Oldskool dual

viili · Post by **viili** » Fri Sep 23, 2005 4:33 pm

Hi! I've had this dual board Tyan Thunder 100 for some time with dual P3-450's, but got a very tempting offer of 40++ euros for the procs and sold them at instant

Now the board has 2x celeron 300A PPGA's on gigabyte GA-GR7 slotkets @504MHz (these procs can do 600MHz default voltage). The boards seems no good for 133MHz fsb (It's a rev.2 440BX)

Anyhow, I've got 4x P3-600E's and thought of putting two of those in this rig. Though I modified the slotkets to work with coppermines (reset pins) they don't even boot dual. (Individual procs work fine with either slotket). The P3's are cD0 stepping. I guess more modding on the slotkets is needed.. Help/experience any1?
Edit: just my bad english..

Dave Rave · Post by **Dave Rave** » Sat Sep 24, 2005 1:59 am

take out the edit ..
if you hadn't said u bad, I'd never have noticed

viili · Post by **viili** » Mon Sep 26, 2005 1:25 pm

Had some time to tinker with the soldering iron.. The article about dual coppermines on BP6 was helpful, my slotkets just needed the BR1-signal routed to N33. Now it's up and running, but even at 100MHz fsb burnBX (somekind of chipset testing program) fails. Maybe a fan on the chipset or slight bump of 3.3V-rail would help.. I'll post about that later, for now 6x75MHz @ 1.30V is enough (passive cooling of course

)

Dave Rave · Post by **Dave Rave** » Mon Sep 26, 2005 2:11 pm

the 3.3v rail is for memory and/or vga card.
shouldn't need to play with that.

post us what the std voltage and fsb is and then where you are and I'll tell hyou, if you overclock you will need to up the voltage to the cpu.
you might get away without, but I'd bet it's not stable

viili · Post by **viili** » Mon Sep 26, 2005 5:31 pm

I know quite a bit about BX-boards, stuff like the BX-chipset requires 3.3V to work, and that on this board the 3.3V rail is used directly without additional circuitry to generate it for the northbridge. And memory, too.

The processors are P3-600E with default voltage 1.75V. Tried overclocking these individually on a Abit ST6-Raid and 1GHz (166MHz FSB) was no sweat without raising voltage. So I could bet it's not the processors' fault.. Now the dual system is prime stable at dual 500MHz (6x83MHz). I just don't like the thought of 40MHz++ PCI bus on my server..

On to the memory.. 4 sticks, total of 640MB, 2x256MB + 2x64MB. Mixed PC100 and PC133. Latencies set at 3-3-3 at the moment, so not too tight. I could try dumping the small 64MB sticks, see if it helps..

My main suspect is the northbridge. It's an early stepping BX from 1998. Also it has a heatsink glued on that's thinner than a regular greenie. And the northbridge is located right below the heatsink of the first CPU. And it can't run @133MHz FSB. BOOTs, but freezes. Tested with PC133 CL2 mem and multiplier unlocked P2.

I know that later made BX-chips overclock better. I have two extremely well overclocking Abit BX133s, one that could do 199MHz with a huge heatsink and fan on the northbridge, though needed 4.15V voltage (!!). The northbridge on that one was made in mid-2000. Actually, it would be kind of interesting to try swapping the northbridge, put better overclocking one there..

No right equipment or courage for it yet, though.

viili · Post by **viili** » Mon Sep 26, 2005 6:38 pm

With two 256MB sticks dual 600MHz seems stable. Rated speed, w00t

Dave Rave · Post by **Dave Rave** » Tue Sep 27, 2005 3:29 am

have you thought of FSB versus rated memory speed ?
FSB > 100 and memory only PC100

or does BIOS allow controlled back to pc100 regardless ?

viili · Post by **viili** » Tue Sep 27, 2005 4:21 am

The BX doesn't have a memory divider.. it's always 1:1 FSB:MEM. My system is now at 100MHz, memory timings at 3-3-3. These two sticks have been running just fine at 140MHz bus for two years in another system.. The system had frozen overnight when I left it running two instances of prime. D'oh.

I think I have to put the 300A celerons back for now..

viili · Post by **viili** » Tue Sep 27, 2005 5:56 am

Tried taking one of the P3s out of the machine -> stable at 112MHz fsb. With the other P3 same result so no faulty processors or slotkets. Both back in -> freeze. Dual P3 seems to stress the system much more than one. Readings from voltage sensors are OK though all the time..