What an extraordinary load of ignorant (insert other dirty word used in this thread already) from a person who is normally so wonderfully well-informed.
Merc, it's way past time you learned a little bit about Cyrix chips, and substituted some facts for blind prejudice. Seriously. You are so far off base with that post that you're standing somewhere outside the ballpark and across the street.
Installed competently, almost every one of the of the Cyrix CPUs functioned as perfectly as any other CPU.
Let's deal with the worst ones first, shall we? The MX-266, and a small number of variants of the more common chips (the 75 and 83MHz versions of the MX-233 are examples). These few examples were indeed poor performers. Just as poor and just as unreliable, in fact, as some of the other dud CPUs, such as the Intel 486-50, Pentium 66, Pentium III 600 Katmai, and Pentium III 1133, or the AMD 486DX-40, 486DX-80, and K6/2-550. Each of these was problematic. The worst of the lot was probably the 486-50 (an Intel part) but none of them was particularly worth having.
Now, let's move on to the meat of your phobia: heat. The main reason some poorly-educated people used to think that Cyrix chips ran hot was simple incompetence. You would not believe how many third-party Cyrix-based systems we used to get come in with problems all stemming to one simple cause: pig-ignorant retailers who set them to the wrong voltage. Most Cyrix CPUs in the 100 to 166MHx class were designed to run at 3.3v, and worked just fine like that. A whole heap of morons used to just leave the default voltage jumpers on the motherboard alone because they didn't know what they were there for, with the result that all their Cyrix systems were running close to 10% over-voltage. Of course the chip would run hot if you treated it like that. So will any other chip.
Even after people started getting the voltage right (the Pentium MMX helped here, because you had the set the voltage jumpers for that chip too - even some of the thickheads started to learn a little bit about their trade at that point) a lot of shops used the wrong sort of heatsink/fan. They bought fans suited to the OEM (tray version) Pentium, and then tried to fit it to the Cyrix parts, which were a different shape. Cyrix responded by bundling an excellent HSF with their chips, IBM didn't bother. Any good retailer, however, was aware of this (it was no secret after all) and took sensible care to use the right part for the right job.
Thirdly, there was the matter of support components. Because Cyrix CPUs were such outstanding value, they became very popular with the bottom-of-the-market places, who proceeded to build systems with perfectly good Cyrix CPUs (because they were cheap) but ruin them with C-grade RAM, PC-Chips motherboards, and all the other usual nastiness. The problems systems had were (in the main) nothing whatever to do with the Cyrix CPU, they were caused by using a cheap motherboard with crappy RAM and an incorrect HSF, but ignorant morons in general, and quite a few otherwise well-educated people who should have known better, rushed in to blame the CPU.
If you took the trouble to build a Cyrix-based system with the same care, the same product knowledge, and the same quality of components that you put into an AMD or Intel-based system, you would up with a system that was equally reliable, a little bit cheaper, and (for most but not all purposes) quite a lot faster. During the 100 to 200MHz period, Cyrix were simply the best there was.
Notice that I said "the same quality of components", not "the exact same components". I mentioned cooling already. Some motherboards and motherboard chipsets worked great with Intel CPUs and sluggishly with Cyrix ones (the Intel HX is a prime example); with other ones the reverse obtained (VIA 580-VP is an example), with others again it didn't seem to matter either way. Many an Intel-coddled baby plugged a 6x86-166 into an HX mainboard (because he already "knew" that the HX was "the best", and didn't stop to consider the fact that he'd made that judgement using an entirely different CPU design), measured the performance, and concluded that he had "a slow chip". In fact, of course, he had a poorly chosen CPU and chipset combination. Put the HX with a Pentium and it would fly. Put the Pentium on a VIA board and it went slower. But put the VIA chipset together with the Cyrix chip (i.e., build your system with matched as opposed to mis-matched components) and you got a seriously fast result. (I still have benchmark figures for a lot of this lying around somewhere. The differences were quite substantial.)
I don't begrudge you your $150, but those customers of yours got ripped off. If you had known this part of your trade as well as you know most other parts of it, you would have been able to diagnose the actual cause of the problems brought to you (probably incorrect voltage or wrong HSF 75% of the time, running a motherboard chipset out of spec in the other 25% of cases) and fix it for more like $30.
No matter, it's all water under the bridge now.
But make no mistake: it was Cyrix who ushered in the price-performance revolution late last decade, and the Cyrix legacy that continues to provide you, me, and every other PC user with cheaper, faster systems than we would otherwise have the benefit of.
The directors of National Semiconductor, by the way, should have been shot for gross incompetence. They truly had the Novell touch.
