The reason seems to be two-fold.

1) Windows NT 4.0 would deliberately disable the L1 cache on Cyrix 6x86 CPUs below revision 2.7 due to an issue with "reflections", which I assume is referring to transmission line reflections.

2) The Cyrix 6x86 requires too much current (hence, power) compared to the Intel Pentium, and the motherboard's onboard voltage regulator may not be able to handle this, causing CPU instability.

reference: http://www.tomshardware.com/reviews/cpu-guide,13-2.html

Gona wrote:

Now I have found on the net that the Socket 7 ATX/AGP board Asus P5A (ALi Aladdin V) know 40MHz fsb but manual writes only 6x86-PR166+ and later Cyrix CPUs and for 166+ they writes: "The only IBM or Cyrix 6x86(L) (or M1) that is supported on this motherboard is revision 2.7 or later".

60MHz FSB is the minimum. Here is the slowest MHz result with that setting, using a Cyrix 6x86L at 1.0x multiplier.

gerwin wrote:

Gona wrote:

Now I have found on the net that the Socket 7 ATX/AGP board Asus P5A (ALi Aladdin V) know 40MHz fsb but manual writes only 6x86-PR166+ and later Cyrix CPUs and for 166+ they writes: "The only IBM or Cyrix 6x86(L) (or M1) that is supported on this motherboard is revision 2.7 or later".

60MHz FSB is the minimum. Here is the slowest MHz result with that setting, using a Cyrix 6x86L at 1.0x multiplier.

Thanks. Yes, it is not 1.5*40 but 1*60. First I have not found this "CPU E → 1.0x(1/1)" "CPU E: IBM/Cyrix 6x86, IBM/Cyrix 6x86L".

feipoa wrote:

The reason seems to be two-fold. […]
Show full quote

The reason seems to be two-fold.

1) Windows NT 4.0 would deliberately disable the L1 cache on Cyrix 6x86 CPUs below revision 2.7 due to an issue with "reflections", which I assume is referring to transmission line reflections.

2) The Cyrix 6x86 requires too much current (hence, power) compared to the Intel Pentium, and the motherboard's onboard voltage regulator may not be able to handle this, causing CPU instability.

reference: http://www.tomshardware.com/reviews/cpu-guide,13-2.html

Thank you much. Might this "The voltage regulator has to supply much more current" has killed my TMC board. Can I repair the board by replacing the voltage regulator? What type of motherboard has you used for testing your 6x86-P90+GP?

You'll want to review the specs on the Cyrix 6x86 for how much current it draws and at which voltage. I believe chipdb.org has the spec sheets on the 6x86. Look at the markings on your voltage regulator and look up the spec. sheet to see how much sustained current it can output.

If it is, in fact, the VRM on your board which blew, then replacing it should be fairly straight-forward, e.g. desolder and resolder the new one. In the past, I have replaced VRM's with units which can output more current, provided the pin-outs are the same.

On one of my motherboards, I operate the VRM near its current limit (socket 7 board with AMD K6-III-500). I added a small fan to the VRM which keeps it from getting too hot. If your VRM doesn't already have a heatsink, you might want to add one, but first be sure it is necessary by comparing the current spec. for the VRM and the target CPU.

The motherboard I used to test my 6x86-P90 can be found in the Ultimate 686 Benchmark Comparison's, either in the text of the website, or in the PDF. I should have listed all the motherboards with sub- or superscripts denoting which CPU used which motherboard.

And did you confirm that your VRM is busted by testing out a simple Pentium 75?

There is one of these Cyrix 80mhx 686's on ebay right now.

http://www.ebay.com.au/itm/Vintage-Cyrix-6x86 … YUAAOSwzoRZfK-u

feipoa wrote:

You'll want to review the specs on the Cyrix 6x86 for how much current it draws and at which voltage. I believe chipdb.org has […]
Show full quote

You'll want to review the specs on the Cyrix 6x86 for how much current it draws and at which voltage. I believe chipdb.org has the spec sheets on the 6x86. Look at the markings on your voltage regulator and look up the spec. sheet to see how much sustained current it can output.

If it is, in fact, the VRM on your board which blew, then replacing it should be fairly straight-forward, e.g. desolder and resolder the new one. In the past, I have replaced VRM's with units which can output more current, provided the pin-outs are the same.

