any length of time, until the processor was changed.

I'll assume that 'curve ball' was for my benefit... :roll: there's no need to get 'narcy' because others have had better success with a processor which you've been biased against from the outset, I suppose it's like a GMH Vs. Ford thing eh? :lol:

FYI Tom's was used as a mere example (for that particular GPU post) for two reasons... one is was a useful snapshot of that period (and a quick google result), and secondly to assist the manifestation of pessimists (like guess who? ), oh and it is a well know fact that THG's reviews can become biased even over @ the 'PC modder' free-for-all style forums

All reviews have bias, it's a well known fact... we're human after all some are just less obvious than others for example during Skt. A's peak nVidia (whom you love to hate) had the best AMD chipset solution namely the nForce2 Ultra but depending on which reviews/forums you perused... the Asus, DFI, Abit was 'king of the hill' like I said human nature, we like different things for different reasons.

BTW as far as hobbyist computer tech slag title goes, I actually achieved an Advanced Certificate in Information Technology back in '89, and have been dabbling ever since

Mmmmkay? :eekers:

 

[Mercutio]

You really should stick to posting on topics that you understand.

I understand how to build a decent Socket 7 system just fine, Tannin. I probably built a few hundred on my own. Which is not bad for someone who never worked in retail. Certainly enough to be qualified enough to know shit when I smell it.

I saw Cyrix chips mounted on Asus, Shuttle, Biostar, PC Chips and Octek boards most often, Epox or FIC if I was lucky. Lots of SiS and ALi chipsets. Low end products all, but even then, I could switch a K6 or K6-2 for a 5/6x86 and see a drastic change in how well that PC would operate.

Do that a few dozen times and a clear pattern emerges.

 

[time]

Tazwegion,

Mercutio is a qualified instructor for several IT certifications. He has a copious memory and a formidable intellect.

He's also sometimes spectacularly wrong, but I can't recall him ever admitting that.

I *know* the Cyrix CPU could be a viable platform. So did IBM - AFAIK they manufactured them and sold a similar amount under their own brand.

As I sit here, contemplating a mix of nVidia and ATI graphics cards, I wonder who is right - Tannin or Mercutio? :-?

 

[Handruin]

I had little to no problems with a Cyrix PR166 mounted on an Abit SM5-A (VX chipset) many moons ago. Heat was never an issue that I can remember. It may still be kicking around, I'll have to check with my parents. They used it for a number of years until I gave them my PII 350, which is still being used.

 

[tazwegion]

I saw Cyrix chips mounted on Asus, Shuttle, Biostar, PC Chips and Octek boards most often, Epox or FIC if I was lucky. Lots of SiS and ALi chipsets. Low end products all, but even then, I could switch a K6 or K6-2 for a 5/6x86 and see a drastic change in how well that PC would operate.

Do that a few dozen times and a clear pattern emerges.

Yeah, you really should stop working with low end components!

Seriously though... it's not fair to compare a Cyrix MX-200 to a K62, they were Pentium Pro killers yes, but as Tannin's already conceded that Cyrix progressively lost credibility above PR233, let's keep comparisons @ around spec

Tannin I'm truly saddened to hear you call the K62-550 "a dog" mine is still running 'fuss free' on an Aopen AX59pro (VIA Apollo chipset) crunch WU's for F@H & SETI

 

[Mercutio]

Taz, my mention of graphics cards at THG had nothing to do with you but rather my objection to the people who get horny because they think Blue LEDs inside a computer make it run faster. Or whatever. I see a lot of that type in my "day job" and usually a couple hours a day convincing them otherwise through my classes.

 

[Mercutio]

Yeah, you really should stop working with low end components!

Wasn't my choice. I've been a general-purpose IT Contractor for years. From 1998 to 2002 I supported myself primarily by doing service work for small and mid-size businesses, between contracts to very large corporations (prior to that I did tech work, but as a high school or college student I can't say it was a full-time job). A Fortune-500 company would have IBM, Compaq or Dell managed systems. Sometimes those were great, sometimes not.

But then I'd walk into a business with no standard equipment. White Boxes with PC Chips boards. Presarios. Emachines. And I had to make that stuff work, too. That sometimes meant yanking a Cyrix chip for something that actually works.

