This page is translated by Jeff Lim (Singapore) from my Japanese page. Thanks Jeff !!
BP6 DUAL PENTIUM III over clock
As Vio is now variable, you can try to overclock even further! First, the PEN3 either runs, or in the case where it does not simply increase Vio and Vcore.
The PENTIUM III lot number I am using is the SL4MA, serial number 7041a117.
With Vio and Vcore bring variable, we now can try overclocking up till a FSB of 124MHz.
|
FSB |
Vcore |
Vio |
WIN2K boot |
3DMARK2000 |
Remark |
|
102 |
1.7 |
3.5 |
OK |
Perfection |
|
|
104 |
1.7 |
3.5 |
OK |
Perfection |
|
|
106 |
1.7 |
3.5 |
OK |
Perfection |
To here, it operates easily with the stipulated 1.7v. |
|
108 |
1.7 |
3.5 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
108 |
1.8 |
3.6 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
108 |
1.85 |
3.6 |
OK |
Perfection |
|
|
110 |
1.90 |
3.37 |
No, |
|
It is the data before the Vio variable. |
|
110 |
1.85 |
3.6 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
110 |
1.90 |
3.6 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
124 |
1.90 |
3.7 |
No, |
|
|
From the results above, running the PEN3 at 848MHz with a FSB of 106MHz usage of 1.75V for Vcore and 3.5V for Vio is a must. With that 3DMARK started hitting the new numerical value of 4154.

The next goal in remodeling the BP6 will be having 3DMARK2000 running successfully at a FSB of 110MHz. Since the BIOS can only provide a Vcore of 1.90V max, perhaps a successful overclocking at a FSB of 110MHz can be achieved with Vcore at 1.95V? With that in mind the next remodeling will be changing the core voltages. If the FSB were to hit 124MHz, we will have a gigahertz processor! But that will be our last goal.
Whilst looking at the BP6 I had decided to take the connector of the ATA66. This is so that I can make a jumper out of it. The dipswitch method, which someone suggested, is a better idea, but because of the fact that you have buy it especially for just this sole purpose it is not worth it. Furthermore why not put the ATA66 connector to good used instead of having it idling?
Have the connector soldered with 8 wires at the rear as shown in the picture below.

As for the wiring, Vss will be the one connecting to all 3 terminals on
the top for both CPUs. For the bottom connection, the extreme right is the
Vid0 for the left CPU followed by the Vid1, Vid2 and the Vid0,
Vid1, Vid2 for the right CPU in sequence.

There is only one jumper set in the photograph,
but if both CPUs need to
be set put in another jumper. Both short-circuiting Vid1 gives a Vcore of 1.80V as a default.
Both short-circuiting Vid2 gives a Vcore of 1.90V as default. If both Vid1 and both Vid2 are short-circuited, Vcore defaults to 2.0V
No short circuting gives 1.70V as a default
After all that is done solder them to the socket370 pins involved.

As for the connection, please look at the pin arrangement shown below carefully. Reference the top right edge of the picture, on the left, below to the top left edge of the socket370 pins in the picture above.


Actual Layout As seen on the BP6 in the picture above

Now the default voltage, for Vcore, has been changed from 1.70V to 1.80V. With this method, Vcore can or set to 1.90V or even more. The experimental results of tweaking Vcore and Vio for overclocking are shown in the table below.
|
FSB |
Vcore |
Vio |
WIN2K boot |
3DMARK2000 |
Remark |
|
102 |
1.7 |
3.5 |
OK |
Perfection |
|
|
104 |
1.7 |
3.5 |
OK |
Perfection |
|
|
106 |
1.7 |
3.5 |
OK |
Perfection |
To here, it operates easily with the stipulated 1.7v. |
|
108 |
1.7 |
3.5 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
108 |
1.8 |
3.6 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
108 |
1.85 |
3.6 |
OK |
Perfection |
|
|
110 |
1.90 |
3.37 |
No, |
|
It is the data before the Vio variable. |
|
110 |
1.85 |
3.6 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
110 |
1.90 |
3.6 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
110 |
1.95 |
3.8 |
OK |
Midway hang-up |
Screen blacks out, have to reset |
|
110 |
2.0 |
3.8 |
OK |
Perfection |
However the GFORCE 256DDR was set at its
default clock settings. |
|
124 |
2.10 |
3.8 |
No, |
|
It is indeed sad that itfs not achievable. |
With the fact that 3DMARK completed successfully at a FSB of 110MHz proves that the system is stable. Well the FSB of 124Mhz becomes a vision for the future.