I once sold off a Cx5x86-120 and the buyer then send me a message that he found out the CPU supports 4X multi!! Now I always test for 4X but it's a no show.

Never saw an IBM 5x86 faster then 100MHz and also only 2-3X multi. Can't say anything about the revisions.

I've been trying to get answers to some of these questions for almost 20 years.
The best I can tell is that the 120MHz part exists only in very small numbers. I've only ever seen these in CPU collections. It's said that originally IBM planned to sell the -120 and -100 like Cyrix/ST, but then scaled them back to -100 and -75. Maybe for heat or stability reasons...I guess a few 120 escaped into the wild before they changed their evaluation process. So in theory the 120MHz marked IBMs are the same as their 100MHz parts (which are known to overclock to 120MHz pretty reliably). Note that IBM rated their chips as 3.3V rather than 3.6V on the Cyrix/ST parts.
I have never heard of any 5x86c chips that are anything other than Stepping 0, Rev 5. Not to say they don't exist, but if they do they're probably about as rare as the 120MHz marked chips.
Same goes for the 3x/4x rated 5x86cs. Not known to exist.

One thing that's also interesting about the 5x86C is that it was in production far longer than the ST version. I think IBM used these in some of their portables until 1998/99. Of course, these were the PQFP versions, so the only way you're going to find them is by cutting them out of laptops or maybe on a PQFP on PGA adapter (these exist). There is an odd version of 5x86 PQFP that is rated for 3.2V rather than the usual 3.3V. I've always wanted to buy one to see if there were any further revisions to the core, but they are also quite rare.

They all seemed to drop out of retail pretty quick in the UK in the 90s, about 6 months after they first appeared. I had buddies wanting a cheap option to max out their 486 in 96 and 97 and they were just gone... possibly there was some spotty availability of ST branded ones. Where you'd see them most often were in surplus/refurbed IBM Aptiva, PS/1, Valuepoint type IBM models, being sold off cheap for a system, but spendy just to get the CPU. Where are all those you may ask? Well, it seems that the BIOS codebase that IBM Microelectronics were fond of at that time for 486 class was not Y2K compliant, whereas many competitors were, so I would imagine scrappage was accelerated by that.

What's weird to me is that, according to what others have said, IBM took over the most of the production for Cyrix 5x86's starting around end of '95/beginning of '96. This also seems to be the time period where most 4x 5x86's were made. It seems odd to me that IBM would maintain two separate production lines for the 5x86 (e.g. 2x/3x 3.3v models & 3x/4x 3.45v models). Does anyone have an IBM 5x86C made around early '96 that they can test to see if it has 4x? Likewise, has anyone found any chips from '97 or '98 to test?

Looks like I may be wrong. I think the late 4x parts are actually ST-made ones, not IBM-made. This would make sense because the ST-branded 4x parts I've seen have identical serial number printing to the 4x Cyrix-branded modules I have. IBM-branded chips all have very different serial number printing. Also, I found a week 13 1996 IBM part and it does not have 4x.

Plausible, I think Cyrix and IBM just wanted their focus on 6x86 and handed off 5x86 and 486 to ST, there was a definite flood of ST 486 chips at the computer fairs in late 96 early 97 when they were going dirt cheap.... I guess they found somewhere to get better money for the 5x86 though.

BitWrangler wrote on 2024-08-27, 16:47:

Plausible, I think Cyrix and IBM just wanted their focus on 6x86 and handed off 5x86 and 486 to ST, there was a definite flood of ST 486 chips at the computer fairs in late 96 early 97 when they were going dirt cheap.... I guess they found somewhere to get better money for the 5x86 though.

ST Always made bank on obsolete CPUS they were still making first gen ST6x86pr166 (3.3volt) chips for dryers into 2k

They had problems producing compliant 6x86 chips into 1996 resulting in the pr90

Anonymous Coward wrote on 2024-08-27, 23:34:

I believe my 5x86c is also from early 1996, and it’s 2x/3x s0r5. It seems 3x/4x was somewhat common with the ST branded chips, but it seems most of those went to Europe…I never saw any in Canada.

They were the low tier producer with meh fab capabilities, Europe was much friendlier with the slower clocks so it was probably a financial thing that the fast IBM/Cyrix chips stayed in the US and the ST produced chips with yield issues became the harder to market PR133 and down
thus being sent overseas in places where people were less focused on the latest fastest cpu.

I believe my 5x86c is also from early 1996, and it’s 2x/3x s0r5. It seems 3x/4x was somewhat common with the ST branded chips, but it seems most of those went to Europe…I never saw any in Canada.

rmay635703 wrote on 2024-08-27, 19:09:

They were the low tier producer with meh fab capabilities, Europe was much friendlier with the slower clocks so it was probably a financial thing that the fast IBM/Cyrix chips stayed in the US and the ST produced chips with yield issues became the harder to market PR133 and down
thus being sent overseas in places where people were less focused on the latest fastest cpu.

From what I can tell, the chips produced by IBM were largely sold as the IBM 5x86C and unlike the 6x86 did not sell in very large quantities vs the Cyrix marked parts. I would guess most of them were used in IBMs own low-end computers. The vast majority of the Cyrix branded 5x86s seem to be made by ST. I think I have five or six Cyrix 5x86-100s and they're all from ST fabs. I've only seen pictures of Cyrix branded 5x86 with IBM markings on the bottom.