 

[P5-133XL]

My main complaint about Cyrix was not if they worked or didn't but rather the extrodinary number of unscrupulous people (in my area of the country) that would sell a machine with a Cyrix processor as a machine with an Intel processor with the customer totally unknowing. These budget machines would be touted as high-end machines and sold at just a slight discount from a true high-end machine. I will not try not to mention the names and number of current white-box machines that use the same marketing strategy: Lets face facts that strategy still works today

P.S. My experiance with Cyrix processors was not good. I generally found machines containing these budget processors to be total junk. I can't blame the processors for that though: If one component of a machine is chosen as a "Budget" component, then generally all the components were picked for the same identical reason and the reliability of the system was then highly suspect. Put in a budget processor, then the MB would be a low-end budget product, as well as the hard drive and the power supply and every other component...

 

[i]

Forgive me for pointing this out but the USA is part of the continent known as North America. And Australia is part of ... well, it's its own continent.

So, umm, what do we have here ... one plus one, carry the zero ... umm ... two. Yes, that's right. Two continents. Two different continents. Separated by the largest ocean this planet currently has to offer.

I'm sure someone here can tell me I'm wrong but ... maybe Cyrix had more than one facility for chip fabrication. Wouldn't that have led to all kinds of interesting product distribution choices! Cyrix could have sent chips from one plant to one region, and chips from another plant to another region. I wonder what kind of effects that might have had on regional perceptions of quality. Hmm.

Or maybe that's ridiculous. Maybe no CPU manufacturer in the 1990's - let alone Cyrix - had more than one manufacturing plant.

Well then maybe they had various quality control groups within that one plant, and some of the quality control groups sucked, and some didn't. And North America usually got the products passed by sucky groups.

Or maybe all the quality control teams were really on the ball, and were able to divert the CPUs with extra-high quality margins to wherever HQ wanted. Maybe Cyrix saw that their greatest demand for the cheapest CPUs was in the USA, so they shipped all the marginal performers there. And maybe they saw that Australians were used to having to pay an arm and a leg (comparatively speaking), so they sent the better ones their and simply charged a slightly higher price. Or maybe they just hated America, and intentionally shipped all of their most marginal products there.

I don't know.

But my point is, it seems to me that there could have been a lot of stuff going on with Cyrix ... given that we're talking about 2 different continents here ... that no one here has a clue about. Unless one of you actually worked in a senior management position at Cyrix during those years of course.

So until someone comes along with the whole story about Cyrix, can I suggest that everyone here might be correct, at least with respect to what they, personally, were seeing?

Thanks. Have a nice day.

 

[tazwegion]

Taz, my mention of graphics cards at THG had nothing to do with you but rather my objection to the people who get horny because they think Blue LEDs inside a computer make it run faster. Or whatever. I see a lot of that type in my "day job" and usually a couple hours a day convincing them otherwise through my classes.

OMG everyone who's anyone knows that it's the RED LED's that go cause a computer to go faster :roll:

White Boxes with PC Chips boards. Presarios. Emachines. And I had to make that stuff work, too.

That sounds like a form of physical abuse dude, I'm so sorry they put you through that ordeal

I totally agree with you P5-133XL 100% if a low-end (read as skimped) system is built all components suffer equally, hence Tannin's systems (during the MX-200 period) were built with 'tried and tested' methodology & quality FIC mainboards, though admittedly I purchased @ the end of the 'Cyrix peak' and opted for a K6 / PA-2007 solution (as I was an avid Gamer @ the time )

Vomit boxes have always sucked because they're built with one performance factor in mind, lowest possible price

An interesting theory there i and a skillful piece of fence sitting to boot! as far as 2+ manufacturing plants, the MX's I located @ CPU-collection would indicate that to be true Canada & USA, however both my 're-badged' (IBM) MX's were USA stamped, it is also reported that during the Cyrix DX2 period they had a plant in Japan, though I don't know when they ceased production of Cyrix processors :eekers:

 

[LiamC]

IIRC, Cyrix was a fabless company. All Cyrix processors were fabbed by IBM. As part of the deal, IBM took x% of the chips labelled them IBM (instead of Cyrix).

Again IIRC, IBM avoided the 83MHz bus chips and opted for another 0.5 multiplier @75MHz to get the same PR rating--> Cyrix would offer 2 x 83 = 166MHz PR200 whilst IBM would offer 2.5 x 75 = 187MHz PR 200.

Again IIRC, the 6x86 ran very hot but the later 6x86L ran rather cool and was the chip to use. The 6x86MX was an L with MMX extensions. But as the (absolute) frequency started climbing again, the L/MX started to run hot again (266 etc.).

