OC help please - e2180, P35, Xigmatek 120mm

M

MrHood22

Supreme [H]ardness
Joined
Aug 21, 2007
Messages
4,312
Hey, I just got the setup in the title and i'm wondering about OCing.

I've read the 2 guides on tweaking/ocing but I haven't really done them step by step since i know how to OC (for the most part) because i had my AMD 4000 running at 2.74ghz (1.3v) for a while.

I have it at 1.5v now and 3.3ghz but it fails after like 3 minutes.

I tried 1.4v 3.2ghz and it fails after 45 minutes.

Any idea how I can increase north/south bridge volts on the P35-DS3L?

Thanks.
 
Oh Btw. I have another question.

Is it possible for a CPU to have a sweetspot?

What I mean is, could it be possible for a CPU to "prefer" to run at say 3.0ghz instead of 3.2ghz. That it could actually perform better at 3.0ghz instead of 3.2ghz? Keep in mind i'm assuming that it's 100% stable at both frequencies.

Thanks.
 
3.0-3.2 GHz is the typical top end for E2xxx series chips. Some get up to ~3.4, but it's very rare. I think you're just expecting too much out of it.
 
Yeah Silent. I try to aim high on a lot of stuff.

Right now i'm gonna try to let orthos run overnight @ 3.0ghz, 1.45v (1.4v in CPUZ).

Sometimes i'll have temps in like the 55c's then all of the sudden it will drop and jump around 40c-47c for a few minutes and then shoot back up to the 50's. Does that mean that the temp is dropping or does it mean that Core Temp is playing tricks on me?

Any word on that "sweetspot" theory? i hear some people say stuff like "Yeah, my chip has a sweetspot of 3.3ghz so I just keep it there" and I can't really tell if it's just a type of expression or if there literally is a sweetspot of CPU's.
 
Sweetspot with reference to OCing is pretty straight-forward -- there's a point of (often sudden, and massive) diminishing returns.

For example, with my X3220 (much like a Q6600) I can get it in to Vista and stable ~2+ hours (I always cut the test) at 3.85Ghz. The problem is, this requires far far more voltage all around -- to the CPU, to the northbridge, to the southbridge -- than 3.5-3.6Ghz and produces far more heat. Where I can kept at ~62C or less full load on all 4 cores at 3.6Ghz, temps get well in to the 70s at 3.85Ghz. Loaded vcore of around 1.5 is enough for 3.6Ghz, but 3.85 requires... well, I don't even want to consider what it might be doing to my chip, and I don't want to test it long-term.

You will notice a point where the vcore required suddenly takes a big jump in a ~500Mhz range. The idea is that you stay on the low end of the range, and have less heat to deal with as well as a potentially longer lasting chip. Also, ~300-500 Mhz either way with these chips does not result in a big performance differential -- anything above ~3Ghz with current software is pure 'gravy' -- so it's just not worth beating up your equipment for.

With the DS3L's fairly weak cooling your main concern is going to be keeping low NB/SB voltage and low FSB.
 
Well I just ordered 2 40mm fans that i'll just twist tie onto the NB/SB.

So basically the sweetspot theory means that you get to a point where the volt increase:performance increase ratio isn't good enough to do the extra performance increase so you should just stay at the lower voltage, sound about right?
 
You must log in or register to reply here.
Back
Top