On one of my motherboards, I operate the VRM near its current limit (socket 7 board with AMD K6-III-500). I added a small fan to the VRM which keeps it from getting too hot. If your VRM doesn't already have a heatsink, you might want to add one, but first be sure it is necessary by comparing the current spec. for the VRM and the target CPU.

The motherboard I used to test my 6x86-P90 can be found in the Ultimate 686 Benchmark Comparison's, either in the text of the website, or in the PDF. I should have listed all the motherboards with sub- or superscripts denoting which CPU used which motherboard.

And did you confirm that your VRM is busted by testing out a simple Pentium 75?

Thank you very much! I have seen specs at chipdb.org and I have found this:
Active ICC (input current) 80 MHz : typical 3.9A; maximum 4.7A
Than I have seen the VRM (Linear Technology LT1085CT) sheet and it seems the LT1085 modules are have output current 3.0A
On the TMC board the VRM has relative big heatsink. I have try to boot with Pentium 120 @ 40MHz without success.

Remembers me that intel made a 16MHz 486SX 🤣

Deksor wrote:

Remembers me that intel made a 16MHz 486SX 🤣

What about the 90mhz DX4? I have one of those 🤣

Well 75MHz DX4 was pretty standard, I had it in some noname laptop.

mv_cz wrote:

Well 75MHz DX4 was pretty standard, I had it in some noname laptop.

75mhz, 100mhz and 120mhz are the standard speeds. 90mhz dx4 is as far as I know only used in one range of OEM systems.
Running at 3x 30mhz it is a one off version.

I have selected an "expendable" mobo for a second test with Cyrix 6x86 PR80, this is PcPhips M560 TopGun. It has Acer ALI M1531 Aladdin IV chipset: http://leoromo.net/mirrors/motherboards.mbarr … hips/m560tg.htm
manual says FSP 60/66/75/83 multiplier 1.5 - 3.5 and says nothing about "only for Cyrix 2.7 or later". First I have try 1.5 * 60 but nothing. Than I have put back the P120 CPU and I can see the mobo sill live. Put back the Cyrix 6x86 PR80 and I have started to play with multiplier settings than "2.5x" the mobo has started with the Cyrix 6x86 PR80 with multiplier 1x so on 60MHz. At last now I know the CPU works.

Nvm1 wrote:

mv_cz wrote:

Well 75MHz DX4 was pretty standard, I had it in some noname laptop.

75mhz, 100mhz and 120mhz are the standard speeds. 90mhz dx4 is as far as I know only used in one range of OEM systems.
Running at 3x 30mhz it is a one off version.

486SX-16 was also an OEM-only chip, iirc. I got mine from a machine I found on the streets years ago (I think it was a Dell), I was like "wth they made these?!? 😵 "

The 486DX-90 was an AMD chip, right?

Gona wrote:

Some OPTi chipset support fsb lower than 50MHz. I have a Socket 5 board with OPTi 82C596/82C597/82C822 chipset (TMC PCI54PV) wha […]
Show full quote

feipoa wrote:

vetz wrote:

I have no clue how they got 40mhz FSB, and god damn that CPU must have been slow! I want to see some comparison benchmarks just for lolz!

While I did benchmark an actual Cyrix 6x86-P90+GP (80 MHz) for the U6BC, the closest to 80 MHz I could get it was 1.0 x 83 MHz, which is not quite the same. I think there were a few socket 5 boards with a 40 MHz FSB option.

Some OPTi chipset support fsb lower than 50MHz. I have a Socket 5 board with OPTi 82C596/82C597/82C822 chipset (TMC PCI54PV) what I can boot with 1.5 * 26.7MHz = 40MHz. There is also 1.5 * 22MHz = 33MHz but freezes at second bios screen.
Some weeks ago my Cyrix 6x86 PR80 has arrived. Almost all pins was bent but noone is missing, I have played a day to fix all of the pins.
I have tried the Cyrix 6x86 PR80 with the TMC PCI54PV board but it seems to me the CPU has killed my board. Now I have put back the P100 CPU and I have put a PC analyser to an ISA slot, but writes nothing...
My CPU has printed "016" on it so it is 3.3V according to:
http://www.pc-atrium.de/html/cyrix6x86.html
I don't thing that the 3.3V is too low for a 3.5V board.
Now I have found on the net ASUS Socket 7 boards where manual writes only 6x86-PR166+ and later Cyrix CPUs and for 166+ they writes:
"The only IBM or Cyrix 6x86(L) (or M1) that is supported on this motherboard is revision 2.7 or later".
Why is this?