Don't ask me who fabbed the chips when Nat Semi bough Cyrix though--but it's probably irrelevant as Cyrix were on the downward slide by then--and Nat Semi's process tech was nowhere near Intel/AMD/IBM's

0.05 (we abandoned 1¢ and 2¢ coins long ago)

 

[Splash]

...but rather the extrodinary number of unscrupulous people (in my area of the country) that would sell a machine with a Cyrix processor as a machine with an Intel processor with the customer totally unknowing.

Yes, this was a *hot* topic everywhere (disguised Cyrix boxes) during the Pentium Classic / "586" era. These unscrupulous jackasses would build a computer system with a cheap 486 mobo, install a Cyrix 5X86 processor in it, and call the system a "586" computer system -- fooling the fools into thinking that this was one better than a 486 (or a Pentium). I can't recall, but didn't a "100 MHz" 5X86 really have a 66MHz or a 75 MHz processor clock?

IIRC, Cyrix was a fabless company.

Correct – fab-less. When Cyrix was in business, I drove by their offices just up the road here a few times. There office building was nothing impressive. It was the only property they owned -- offices full of engineers, marketing types, and executives. Probably even had certain people working out of their houses. :lol:

All Cyrix processors were fabbed by IBM. As part of the deal, IBM took x% of the chips labelled them IBM (instead of Cyrix).

Correct. IBM's processors were referred to -- at least for a while -- as IBM "Blue Flame" processors.

Again IIRC, the 6x86 ran very hot but the later 6x86L ran rather cool and was the chip to use.

Correct. I recall the whole Cyrix processor debacle as being essentially:

Procs started off sort of alright in the early months; procs got slowly faster and hotter over the next several month but also got less and less expensive putting pressure on AMD and Intel; IBM stepped in and began manufacturing Cyrix procs along with National Semi; as time went by the Cyrix procs ran hotter with clock freq increases; now I'm a bit fuzzy on what happened next, but, Cyrix was loosing money big-time tried to sell the whole mess (+debt) to IBM but IBM didn't buy, however National Semi did buy Cyrix patents and technical data but never did manufacture any of their own 6X86 procs because National Semi themselves were now in financial straits!

Don't ask me who fabbed the chips when Nat Semi bough Cyrix though--but it's probably irrelevant as Cyrix were on the downward slide by then--and Nat Semi's process tech was nowhere near Intel/AMD/IBM's

I believe IBM had already ceased 6X86 and other Cyrix-related production by this point in time. There was a lot of excess new/old stock 6X86 processors in warehouses for years after that, though. Towards the end, I sort of recall seeing 6X86 processors selling for something like US$15 each.

As crappy as Cyrix processors were throughout most of their existence, much of the basic X86 microprocessor technology they invented or refined later found its way into AMD K6 technologies which have, of course, evolved greatly over time into Athlon / Opteron technology. Cyrix wasn't ALL bad!

 

[Mercutio]

Cyrix wasn't ALL bad

Actually that is the crux of our argument here. I believe I've staked out the zero good whatsoever position.

Oh, I'm sure there's SOME good. Like the amount of good in a Republican Catholic Priest NAMBLA member playing Warm Oil and Astroglide Twister with a Half Dozen Altarboys and a randy goat.

But most people would say that's not much good at all.

I talked to a Cyrix sales rep once. I don't remember why, since I distinctly recall being very vocal in my displeasure about Cyrix products. He gave me a coffee mug and a t-shirt even after his ass-chewing. I keep screws in the coffee mug and I wipe up cat vomit with the shirt. That is the extent of my use for Cyrix.

As I recall, Via bought both Cyrix and IDT (who made the WinChip). Cyrix is the name on Via processors, but the core tech is all WinChip. In the end, modern Cyrix is just a marketing term.

As I recall AMD bought NexGen. The name on current AMD chips is AMD, but from the K7 on, the engineering has been NexGen.

 

[Dïscfärm]

Cyrix wasn't ALL bad

Actually that is the crux of our argument here. I believe I've staked out the zero good whatsoever position.

I loathed (and still loathe) Cyrix. After my early encounters of Cyrix 5X85 activities back in the day, I made sure to keep my distance from anything-Cyrix and advise others to do the same.

