That's quite interesting. I have both the "L" PR166+ and the non "L" But I've never gone farther than Win98SE on them. How long does WinXP take to boot on one of these? How responsive could it possibly be? I assume it was the release version of XP or at least no later than SP 1?

Interesting

retrofool wrote:

That's quite interesting. I have both the "L" PR166+ and the non "L" But I've never gone farther than Win98SE on them. How long does WinXP take to boot on one of these? How responsive could it possibly be? I assume it was the release version of XP or at least no later than SP 1?

I was just trying XP for fun, no real use. It is SP3 and takes about a few minutes to boot, and slow as hell afterwards 🤣
By the way, Windows 7 SP1 refuses to boot on 6x86 or 6x86L but runs fine (sluggish of course) on P54C. Win7 might requires more Pentium specific instructions that are not supported by 6x86 nor 6x86L.

lazibayer wrote:

retrofool wrote:

That's quite interesting. I have both the "L" PR166+ and the non "L" But I've never gone farther than Win98SE on them. How long does WinXP take to boot on one of these? How responsive could it possibly be? I assume it was the release version of XP or at least no later than SP 1?

I was just trying XP for fun, no real use. It is SP3 and takes about a few minutes to boot, and slow as hell afterwards 🤣
By the way, Windows 7 SP1 refuses to boot on 6x86 or 6x86L but runs fine (sluggish of course) on P54C. Win7 might requires more Pentium specific instructions that are not supported by 6x86 nor 6x86L.

Have you tried Windows 7 on a K6 or Cyrix MII?

sliderider wrote:

lazibayer wrote:

retrofool wrote:

That's quite interesting. I have both the "L" PR166+ and the non "L" But I've never gone farther than Win98SE on them. How long does WinXP take to boot on one of these? How responsive could it possibly be? I assume it was the release version of XP or at least no later than SP 1?

I was just trying XP for fun, no real use. It is SP3 and takes about a few minutes to boot, and slow as hell afterwards 🤣
By the way, Windows 7 SP1 refuses to boot on 6x86 or 6x86L but runs fine (sluggish of course) on P54C. Win7 might requires more Pentium specific instructions that are not supported by 6x86 nor 6x86L.

Have you tried Windows 7 on a K6 or Cyrix MII?

No... But I can try that 😎

sliderider wrote:

lazibayer wrote:

retrofool wrote:

That's quite interesting. I have both the "L" PR166+ and the non "L" But I've never gone farther than Win98SE on them. How long does WinXP take to boot on one of these? How responsive could it possibly be? I assume it was the release version of XP or at least no later than SP 1?

I was just trying XP for fun, no real use. It is SP3 and takes about a few minutes to boot, and slow as hell afterwards 🤣
By the way, Windows 7 SP1 refuses to boot on 6x86 or 6x86L but runs fine (sluggish of course) on P54C. Win7 might requires more Pentium specific instructions that are not supported by 6x86 nor 6x86L.

Have you tried Windows 7 on a K6 or Cyrix MII?

Just tried and they both work. Interestingly Windows 7 can correctly recognize the model of K6 but not MII.

Thanks a lot for an interesting findings. Yesterday, I've realised, that my 6x86 P166+ is in fact IBM 6x86L PR200+ 😎 But it was running for all of it's life downclocked (to 2x 66 MHz) and with overvolted I/O (maybe to compensate for lower clocks 😁) for no obvious reason, apart from wrong mounted heatsink barely touching the IHS.
So my CPU has to run at 150 MHz, fine. On CPU itself, the only frequency printed is the total chip frequency, no mention of multiplier and bus speed like on MII cpus.
I found that Cyrix's PR200+ is listed as 2x 75 MHz whereas IBM's chip is 2,5x 60 MHz (http://www.cpu-collection.de/?l0=co&l1=IBM&l2=6x86)? This confused me a little, since my board has "strange" multiplier settings, in the manual only the exact CPU are listed and on PCB itself there is a little table with "inversed" multiplier (something like on attached picture from another board revision).

So I tried to jumper it as 2,5x 60 MHz, board reports PR166+ and I'm not sure what multiplier is in fact in use. WCPUID is not working (newest version doesn't show anything; older cannot launch), HWiNFO for dos doesn't lounch (not enough conventional memory, although running from safe dos prompt) and sisoft sandra was reporting 120 MHz. Which might be true, as I'm reading this thread. I tried to google it and it seems that 6x86 and 6x86L don't support half multipliers?

