E7400 Overclocking Advice

[Lord-Ilpolazzo]

Hi,

I have an E7400 based computer, having heard these things overclock well I have been toying with a modest overclock from 2.8GHz (stock) to around 3.3GHz which from looking about I reasoned should be easily achievable. I have the stock heatsink fitted, with no intention of buying anything exotic to replace it hence the "modest" part. Appologies in advance for the wordy explanation that follows, I reason that with sufficient detail there is less room for misunderstanding!

The major components are:

E7400, Gigabyte EP-43-DS3L Motherboard, 2GB of cheapo generic PC6400 RAM (I have no intention of overclocking that!), 8600GTS, and a 500W OCZ Stealthxtreme PSU.

I have tried two different overclock configurations so far, in both cases I have intentionally left speedstep on since I reason the computer spends much of its life good as idling, its nice to have it cool down and slow the fan a little when its not doing much. the first was:

-FSB 320MHz, multipler 10.5x, MCH Strap/Multipler 2.5A, ("A" being the 266MHz strap) Vcore = Normal (1.2625v)
This gave me a 3.36GHz for the CPU, with the RAM running at its normal 800MHz. I did all the usual run prime95 for ages, and it passed those tests fine (max temp I ever saw was 61degC), so I used the computer that way for a month during which it worked fine save for couple of unexplained crashes. Seeing as that had never happened before, I decided to review the overclock...

For the second attempt, I did some more reasearch and my new understanding is that the MCH strap you choose ( A=266MHz, B=333MHz etc) defines the timings used between the processor and MCH. Lower MCH straps result in tighter timings. Is my understanding of this correct?

So, based on this I set the FSB to a round 333MHz, and chose the 2.4B strap (2.4x memory multipler, strap B=333MHz). Which brings the memory to a correct 800MHz. I reduced the processor multipler to 10x to get my desired 3.33GHz. I re-ran the Prime95 stress tests overnight, and these came back fine. Maximum temperature was unchanged at 61degC. So I will proceed to use this configuration and evaluate it over the following months.

Out of all this, I have two questions:

First, is my understanding of the MCH strap situation correct?

Second, I have observed that Vcore is not quite what i would expect it to be:

At stock:
Vcore=Auto, Observed 1.232v under full load in cpuz
Vcore=Normal [1.26250v], Observed 1.232v under full load in cpuz
In both cases, the voltage drops as the multiplier drops because of speedstep.

In both the overclocked configuartions above:
Vcore=Auto, Observed 1.3625v in cpuz. This dosent really ever change much, even with speed step enabled
Vcore=Normal [1.26250v], Observed 1.216v under full load in cpuz (dropps lower than at stock 2.8GHz). Voltage varies with multipler due to speedstep
If I manually set any other Vcore value to compensate for the droop, I seem to loose the speedstep thing; the mutlipler changes but the voltage does not. why?

So two things concern me, first is that with the overclock at its present settings the Vcore is actually lower than it is at stock! It appears stable so far, but should I be concerned about this? Would you recomend I manually increase Vcore to bring it back to at least the stock 1.232V under load? Second is that even when i manually set a Vcore value, it is shown as expected in cpuz under light load, but droops considerably at full load.

Any advice would be much appreciated

Thanks

 

[japps2]

Lord,

I'm confused...how did you change the multiplier on a locked chip? On this chip you can only change the FSB frequency to OC. But, to answer you questions: when you set the voltage manually you cannot use speedstep anymore...since the purpose of speedstep is to lower the frequency (by lowering the multiplier) and lower the voltage draw...you basically short circuited this by setting the voltage manually.Don't be concerned about the voltage being too low...you will get BSOD if it gets starved for volts...too high a voltage and you will fry the chip. As for your last question...sorry, I don't know

 

[Lord-Ilpolazzo]

Hi,

Thanks, thats helpfull to know. It just concerned me a bit that the voltage sagged so much! I've settled at leaving it set to normal so that speed step works properly for now.

I changed the multiplier down from 10.5x to 10x as part of the overclock, rather than increasing it which as you say is not possible. Leaving it at the stock 10.5x would have given me Fcpu = 3.49GHz which was a bit more than I was after! so I reduced the multipler a tad to get back to my original 3.3GHz or so.