Still, I will make a *rare* admission that Cyrix did in fact introduce microprocessor innovations that have been picked up by various CISC processor designers along the way, such as introducing VLIW processing and a primordial form of "NetBurst" architecture. Unfortunately, Cyrix was just a bit too far ahead of their time. They should have had IBM helping out earlier on.

Otherwise, include me in the "Cyrix Is CRAP" camp.

As I recall, Via bought both Cyrix and IDT (who made the WinChip). Cyrix is the name on Via processors, but the core tech is all WinChip. In the end, modern Cyrix is just a marketing term.

As I recall AMD bought NexGen. The name on current AMD chips is AMD, but from the K7 on, the engineering has been NexGen.

OK, now that makes sense. I knew National Semi dumped Cyrix off to somebody, I wasn't sure who, though. I *did* know about Via buying IDT.

NexGen was around even before Cyrix. However, after 2 or 3 years of production, design, and redesign, NexGen couldn't even manufacture anything that was stable enough to sell to anybody. In the end, AMD bought them for their technology. There was some crossover in technologies between Cyrix and NexGen, but NexGen's crown jewels was its RISC-like X86 "emulation" technology. I’m sure that I'm glossing over a LOT in my descriptions.

 

[LiamC]

As I recall AMD bought NexGen. The name on current AMD chips is AMD, but from the K7 on, the engineering has been NexGen.

K7 is Alpha (think DEC) not NextGen. Google Dirk Meyer

The original K8 design was being worked on by the K6/NextGen team--but this was scrapped in favour of K7 on steroids.

Elements of the Nextgen K8 design and team were rumoured to be involved in what became K9. This too has slipped, and in order to fool the street, K9 is now a multi-core K8. K10 is where these boys and gals are currently at--but at best, this may emerge H2 2007. My understanding is that it is a very ambitious design--hence all the slippage.

K7, K8 & possibly K9 have roots in Alpha.

 

[Tannin]

Damn it, I just lost a whole long post. (Stupid laptop put itself into standby, despite being docked and on mains power. I really, really hate the way that laptops use the same button for "wake up from standby" that they use for "power down now".

 

[Tannin]

Now, let's start dealing with some of the misinformation above — and there is quite a lot of it to deal with!

Bill is quite right in saying that Cyrix was always a fabless manufacturer, more or less, but both Bill and the usually well-informed iGary get the details pretty compregensively garballed.

There were essentially three stages to the Cyrix fabbing arrangements.

The maths co-pros, the 486SLC and the 486DLC were fabbed by Texas Instruments, and also (I think) by SGS-Thompson. TI were certainly the main source. They were pretty much all sold with a Cyrix badge.

The 486DX/2, 486DX/4, 5x86, and 6x86 Classic were mostly but not exclusively fabbed by IBM. The IBM-made chips were sold under both brands (IBM and Cyrix) in roughly equal numbers. And yes, Bill, there was a no-money-changes-hands arrangement which meant that for at least some of the several contracts, IBM got paid in kind: Big Blue got to keep half the finished parts. But there were various different contracts with different terms, the details of which I don't know.

By no means all of the Cyrix chips of this era were IBM fabbed, however. Somewhere at home I have a 486DX/2-80 colorfully marked with the Texas Instruments name, and I assume (but don't know for sure) that it was manufactured by TI. More significantly, SGS-Thompson also made Cyrix-designed parts, notably the 6x86 Classic. This may or may not have been sold with a Cyrix badge, but the SGS-Thompson branded and manufactured 6x86-166 sold in reasonable numbers late in the market life of the various 166 classic CPUs - i.e., the 6x86-166, Pentium Classic 166, and K5-166. These were all replaced quite soon afterwards by the Pentium MMX 166, 6x86 Classic 200, and K6-166. The 6x86MX came along a few months later.

The 6x86MX began in the same mould: Cyrix design, IBM manufacture, sold both ways. This applied to the 166, 200, and 233. I don't believe that anyone else except IBM fabbed these parts.

But starting with the 266, things changed. For starters, the original release of the 83MHz bus 6x86MX-266 was, for the first time, not shared equally between the two companies.

IBM avoided the 83MHz bus chips and opted for another 0.5 multiplier @75MHz to get the same PR rating--> Cyrix would offer 2 x 83 = 166MHz PR200 whilst IBM would offer 2.5 x 75 = 187MHz PR 200.