Motherboard manual shows Cyrix 6x86 P200+ configured with 50MHz bus. So the only way to jumper it on my board (it is not able to run at 75 MHz bus speed, or it is not documented) will be 3 x 50MHz, am I right? If so, then the http://www.cpu-collection.de website is wrong, since 2,5x multiplier is not possible.

mv_cz wrote:

Thanks a lot for an interesting findings. Yesterday, I've realised, that my 6x86 P166+ is in fact IBM 6x86L PR200+ :cool: But […]
Show full quote

Thanks a lot for an interesting findings. Yesterday, I've realised, that my 6x86 P166+ is in fact IBM 6x86L PR200+ 😎 But it was running for all of it's life downclocked (to 2x 66 MHz) and with overvolted I/O (maybe to compensate for lower clocks 😁) for no obvious reason, apart from wrong mounted heatsink barely touching the IHS.
So my CPU has to run at 150 MHz, fine. On CPU itself, the only frequency printed is the total chip frequency, no mention of multiplier and bus speed like on MII cpus.
I found that Cyrix's PR200+ is listed as 2x 75 MHz whereas IBM's chip is 2,5x 60 MHz (http://www.cpu-collection.de/?l0=co&l1=IBM&l2=6x86)? This confused me a little, since my board has "strange" multiplier settings, in the manual only the exact CPU are listed and on PCB itself there is a little table with "inversed" multiplier (something like on attached picture from another board revision).

multi.png

So I tried to jumper it as 2,5x 60 MHz, board reports PR166+ and I'm not sure what multiplier is in fact in use. WCPUID is not working (newest version doesn't show anything; older cannot launch), HWiNFO for dos doesn't lounch (not enough conventional memory, although running from safe dos prompt) and sisoft sandra was reporting 120 MHz. Which might be true, as I'm reading this thread. I tried to google it and it seems that 6x86 and 6x86L don't support half multipliers?

Motherboard manual shows Cyrix 6x86 P200+ configured with 50MHz bus. So the only way to jumper it on my board (it is not able to run at 75 MHz bus speed, or it is not documented) will be 3 x 50MHz, am I right? If so, then the http://www.cpu-collection.de website is wrong, since 2,5x multiplier is not possible.

I think you had overvolted the core voltage rather than the i/o voltage.... The latter should be roughly the same (3.3 - 3.5V).
The IBM chips should run at 75x2 as well. From what I read and tested 6x86s and 6x86Ls do not take half multipliers. 50x3 is way too slow; it's slower than 66x2. What's your motherboard? And more importantly, what's the clock generator on your board?

The core voltage was correct (2.8V), but the I/O voltage was set to 3.5 instead of 3.3V. No big deal since it apparently worked for few years 😀
CPU itself is indeed specified for 2x multiplier, decoding from model number, so it has to be 2x75 MHz. Maybe previous user saw, that the board does not support higher FSB, so he set it to 2x 66 MHz.
My mobo has only two jumpers for FSB (JP41, JP42 - http://support.pcpartner.com/support/man/intel/832003.pdf ) and all possible combinations are listed for FSB 50/55/60/66. Above those jumpers is "reserved" JP45 jumper which might be used for another combinations of FSB speeds?
(simmilar situation is with multplier jumper JP4 which is also reserved but serves as third multiplier pin, it is verified by another user on the net)

The best thing with the 6x86 is the awesome speed!

Like this Superpi 1M score! (Familiar? Yea this channel often does reruns these days)

There is no cheating involved! Both L1 and L2 cache are activated and working. All BIOS settings are set to the optimal (fastest) values and even Linar Burst is supported and working.

mv_cz wrote:

The core voltage was correct (2.8V), but the I/O voltage was set to 3.5 instead of 3.3V. No big deal since it apparently worked […]
Show full quote

The core voltage was correct (2.8V), but the I/O voltage was set to 3.5 instead of 3.3V. No big deal since it apparently worked for few years 😀
CPU itself is indeed specified for 2x multiplier, decoding from model number, so it has to be 2x75 MHz. Maybe previous user saw, that the board does not support higher FSB, so he set it to 2x 66 MHz.
My mobo has only two jumpers for FSB (JP41, JP42 - http://support.pcpartner.com/support/man/intel/832003.pdf ) and all possible combinations are listed for FSB 50/55/60/66. Above those jumpers is "reserved" JP45 jumper which might be used for another combinations of FSB speeds?
(simmilar situation is with multplier jumper JP4 which is also reserved but serves as third multiplier pin, it is verified by another user on the net)

I thought the previous owner set the core voltage to 6x86's 3.5 volts. Yes setting vio to 3.5v is no big deal.
I am not familiar with this board. JP45 does look like bus speed related. Could you please take a picture of the clock generator chip, or post its model number?