This Shuttle HOT-541 430FX board also looks like it can handle 40 MHz FSB and it is listed as Socket7 😕 http://mail.lipsia.de/~enigma/neu/pics/mainbo … rd_i430fx66.jpg

mv_cz wrote:

This Shuttle HOT-541 430FX board also looks like it can handle 40 MHz FSB and it is listed as Socket7 😕 http://mail.lipsia.de/~enigma/neu/pics/mainbo … rd_i430fx66.jpg

according to its PLL datasheet it should be able to reach 16MHz at the bottom.

mv_cz wrote:

This Shuttle HOT-541 430FX board also looks like it can handle 40 MHz FSB and it is listed as Socket7 😕 http://mail.lipsia.de/~enigma/neu/pics/mainbo … rd_i430fx66.jpg

Good finding. I have also find an 430FX board also looks like it can handle 40 MHz FSB. It is DTK PAM-0054I and it is also Socket 7. Manual writes not only 40MHz but also writes 6x86 P90:

And I have a DTK PAM-0054I board:
https://gona.mactar.hu/6x86-80GP/Dtk_pam0054i.jpg
So I have tested my 6x86-80GP CPU:
https://gona.mactar.hu/6x86-80GP/6x86-80GP.jpg
First It was frozen in the beginning of the boot. I have turned off and touched the VRM. It was really hot. I have installed temporary an intensive cooling than it worked:

https://gona.mactar.hu/6x86-80GP/pam0054i_aida.jpg
I think HOT-541 has more strong VRM.

wow and how well/bad it performs? this CPU was originally marked as PR80 and later changed to PR90?

mv_cz wrote:

wow and how well/bad it performs? this CPU was originally marked as PR80 and later changed to PR90?

I have tested only Wizardry VII: Crusaders of the Dark Savant to find out that this CPU is affected by the Pentium reboot error or not, and it is affected. This CPU originally marked as 6x86-80GP to 1995 week 51 (in the numbers the "551" is the manufacturing date, the "5" is the year and the "51" is the week). This is my one:

and from 1995 week 52 marked as 6x86-P90+GP. The "PR" has come later, the "PR90" is not exist, there are 6x86-PR120+GP and higher numbers.
http://www.pc-atrium.de/html/cyrix6x86.html
The overall performance is reach the Pentium 90 according to feipoa Ultimate 686 Benchmark Comparison (here the CPU has run on 1x75 and 1x83MHz):
download/file.php?id=11671

Oh nice, you have a board which supports the 40 MHz FSB for these obscure chips. I wish I had such a board when I ran these benchmarks, so that I can see how the 40 MHz FSB reduces performance of these chips. I suspect the overall performance of the Cyrix 6x86-80 at 2x40 might be closer to that of the Pentium 75. Remember, at FPU operations, the 6x86-80 is more like an Am5x86-133.

Does anyone know what criteria was used to establish the P rating of these? I suspect mostly, if not exclusively, integer performance.

Since you have a 40 MHz FSB supporting board, perhaps you can run the benchmarks specified in the U6BC to see how it performs to my 1x75 and 1x83 tests?

I was surprised to see a Week 44, 1995 Cyrix 6x86-80 on that pc-atrium site. I always thought that the Cx6x86 chips were produced starting in January 1996. Its too bad they had to cancel production of the 5x86 chips to make room for the 6x86 chips. Also a bummer that IBM, who I think produced the Cx5x86 chips into 1998, did not produce 4x-capable 5x86-133 chips. I wonder if there was a clause stating that IBM could not produce the 4x chips.

I have no PCI G200. If I find one for cheap I can run the benchmarks. Now I see a DTK PAM-0054i on eBay in my country: http://www.ebay.com/itm/282658850728. It is a bit older than mine because this has VRM connector but my has not.

Yes Week 44, 1995 is very early, I was surprised too. I fully agree, it would have been much better to manufacture Cx5x86 than 6x86. It would be better business in that times and also would be better for us.