Wrong. IBM wanted to release the part, but Cyrix didn't believe it was stable enough yet. So IBM released it, Cyrix stayed with 233. Cyrix were right, IBM were wrong. The part was very quick (by the standards of the day) but a very tricky damn thing to get right. (83MHz was an absolute bastard of a bus speed.) The manufacturing, once again, was all IBM. Cyrix did release a 266 (renamed M-II 266) a couple of months later, but sold only a handful of them as by that time the (vastly improved) 300 part was where the action was.

The 6x86/M-II 300 was the last of the great Cyrix chips. For the first time in Cyrix's history, their PR rating was way too optimistic - the 300 was barely PR-266 - but (unlike the 266) it was simplicity itself to work with, as reliable as sunrise, and priced like a PR-200. This last of the traditional knock-down, drag-out Cyrix bargain parts was manufactured by IBM to begin with, but then by National Semiconductor as well. Eventually, all M-II production was on the National Semiconductor fabs. By this time, Cyrix had been under a new management team for quite some time, and was headed up by an ex-Compaq CEO with a strictly sales background. This was a significant change, it saw good sales growth, but staff morale plummeting and the brightest stars on the design team leaving for greener pastures. So the last of the old-school Cyrix products was selling well, but there was nothing workable by way of new parts in the pipeline. They got up to the 333 .... and stayed there .... forever.

After that, there was nothing except promises. National Semiconductor was an absolute disaster for Cyrix. No continuity, no sense of direction, no new product: before too long there was no design team either. NatSemi made a handful of M-II 366 parts, and then degenerated into an organisation barely capable of making a press release.

It was a very sad day for the industry. Cyrix brought PC prices down further and faster than any other company before or since. Cyrix was also the first company in two decades to produce a desktop CPU faster than Intel's best. That alone was a huge achievement. It would not be matched again until the K6-233, which only held the speed crown for 6 weeks and was in any case a tricky damn thing at the best of times, then the K6-3/450, which had a similarly short life as the fastest desktop chip on the planet. The 6x86-200, in contrast, was the fastest thing money could buy for six straight months. It wasn't until the Athlon arrived in 1999 that a longer-lasting challenge to the former monopolist would succeed.

 

[Tannin]

Procs started off sort of alright in the early months; procs got slowly faster and hotter over the next several month but also got less and less expensive putting pressure on AMD and Intel.

Nope.

First, let's define what we mean by "started off". If we mean the start of the 6x86 (the part that Cyrix is most famous for), then quite cleary it was not like this.

The very first 6x86 parts did run hot, about the same as a Pentium-60. Cyrix and IBM went through a series of rapid revisions and took about a month, maybe two months to get the temperatures down to something reasonable. These early parts were manufactured in tiny volumes, and the only people who got to buy them were, by and large, the very keen early adopters. These were the 6x86-100 and the 6x86-120, plus a tiny handful of 6x86-133, which never sold in volume as it required a weird bus speed (55MHz) that practically no motherboards supported. The 166 was just a marketing man's dream at this stage, a paper part.

There were two key requirements for these chips: you needed to set the voltage right (of the 3rd-party builds we saw, probably two-thirds were set too high), and you needed a suitable heatsink/fan, which was hard to find at that time. The motherboards of this period, by the way, were generally crap. With any CPU, you could be fairly confident of having a system come back once or twice because the COAST RAM module was playing up, or because the board had just decided not to POST anymore, or for a hundred other reasons. The late-model 486 boards that came before were much more reliable, as were the next-generation Socket 7 boards that followed.

The original 166 and (particularly) 200 parts did run hot, and you had to take care with them. But they were perfectly managable with proper care (easier than a K6-233, about as difficult as a Thunderbird 1333 or 1400). In any case, they were not around for very long: the low-voltage revison (6x86L) dropped temperatures through the floor. You could use a crappy HSF on a 6x86L and it would go just fine.

Generally speaking, the first generation 6x86MX parts were very forgiving. Temperature was not particularly an issue with them. Only the 233 got close to the limits, and it was still substantially cooler than a K6-233.

The M-IIs were, in general, fine. The 266 had problems, but they were entirely attributable to the stupid 83MHz bus speed, not cooling. The 300 was as easy to work with as any chip has ever been. The 333 was a variable package: early on they could be quite difficult; later revisions were fine. I have always assumed that this had to do with NatSemi not getting the production process right for a while. I'm not sure now if heat was an issue for them or not. I remember fiddling with them more than we did the the 300s, but don't remember why. In any case, we didn't use all that many of them: the wonderful K6-2/300 was sweeping all before it at this time, and the Celeron-A 300 was another outstanding part.

