bloodem wrote on 2020-04-02, 09:00:

Hello guys, […]
Show full quote

Hello guys,

I have what I consider to be a very nice 440BX board (MSI MS-6163 ver 1.0), which is probably as close as it gets to perfection for Pentium 2 Deschutes / Pentium 3 Katmai CPUs.
Now, since the board supports FSB frequencies up to 150 MHz (I'm actually running a Pentium 2 350 MHz CPU @ 504 MHz with a 144 FSB setting and it's unbelievably fast and stable), I really want to try out a Coppermine CPU. I also updated the BIOS to the latest one (which is actually for the Pro version of the board, but it's also supported by ver 1.0).
The issue is that the board only supports 2.0 - 2.4 CPU voltage (configurable in BIOS) and Coppermines are safe up to about 1.8V max.

So you can understand my frustration that, even though I'm 100% sure this board could perfectly handle Slot 1 Pentium 3 CPUs, my hands are tied because of minimum voltage.
"
Does anybody know of a miraculous solution to this problem, anybody else that might have the same or a similar board?
I know about the Powerleap slotket adapter which supports lower voltages, but I actually want to run pure Slot 1 Coppermine CPUs.

Coppermine CPUS have no issues with 2.0V with active cooling

"All" slot1 boards that support Deschutes and Katmai support 1.8V if they follow the official design guidelines. You might need a slotket with VID selection to force the VRM circuit to provide 1.8V though unless you have one of few Coppermine CPUs with 1.8V VID. CPUs with VIDs below 1.8 will confuse the motherboard unless used with said slotket.

I guess there could be boards that won't go below 2.0V eventhough they "should" support it and they will probably refuse to boot if the CPU (or slotket VID override) asks for a voltage below 2.0V.

(Also I think there is a good chanse that your motherboard supports the full Coppermine voltage range (1.5V - 1.8V) and you are seeing 2.0V as the lowest setting just because it's the default voltage for your CPU)

Thanks.
Yeah, that's what I was afraid of. Anyway, I will try a Pentium 3 700 MHz CPU in a few days and see if I can get this baby to work 😀

I was wondering if Intel's 180 nm parts tolerated 2.0-2.2V as well as IBM, NS, and AMD's, though I suspect some over-voltage issues are also tied to the surface-mount caps or resistors or diodes (I'm not sure what all of those are or when/why they're used other than surface-mount jumpers) might actually have a lower limit than the native silicon and interconnect.

But that doesn't seem relevant here, or at least up to 2.0V. I'm not sure the same could be said for 250 nm slot one chips run at 2.8V (or higher if boards allowed that). Where I'm pretty sure the 350 nm parts tolerate 3.3-3.5V fairly well like their Socket 7 counterparts.

Though adequate cooling for such overvoltage is always the other concern. (and overheating combined with overvolting can be destructive where one or the other alone may not be: and running hot means you're closer to thermal runaway, too) That and potentially damaging the board/chipset and not just the CPU.

2.0V for a 1.8V part is no drama, indeed if you want to overclock significantly you might well want to increase the voltage anyway. Just make sure cooling is good. I find big later SoA coolers with 80mm fans are great for keeping Coppermine CPUs nice and cool without too much noise.

Well, that's the issue, I will try to run a 1.65V part, not a 1.8V 😁.
The thing is, the BIOS also has an "Auto" voltage setting and I'm curious to try that out, although I don't have much hope that it will work for voltage settings below 2.0.
Cooling won't be an issue, the CPU cartridge looks like this:

Also, I have a buch of Noctua fans of all sizes, ranging from 50mm up to 120mm.
Still, I'm thinking that a 20%+ CPU voltage increase (0.35V) is probably way too much...

Later edit: according to this, I might actually have a shot. 😀

Your 'auto' voltage may be the key here. It is likely to work and if not the system will not boot. There is not a big chance on risk there. If it does not work your best option is to get yourself a 'Pro' version of that board. The one I had in the past ran Coppermines up to 1.100 Mhz 'out of the box' with the latest BIOS.

PARKE wrote on 2020-04-02, 16:13:

If it does not work your best option is to get yourself a 'Pro' version of that board. The one I had in the past ran Coppermines up to 1.100 Mhz 'out of the box' with the latest BIOS.

Nah, there's no fun in that ...
I already have too many Coppermine/Tualatin boards that work out of the box. Now I want to see one working, even though it's not supposed to work. 😉

'Fun' is relative in this case. It works or it does not. There is not much ingenuity involved unless you are willing to mod the VRM and the BIOS in case it does not work. But I understand the itch that needs scratching 😀

Yeah 😁
I think the BIOS is fine, the VRM is the one I'm concerned about. I'll update this thread as soon as I do the tests 😀

If you don't have a laser thermometer, you could use the finger-test method on a bench or open-case test run of the board and see how hot the regulators get.

Other than that, if the original capacitors are still there, that might be a concern, though I think less for destructive failure and more just for stability. (and non-bulged, non-visibly leaky caps can still dry out from evaporation in the long run)

That said, most boards with old electrolytic caps seem to do fine anyway. (and some with bulged caps still seem to do OK too)

From my observations, Coppermines can work between 1.5v and 2.0v. Some lucky 1Ghz Coppermines will work at 1.55v and almost everything below 750 Mhz.

