2500+ overclocking multiplier or FSB?

[540mb]

Ok, I have a 2500+ and it runs completely stable with a multiplier of 13. Gives me 2.16 ghz. Should I keep it as is or go back to 11 for multiplier and bump up the FSB? I know bumping up the FSB will give me more memory bandwidth but I just have pc2700 so I'd have to run at a ratio to get the FSB real high. With the multiplier at 13 I have to run the vcore at 1.7.

 

[BlueWeasel]

Well, your RAM is probably going to be the limiting factor, but quality PC2700 can hit 200 FSB. It will probably require a voltage bump of the memory to 2.8v or so.

I'd lower the multiplier to 11 and start bumping the FSB up in increment of 10 (increase the RAM voltage as necessary for stability).

 

[Concillian]

Yeah, it depends on the memory, but since you are unlocked, find out what your memory can do:

1) download superpi, memtest86 and prime95
2) lower the multiplier to something very low, like 9
3) increase FSB about 3MHz at a time... when you boot into windows, run superpi at 2M digits.
4) If superpi fails, then either increase latencies, or drop down 3 MHz and call this your best achievable RAM speed... run the other programs to insure stability. Prime95 blend for 8-12 hours is probably the most robust test.
5) if you get to 200 then all is well, most nForce2 motherboards won't go much further.

SuperPi is a very good test for a first pass. It's VERY quick, takes like 2 minutes to run 2M digits, and it will usually error out ver early on if there is an issue. This allows you to hone in on what speed your memory can do at what latencies pretty quickly and still take small enough steps that you aren't likely to create a situation where you can't boot and have to clear the CMOS.

I have 2 sticks of PC2700.
First is a Kingston Value RAM that comes up at 3-5-5 if I set the BIOS to use the SPD latencies. It runs 200 MHz at 2.5-3-2 all day long, even at only 2.6 volts (the lowest my MB will do)
Second is OCZ PC2700 2-2-2 v 3.2 It will do 2-2-2 up to around 180 MHz and will do 2-3-2 at 200 MHz

So I'm 2 for 2. You may get lucky and have good PC2700 sticks like I did. You could strikeout too, who knows. No matter what it's a time consuming process finding exactly what your memory can do well. Especially because memtest and Prime95 take hours, and they're the only way you can really ensure stability at a given speed/latency.

You may want to start by lowering latencies one at a time at PC2700 until you either can't boot or superpi fails to find how well your memory can do PC2700. Often it's better than rated, even for value memory. I bet if you have a reasonably recent DIMM, it will do PC3200.

 

[540mb]

Thanks guys for the replys. I appreciate the info.

 

[Manzelle]

I am going to try the same thing...

I run an AXP-M 2600+ @ 2.6GHz (200x13) but I would like to try my hand at a higher FSB...I wonder if 220x11 would be faster or slower than 200x13...

 

[Concillian]

THe idea is that after you find what your memory can do you then increase the multiplier is where your processor was before, but now the memory is running faster.

Manzelle, that would likely be slower. 216x12 on the other hand...

 

[trexpesto]

216x11.5@1.775V all day long, but Manzelle has low voltage magic. Good Luck everyone!