 

[Tea]

Tannin, can you read?

(Huh? Of course I can. Why do you ask?)

Well, according to a website I know about, what you said about Bill being wrong was not correct.

(Gahh. What does the average website know about CPUs and history?)

Well, Tannin, the site I looked it up on says that what you said about IBM releasing an 83MHz bus chip that Cyrix didn't think was ready for prime time is ... well, you said that about the 6x86MX-266, but it was actually the 6x86MX-333.

(Bahh... You shouldn't believe everything you find through Google, Tea. Half of it is crap. What moron wrote that nonsense?)

You did.

(Oh.)

 

[tazwegion]

Never mind Tannin we could debate this till the cows came home, and still nobody would change their position :roll:

Like 'em or not... they never experienced FPU errors like *cough* Intel's Pentium *cough* :lol:

 

[Tea]

Possibly because the poor little Cyrix FPU started doing the floating-point bug test suite in December 1997 but it hasn't got to the end of the third loop yet. Tune back in around 2008 and we will see if it passed or not.

 

[Tannin]

Tea! I thought you were supposed to be on my side.

 

[Tannin]

These unscrupulous jackasses would build a computer system with a cheap 486 mobo, install a Cyrix 5X86 processor in it, and call the system a "586" computer system -- fooling the fools into thinking that this was one better than a 486 (or a Pentium).

Exactly how is this less honest than Intel selling a $2000 "high-end 1400MHz" chip that couldn't even outperform AMD's $200 low-end 900MHz chip?

In any case, the 5x86 was one better than a 486. Easily faster.

Didn't a "100 MHz" 5X86 really have a 66MHz or a 75 MHz processor clock?

In a word, no. It was a 100MHz part.

 

[Tannin]

I opted for a K6 / PA-2007 solution

And an excellent choice it was too. At this later time, the Cyrix CPUs still offered the best mainstream value for money, but for non-business oriented systems, or just for maximum possible performance, K6 was the way to go, and the PA-2007, though expensive, was the best darn board money could buy. Best one that I ever met, anyway. Strangely enough, it was an under-performer at 75MHz, a bit like the HX boards really, but with a K6 or a Pentium MMX it flew.

as far as 2+ manufacturing plants, the MX's I located @ cpu-collection.de would indicate that to be true Canada & USA, however both my 're-badged' (IBM) MX's were USA stamped.

Not strange at all. The MX parts you saw were made by National Semiconductor, the IBM badged ones by IBM Microelectronics.

 

[tazwegion]

Possibly because the poor little Cyrix FPU started doing the floating-point bug test suite in December 1997 but it hasn't got to the end of the third loop yet. Tune back in around 2008 and we will see if it passed or not.

Fair crack-of-the-whip Tea even my Intel Over Drive (PR66) isn't that slow... but perhaps my Commodore Vic 20 is? :lol:

No banana's for you this week you wicked simian! :evil:

 

[LiamC]

(Bahh... You shouldn't believe everything you find through Google, Tea. Half of it is crap. What moron wrote that nonsense?)

You did.

(Oh.)

:rofl: I knew I read that somewhere reliable! You had me worried for a second or 60

 

[Ted]

I had one of those horrid P60's and to this day it has given me hesitation when thinking of buying anything new that's Intel based again because I'm still pissed off about it.

I subsequently had several early Cyrix and AMD CPU based systems, some with chipsets other than Intel (when available) and they sufficed for my computing needs without suffering any great penalty in stability problems that I can recall, but maybe I was lucky in hitting the right combination?.

In continuation of my own personal protest against Intel I oneday purchased a new machine based around an M2-300 CPU and a SiS TX Pro based motherboard, (PC Chips M571) What can I say it was very CHEAP

That motherboard and CPU combination was one of the most frustrating hair pulling !!&$!@@ expletive crash happy lemons I had ever owned as it would irregularly crash and or lockup with no discernable reason until oneday several months later I wanted a bit more grunt in the FPU department (games of course) and decided to try an AMD K62-380 processor in it (underclocked)

Low and behold the crashes suddenly reduced to what is quite ordinarily normal for a Win 98 based machine. I eventually started playing around with the present but unsupported 83MHz FSB settings on that motherboard and managed to clock and run the K62-380 at 400 Mhz rock stable. To this day many years later that PC's motherboard with its chipset and PCI bus running out of spec with a CPU it was not designed to support is still in the hands of a family member and running as stable as anything of that era can.