UPDATE:

So, I finally got the chance to play with the Pentium 3 Coppermine 700 MHz CPU on the MSI MS-6163 ver 1.0 board.
First thing I did (with the old Pentium 2 CPU) was to set the BIOS voltage to AUTO and the frequency to 100MHz x 7.
I rebooted, powered off the board, installed the P3 700 MHz CPU and.... BINGO! To my amazement, not only did it instantly boot, but the actual voltage switched to 1.65V 😁
I accessed the BIOS and the voltage/multiplier options have changed, compared to before, when the P2 was installed. I can now choose between 1.65V - 2.00V, and multipliers up to 8.5.
I immediately set the FSB to 122MHz (fun fact: in order to get an effective 133MHz FSB, you need to set it to 122MHz in the BIOS - this is probably a bug caused by the latest BIOS update, which is in fact for the Pro version of this board).
Also set the voltage to 1.75 (but I will also test with 1.7, it could work too), booted into Windows, benchmarked the hell out of it for more than two hours and it works like a charm. 😀
So, as I guessed, this board is a keeper!

UPDATE 2:

Tested the CPU with 1.7V ---> stable.
So I went for 1.65, the default, and so far it's also stable. Not sure if this is common with these chips, but I'm impressed. It seems that this CPU could have easily been sold as a 933 or even 1 GHz part.

The voltage depends on the stepping and the speed, see the attached.

I have 2 x Pentium IIIE (Coppermine) 1.00GHz (10 x 100MHz) Stepping 6 in my Gigabyte GA-6BXD system and they run at about 1.744 volts.

bloodem wrote on 2020-04-16, 18:06:

UPDATE 2:

Tested the CPU with 1.7V ---> stable.
So I went for 1.65, the default, and so far it's also stable. Not sure if this is common with these chips, but I'm impressed. It seems that this CPU could have easily been sold as a 933 or even 1 GHz part.

What's the manufacture date of the chip? This isn't surprising if it was manufactured later in the PIII production run.

red-ray wrote on 2020-04-16, 18:30:

The voltage depends on the stepping and the speed, see the attached.

I have 2 x Pentium IIIE (Coppermine) 1.00GHz (10 x 100MHz) Stepping 6 in my Gigabyte GA-6BXD system and they run at about 1.744 volts.

Ehm... not sure why you are mentioning this, but... OK 😀 Yes, I know that higher frequency Coppermines run at a higher voltage. My 700 MHz CPU has a 1.65V default voltage, though.

Katmai500 wrote on 2020-04-16, 18:37:

What's the manufacture date of the chip? This isn't surprising if it was manufactured later in the PIII production run.

Yes, it's a later stepping (cB0), manufactured in 2000.

bloodem wrote on 2020-04-16, 16:36:

I immediately set the FSB to 122MHz (fun fact: in order to get an effective 133MHz FSB, you need to set it to 122MHz in the BIOS - this is probably a bug caused by the latest BIOS update, which is in fact for the Pro version of this board).

What was the BIOS that you used before you upgraded to the PRO version ?

I do believe it was the board's original BIOS. Don't remember the exact version, but it was from the beginning of 1998.

Maybe you are better off with an original BIOS for the non-PRO board ?
This is what you likely have:
ooooooooooooooooooooooo
BIOS MS-6163 PRO Version 3.9
w6163v3a.exe (283KB)
Release Date: 2000-09-04
Description:
-Fix CPU clock show error when FSB = 110.
-Fix CPU clock 90x8.5 will show ?42.
-Support MS-6915 ATA-100 Card.
-Fix iomega ZIP 250 ATAPI under POST shows error.
-Fix system hang when IDE HDD size over 65GB.
-Support new CPU. (500A & 533A MHz for Coppermine /128K)
-Fixed frequency error when clock is over 999MHz.

or:

Version 3.A 2000/10/11
This BIOS fixes the following problems of the previous version:
Fixed CPU L2 Cache error problem.
ooooooooooooooooooooooo

This is what should be installed according to MSI guidelines:
ooooooooooooooooooooooo
BIOS MS-6163, Version : 2.7
w6163v27.exe (269.9 KB)
Fixed Western Digital 153AA Hard Drive compatibility issue.
Fixed iomega ZIP 250 ATAPI under POST shows as ZIP-100, Mode 0.
Fixed system hang when IDE HDD size over 65GB.
Display CPU speed as 500A ; 533A MHz for Coppermine /128K (new Celeron) with 66MHz FSB frequency.
Fixed wrong CPU frequency display if the clock is over 999MHz.
ooooooooooooooooooooooo

As you can see the 2.7 and 3.9 versions share a couple of fixes and are likely from the same release period.

full release listing for the non-PRO version:
BIOS MS-6163, Version : 2.7 driver w6163v27.exe
BIOS MS-6163, Version : 2.6 driver w6163v26.exe
BIOS MS-6163, Version : 2.5 driver w6163v25.exe
BIOS MS-6163, Version : 2.4 driver w6163v24.exe
BIOS MS-6163, Version : 2.3 driver w6163v23.exe
BIOS MS-6163, Version : 2.2 driver w6163v22.exe
BIOS MS-6163, Version : 2.1 driver w6163v21.exe
BIOS MS-6163, Version : 2.0 driver w6163v20.exe