Overclocking a 1.4 Tualatin CPU

[Edward2]

I just got a Celeron 1.4 GHz Tualatin CPU to use in one of my Folding rigs. So far I have been able to run the chip at 1.54 GHz (110 FSB) at default voltage and 39C load, and 1.63 GHz at (116 FSB) at 1.55 Vc and 40C load. I am going to let it run at this speed for a couple of days before I try to go any higher.

I doubt the chip will make it to 1.73 GHz (124 FSB, where the 1/4 PCI divider kicks in on the ST6), but I am curious how high it will go anyway. If I make it to around 120-123 FSB, what problems might I expect with the PCI bus running at so high (40-41 MHz)?

This chip has the following specs:
Philippines, 1400/256/100/1.5, 7224A769, SL68G

Thanks.

 

[breez]

SL68G? Does that mean that the chip is b1 stepping? If it is, you can expect it to overclock pretty well.

 

[Edward2]

No, it's a tA1 stepping chip. It is listed as pre-conversion on the Intel Product Change Notification document.

I'm hoping to reach 120-123 FSB (1.68-1.72 GHz), but I'm worried about the PCI bus running at 40+ MHz.

 

[airflow]

I have a Tualatin 1.2 in an ST6 and I've ran it at 120 bus using the 1/4 divider, so the pci was actually 30mhz. (this was after overcoming the initial problems I had using the 1/4 divider)

I'd also ran it at 124 bus with 1/3 divider, and it was totally stable, but I'd still rather be conservative for now. I'll aim closer to 1.6Ghz when I get a better hsf.

I would say just try it at 120+ and see if boots and survives. Video cards tend to get affected by a high pci, and the onboard sound stops working at around 37mhz pci, I don't know what else gets problems, hard drive maybe?

 

[Edward2]

Hey thanks Airflow. I had forgotten that the ST6 allows you to use the 4:4:1 setting at lower FSB speeds. I was thinking about my VH6T which does not have 1/4 divider until 124 FSB.

I have heard others mention having problems with the 1/4 divider on the ST6 mobo. What sort of problems did you have and how did you fix them?

 

[airflow]

The problem I found was that I would not even post using the 1/4 div. Just no signal to monitor.

I read though the posts and the suggestion I found that worked was posting at normal speed, go into bios, set bus to 133 and divider to 1/4, (I also set voltage to max), post again(hopefully), then go straight back into the bios and set the bus speed down to desired value.

I'm not sure why it works but It might have something to do with tricking the board into thinking the cpu is -supposed- to have 133 bus.

If you can't post at this speed you'll have to find another way....

good luck

 

[Edward2]

Hey thanks for the tip. I am currently trying 118 FSB (1/3 divider) with 1.575v and it appears to be running ok so far. It's only been about 10 minutes though.

(edit) Well, it locked up after about 15 minutes. Just of out curiousity, I tried to boot at 133 FSB with 1/4 divider and max voltage, but it would not boot at all. I am currently trying 120 FSB with 1/3 divider and 1.65v. Even if it is not stable at this speed (1.70 GHz according to MBM), I am quite impressed that it will boot and load Windows 98.

 

[deez]

What kind of temps do you have and what are you using for cooling?

Can you boot at 124? If temps are good enough then perhaps you can try a voltage mod

 

[Edward2]

Well after playing with it for a couple of hours, here is what I have found so far. Note that the ambient temp in my office is warm tonight about 26C and the case temp is 30C.

115 FSB, 1631 MHz, 1.550v, 42C, very stable
118 FSB, 1674 MHz, 1.575v, 44C, ran for about 15 min
120 FSB, 1694 MHz, 1.650v, 46C, ran for about 15 min
120 FSB, 1694 MHz, 1.675v, 47C, ran for about 15 min

I am running this chip in the 3rd rig in my sig. It has an AX-7 heat sink, ThermalTake Smart Fan II, and AS-3 for cooling on an ST6 motherboard. However, this is my wood case setup which probably does not help the cooling situation, but it is the easiest rig to make changes to.

I think if I can lower the CPU temps, it will probably help the stability (at least some). It is suppose to get a lot cooler later this week, so I will try again then. Hopefully I can make 1700+ MHz.

(edit) Obviously, I did not include data for all the combinations (FSB/voltage) that did not boot or make it into Windows.