I think maybe 5x86 got included in that IBM/Intel deal where IBM could make CPU with Intel IP for it's own use, sold on board, upgrade boards, or in systems but not as loose CPU. I have heard that everything 386 and 486 class was included in that, despite maybe not having any or much at all Intel IP. However, it was a reason initially that Cyrix was partnering with IBM because IBM had intel manufacturing rights. Anyway that was definitely affecting how IBM marketted or didn't CPU between, I wanna say 92 and 97. IDK whether there's a full account of that story anywhere or whether you have to piece it together from press releases and newsettes in the preserved magazines. Doesn't help that there's almost nothing online about IBM Microelectronics before about 97 when they went all in on PowerPC, it was all on Prodigy, CompuServe, dial in BBS etc. The one or two early web pages that Archive.org manage to capture just sent you there. Modern equivalent is "Hi this is the official website of Xxxxxxx, follow us on Twitter, Instagram, Facebook.. " while being otherwise content free.

Edit: I brought up IBM microelectronics, because they are the ones that did the motherboards and sales to system integrators, so we'd know more about where 5x86 went if we knew what they were offering.

Recently, there was on eBay a 5x86 120 MHz... the seller claimed it supported the 4X multiplier and asked 550$ as a starting price for it (it was put on auction). I forgot to bookmark the page, but since I can't find it anymore, I guess -in the end- it was sold for that exorbitant price.

I've a couple of 5x86-100 that run fine at 120 MHz and also boot to DOS at 66x2, although the boards I have don't seem to be very stable at that frequency. One is step 0 rev 5 and it has a green metal heatsink with the Cyrix logo, the other is step 1 rev 3. and it has a blue metal heatsink with the IBM logo. Never had the pleasure to find a 4X variant.

Could you share more information about your IBM marked S1R3? All I’ve seen up till now we’re S0R5.

Anonymous Coward wrote on 2025-07-11, 00:05:

Could you share more information about your IBM marked S1R3? All I’ve seen up till now we’re S0R5.

The cpu comes with a blue heatsink attached. On it there's written in white "IBM 5x86C IBM26 5x86-3v3 100HB"

On the bottom of the CPU there are these markings: "Copyright USA 1995 Cyrix - IBM9314 P40003 - 50H5750 PQP"

The chip is a Step 1 Rev 3, as reported by CHKCPU

I believe this CPU is from week 40 of 1995. That seems to be pretty consistent with all (most?) of the Cyrix marked S1R3s being earlier in the production cycle. But this is the first time that I can recall seeing one from the IBM fab rather than ST.

MSxyz wrote on 2025-06-30, 16:11:

Recently, there was on eBay a 5x86 120 MHz... the seller claimed it supported the 4X multiplier and asked 550$ as a starting price for it (it was put on auction). I forgot to bookmark the page, but since I can't find it anymore, I guess -in the end- it was sold for that exorbitant price.

I've a couple of 5x86-100 that run fine at 120 MHz and also boot to DOS at 66x2, although the boards I have don't seem to be very stable at that frequency. One is step 0 rev 5 and it has a green metal heatsink with the Cyrix logo, the other is step 1 rev 3. and it has a blue metal heatsink with the IBM logo. Never had the pleasure to find a 4X variant.

Out of curiosity - set them to 3 x 50 instead at 4 volts and see if they post. Looks like a few QFP ones can do this, no word of any success with PGA so far

Anonymous Coward wrote on 2025-07-11, 00:05:

Could you share more information about your IBM marked S1R3? All I’ve seen up till now we’re S0R5.

I second this. Never seen an IBM in that stepping, only the odd cyrix branded one. One was a 100, and one a legit 120 (which had issues). I suppose this is proof they do exist.

We all know the IBM variety is almost universally able to run at 120mhz whereas the cyrixes often can not...I wonder if the S1R3 had trouble passing IBMs more stringent tests leading to very few making the good bin and hence very few out there.

Seeing it DID make the good bin, I'd definitely be trying that bad boy with branch prediction enabled 😀

Under DOS I always enable FPU Fast, Out of order load/store and Branch Target Buffer.

These three gives the best improvements and I never had a problem.

If I remember correctly, step 1 version 3 CPUs are supposed to be stable even under Windows with Branch Target Buffer enabled.

MSxyz wrote on 2025-07-13, 20:21:

Under DOS I always enable FPU Fast, Out of order load/store and Branch Target Buffer.

These three gives the best improvements and I never had a problem.

If I remember correctly, step 1 version 3 CPUs are supposed to be stable even under Windows with Branch Target Buffer enabled.

From my experience, BTB is not 100% stable on S0R5 CPUs, even under DOS. Most games will run fine for several hours without any problems, but others will crash/hang much sooner, e.g. Transport Tycoon Deluxe. S1R3 CPUs worked fine with BTB enabled for me, but they tend to overclock less than their S0R5 counterparts. Do you enable only these three registers or more when turning BTB on?

Stesch wrote on 2025-07-14, 07:16:

From my experience, BTB is not 100% stable on S0R5 CPUs, even under DOS. Most games will run fine for several hours without any problems, but others will crash/hang much sooner, e.g. Transport Tycoon Deluxe. S1R3 CPUs worked fine with BTB enabled for me, but they tend to overclock less than their S0R5 counterparts. Do you enable only these three registers or more when turning BTB on?

Only those three. As for overclocking, I usually run this CPU at 120MHz, stock voltage, without issues.