Looks like the clockgen is IMI SC653DYB, nothing I'm familiar with 🙁
Looks like it supports 75 MHz http://www.ic72.com/pdf_file/i/26717.pdf

mv_cz wrote:

Looks like the clockgen is IMI SC653DYB, nothing I'm familiar with 🙁

Good news! It takes 3 inputs and can produce 75MHz bus speed. It can also fix the PCI speed at 32MHz so the PCI devices can feel less painful at 75MHz bus speed. 😁

http://www.ic72.com/pdf_file/i/26717.pdf

-- EDIT --
You got it before I did! 🤣

Wow it works, as I found out, the FSB table from the manual corresponds with data sheet, except from first pin S2, which is in fact the undocumented JP45 jumper. 1 = 1-2 jumper setting and 0 = 2-3 jumper setting.
Since I've jumpered it yesterday to 60 MHz FSB, all that was needed to do was moving the undocumented jumper JP45 from 1-2 to 2-3 to achieve 75MHz fsb speed with fixed PCI speed. And finally the board recognized my CPU as PR200+ and also CHKCPU and Speedsys are reporting it as 150 MHz. I don't know whether the performance corresponds to 6x86 PR200+ or not. Speedsys shows 100.49 points, which is a little more than P-133 and far from 6x86-233.

mv_cz wrote:

Wow it works, as I found out, the FSB table from the manual corresponds with data sheet, except from first pin S2, which is in fact the undocumented JP45 jumper. 1 = 1-2 jumper setting and 0 = 2-3 jumper setting.
Since I've jumpered it yesterday to 60 MHz FSB, all that was needed to do was moving the undocumented jumper JP45 from 1-2 to 2-3 to achieve 75MHz fsb speed with fixed PCI speed. And finally the board recognized my CPU as PR200+ and also CHKCPU and Speedsys are reporting it as 150 MHz. I don't know whether the performance corresponds to 6x86 PR200+ or not. Speedsys shows 100.49 points, which is a little more than P-133 and far from 6x86-233.

That sounds about right

The scaling is identical to my 6x86 33 MHz.

100/150 is 2/3

22/33 is also 2/3

6x86 33 MHz. Speedsys dosn't get the FSB right. The FSB is 33 MHz, driven by the clock signal for the PCI bus. Some VIA VPX boards let you do awesome things.

I remember trying Windows XP on an IBM 6x86L PR200+GP back around 2001/2002. Not for very long mind you, it wasn't a pleasant experience. I stuck with 98SE & 2000 dual boot on that PC instead.

I do recall once running into a piece of software that refused to run due to incompatibility regarding Pentium extensions, or it's perceived lack of them. Can't remember what software it was unfortunately.

I still have the CPU in storage too, along with an IBM 6x86MX PR300 & a Cyrix MII-300GP. I don't even recall how I acquired all those, only that I either got them free or very cheaply. I think I must have used them as a temporary upgrade for a pre-MMX Pentium 100MHz system, before finally acquiring and settling on an AMD K6-2.

About a year ago I tested all 3 Cyrix CPUs out, on both a Socket 5 motherboard and a Socket 7 motherboard. They all worked great on the Socket 7 of course, but the Socket 5 would freeze before starting the memory test at POST.

Skyscraper: thanks a lot, it seems good

Yep the 6x86 wasn't that fast, but it was really the best bang for the buck if you were poor student somewhere in Central Europe (like me 😀 ) and wanted to upgrade your PC. I bought Cyrix MII-300 as new, it was really really cheap even as new (1998-1999) and it was the max option I could insert into my pcchips sc7 board (max 3.5 x 66) without exchanging the whole thing. I upgraded from 6x86MX PR166 (bought as used, again for cheap) which wasn't that big difference and still the MII was nowhere near Intel or AMD in gaming, but I was even able to play the original Unreal with my Voodoo1. Although the game needed a patch to run on 6x86/MII.

Still cyrix has compatibility issues, for example I now cannot install DirectX 8 (but DirectX 7 was OK) on the PR200+ since it says my CPU is not pentium class or higher. Don't know how to bypass this, maybe it would be better to drop P55C in the board, install DirectX and put the 6x86 back in 😈
I didn't intend to use this 6x86 in my setup, but was in the system from the previous owner and when I realised, that it wasn't PR150+ but it was in fact an underclocked PR200+ I tried to jumper it correctly and see what it can do. Obviously whoever installed this CPU into the board was suprised it officially didn't support 75 MHz FSB (let alone with fixed PCI speed 😲 ), jumpered and downclocked it as 2x 66 MHz and didn't really care much about the motherboard user manual saying it should be configured as 3x 50 MHz.

Older versions of Matrox display card drivers (e.g. 4.33) are required on the Cyrix/IBM 6x86 and 6x86L chips, whereas the original Pentiums (non-MMX) could use later versions (e.g. 6.28). I'm not sure why this is. MX and MII verisons of the Cyrix/IBM 6x86 could use later Matrox display card drivers.

It is also very interesting that Cyrix and IBM chips respond differently to the BFO clock multiplier selection. I figured this of all things would have been uniform. The IBM's BFO's were a little closer to Pentium's values.

Like this Superpi 1M score!

Huh? Does it really take over an hour? Pentium MMX only takes 15 minutes.

Yep the 6x86 wasn't that fast, but it was really the best bang for the buck

6x86 was good for playing old emulators like NESticle, Genecyst, Magic Engine, ZSNES, early MAME, etc.