That motherboard and AMD CPU combination was a joy to play with and learn from and though I still managed to crash it when I pushed a little too much, it always ran more stable while even overclocked than the previous M2-300 CPU at stock settings with no other hardware changed.

Without having the technical (qualified) background that some of you guys have in PC's I can only come to conclusions based on personal use and gut feeling, and that is that the M2-300 and subsequent M2-366 at least were possibly faulty in design (P60 again?).

I personally know several other people that owned PC's with M2-300 CPU's and without exception they were crash happy pieces of crap. Ditch the CPU for an AMD K62 or K63 and they were transformed into reliable 'useable' machines.

Since then I've had various VIA based socket A boards with AMD CPU's (still hate Intel with a passion) and suffered/endured the limitations and bugs of those chipsets. Currently I'm still running two of several nForce 2 based boards and have not regretted it having had no major issues. These days VIA 'and' Intel are both off my shopping list.

I'm hopeing the ULi 1695 chipset is seriously adopted by first tier MB makers as that is the only thing to come along in a long time to make me seriously consider ditching nForce 2.

 

[Tannin]

Hi Ted. Something was wrong with your M-II or the installation of it. It was probably the (quite uncommon) 75MHz version, as the (much more common) 66MHz bus one was practically bulletproof. We sold just over 100 M571 boards and didn't have trouble with a single one of them. (Nope, not one RMA. Amazing for a PC-Chips board! Probably the only good board they ever made.) I don't have a record of what CPUs we put on them, but it's a fair bet that 80 or 90 of the 100+ were MII-300s. There might have been a handful of 75MHz bus 300s, and right at the end we had some of the 333s. I doubt we'd have wasted a K6-2 on one of those boards (for the K6-2 we used FIC VA-503+ if we could get it, or any of several similar boards from other makers - people willing to pay the extra for a K6-2 were also the sort of people willing to shell out for a stand-alone video card). And doubtless there would have been some M571s used with older CPUs - a Pentium MMX 166 or a Winchip 200, for example - to replace a board that had failed.

There was a later revision of the M571 board that reverted to the usual PC-Chips crap, and it was a very different story.

That's interesting what you were able to do with a 571 at 83MHz. Hell, I never trusted them over 66!

Now for the $64 question: what, exactly, was wrong with your VIA chipset boards? It never fails to amaze me how many people say that VIA chipsets are crap, and how rare actual, documented problems are. Me, I'd trust a VIA chipset a long long way before I'd trust an Nforce 1 or Nforce II (and a shorter but still measurable way before I'd trust an Intel or SiS chipset).

The Nforce IIIs and Nforce IVs we have had these last few months, on the other hand, have (touch wood!) been completely flawless. So far, so good .... but after the random weirdnesses of the Nforce IIs and Nforce Is, which both seemed to go OK to begin with, I'm not convinced just yet.

 

[tazwegion]

That's interesting what you were able to do with a 571 at 83MHz. Hell, I never trusted them over 66!

That was NOT uncommon with the M571, I personally owned the 3.2a revision... and while the on-board audio/video sucked the chipset was quite stable, even @ 83Mhz... as for the K62's not being supported by the M571, that I'm afraid is a fallacy! remember AMD remapped the K62 package to allow 6x multiplier @ 2x settings? this was done so people could benefit from the latest AMD offering without upgrading from their 66Mhz boards at the same time, my K62-350 purred @ 4.5x83 though it would run @ 6x66 also, I was just greedy for more fsb grunt

BTW if you're looking for any support with that mainboard down the track Ted, EYO forums have quite a collection of M571 enthusiasts :lol: and they even maintain a dedicated web page @ M571.com :aok:

Without having the technical (qualified) background that some of you guys have in PC's I can only come to conclusions based on personal use and gut feeling, and that is that the M2-300 and subsequent M2-366 at least were possibly faulty in design (P60 again?).

Hmmm... with adequate cooling my MII 366 & GA-586TX2 combo is still AFAIK running smoothly for my niece, least I haven't had any complaints concerning it during the last 2 years :mrgrn:

but after the random weirdnesses of the Nforce IIs and Nforce Is, which both seemed to go OK to begin with, I'm not convinced just yet.

I'm really happy with my nForce2 Ultra 400, but I hear the first generation chipset was certainly NOT the one to own :roll: