As you can see it supports a 66MHz, 100MHz, 133MHz, and 200MHz FSB! Due to the fact that it would be impossible to hit a 200MHz FSB I will not be covering it... however after looking at the following you should be able to work out a way to do it if you so wish to try it. For those questioning about why Im saying a p4/celeron has a 100/133MHz FSB and not a 400/533MHz FSB is because the p4/celeron operate on a Quad Data Rate bus so that's why they say it has a 400/533MHz FSB. Its the same with the 200/266/333MHz FSB of the Athlons/Durons but they operate on a Double Data Rate bus. What's QDR/DDR mean is that the ram is access 4 or 2 times per a clock cycle which means in theory its running as if it were a faster bus than it is.

As you can see the FSB can be set three ways.. I will only be covering the first two as I don't think the third option is what Trigem used.. however if you are reading this for another motherboard I cannot say for fact which option was used. In truth the only difference in option 1 and option 2 is in option 1 FS2 is set to logic 0 (false/off) and in option 2 FS2 is set to logic 1 (true/on).


Here is an image of the ICST 950201 PLL-IC Pin out:

Ok I now have a new jumpering setup that doesn't use the old traces and will make it a lot easier to set the various FSB's. Here's a picture of what a jumpering system for this might look like. (I drew this one extra big to make it easy to read.)

Blue = 0, Red = 1, Green = FS1, Purple/Pink = FS0


For this you will need 2 10K resistors. What you do is disconnect FS1 and FS0 from the motherboard as in the last one and then to the center pins of some jumper blocks. However this time in-between the pins and the jumper blocks you will use 10K resistors as shown in the picture. This time we will be using GND (ground) to set logic 0, so LEAVE the ground connected to the motherboard but connect a wire to it and then to one end of the jumper block. I used GND on Pin 47.. but any GND will do. To set logic 1 we will be using the VDDREF on Pin 1.. this outputs 3.3v. LEAVE VDDREF connected to the motherboard but connect a wire to it and then the last pin on the jumper block. Now all you need to do is set the jumper to set the FSB... this even makes it very easy to try for a 200MHz/800MHz FSB.. if you so wish to think you can get your machine to even boot at that.
07:30 AM EST - Revision 1.0 - Added new Jumpering design.
04:15 AM EST - Revision 0.2 - Added jumper diagram.
05:50 AM EST - Revision 0.1 -

Written at: 15:16 PM EST - Revision 2.0 - Got rid of old content and moved to one single mod.
Minor layout changes.
06:30 AM EST - Revision 0
Written on: Saturday, April 26, 2003 - Revision 2.0 - Got rid of old content and moved to one single mod.
Monday, December 09, 2002 - Revision 1.0 - Added New Jumpering design.
Sunday, December 01, 2002 - Revision 0.2 - Added jumper diagram.
Wednesday, November 20, 2002 - Revision 0.1 - Minor layout changes.
Friday, November 15, 2002 - Revision 0
Written by: Enigma Deadsouls
Website: http://www.geocities.com/enigmadeadsouls


Names/Info from people who have emailed me their test results:
Josh Koontz - 9 June 2005

Successful ICS950201 mod on Intel D845HV

I'm hopeful you still had interest in the ICS950201 success reports, but I apologize in advance if I was wrong on that assumption. So if it's only nice spam then it's just nice to say something positive to someone who actually took the time to actually think intelligibly about a problem, look up the PLL chip whitepapers and make a hypothesis with guide. Especially in this day and age where thinking has gone out of fashion.

Anyhow, bla bla. Here is an overclock the Intel d845hv and d845wn motherboard. Yes they are old and they are rather lackluster in any respect (SDRAM on a P4? enough said) but there are a TONS of these in service, mostly OEM remarks like
gateway and such.

The ghetto way I did this is to just to take the corner of a razor blade, cut PIN55 close to the motherboard and bend the lead to short with PIN54. This is a nasty nasty hardware hack that I'm sure is violating the voltage/signal tolerances on the PLL. But so is bridging Athlon bridges with a graphite pencil.

Also beware that Intel being the overclocking friendly folks they are didn't include a mechanism for the bios to figure out the memory clock divider, so you get the (hard coded?) 4/3 ratio. Hope your SDRAM likes 166mhz. Might be best to set the bios jumper to the funky "config mode" Intel has and retard all of the memory timings before doing this.

Also I'm not sure if you can fill all 3 slots since you're owning the 845 memory controller - at least not without a volt mod. I've been running well with 2 sticks of generic Hynix OEM pc133 cas3 - It runs prime95 for a few hours, that's enough for the office apps I use all day. So I consider myself lucky - Hynix must have been getting some sweet yields at the time.

In all for a simple cut and bend of a pin, I have no complaints. I'm still giddy it runs at all ;P

Enjoy.


Notes: Maybe soon to come a mod for the dc/dc converter used on the Trigem Imperial if stock voltage isn't enough for overclocking. If this works this project will have been much easier than my project on the Trigem Florida TG/TGA (and C might also work) which should be posted soon and I'll have a link to it here.