e6400 overclock issues

[c_loke]

Hey Everyone,

Just wanted to get your opinion on what I should do next.
I have an overclock set at
FSB: 415
VCore: 1.375
VTT: 1.32
Northbridge: 1.38V
Ram voltage: 2.2V
Ram timings: 5-5-5-16
FSB: DRAM: 4:5

Temps: idle 40,38 load, 56,57

I've ran P95 small ftt test for 12 hours and it passed fine. However when I run the blend test, it always fails. The best I have done is about an hour before failure. I have tried bumping up the voltages for the northbridge and ram and I cant seem to get the blend stable.

I'd like to keep the FSB at what it is right now and try to get things stable that this speed. What do you guys think I should try?

 

[limitedaccess]

FSB : DRAM 1:1?

 

[c_loke]

I did more testing, with the ratio set to 1:1. It didnt seem to help much. I dropped it back to 400 and it was failing (never use to before when I had it previously OC'ed), seems to be quite happy at FSB 375.

Its an older conroe processor. Not sure how much more I can push out of it. Since I passed small ftt's, and failing on the blend, its safe to assume that the issue is not enough memory/nb volts or memory not able to operate that speed?

 

[limitedaccess]

Can your motherboard drop multiplier? If not you could use RMclock or something to drop multiplier on your CPU, that way you can test if the limitation is FSB, Memory or CPU. This is would be the best way to rule things out.

 

[BillParrish]

You are on the right track I think as Blend tests the CPU to NB to Memory by including memory in the testing. So NB, FSB voltage (Vtt ?) and if you have it - skew settings for memory or Northbridge is where I would be testing. FSB voltage (Vtt) can be a double edged sword sometimes as you are messing with the voltage the devices on the buss see as a logic one or zero.

If it where me I would set ram timing to 5 5 5 18 (16 maybe a typo ?) and ratio to 1:1 (for now) back off 5 MHz or so and get it rock stable. Then taking notes I would reduce NB voltage until it failed, if it failed, and take note of that value. Add NB voltage one step at a time and test until stability came back and note that voltage. With NB voltage at the lowest stable setting do the same with Vtt.

You should then have a baseline of NB and Vtt for solid stability at 410MHz FSB or thereabout at 1:1.

Now start increasing you options one at a time, one step at a time taking notes at a 415FSB and see if you can zero in on what helps and what does not. If nothing helps try 413. Don't forget the skew if you have it.

I hate recommending just piling on more Vcore but at 3.2 you are right where the "normal" OC of these chips starts to peter out and you have to start really piling on the voltage. If your cooling can handle it also increase Vcore to 1.4 as you do the above testing, since you pass small FFT there is a good argument that more Vcore is not needed but it might juice up the drivers/buffers on the FSB I/O of the chip and make that tiny little bit of difference.

Sorry, all kinda standard, "screw with it over the course of several weeks taking careful notes " advice I always give. Not xexy or quick but usually will work .

A side note, you video card is already OCed (exact same as mine BTW) but if you are a gamer and looking for more performance, even 5 or 10MHz more out the the video card is worth the testing time. (here I recommend using your game(s) as stress test as running them is the goal anyway. ) And if you get disgusted and decide to upgrade on a limited budget the vid card would be the first thing IMO the new 1GB 260 looks sweet once the price comes down from initial frenzy.