cant seem to get a stable oc

[fixxxer0]

okay so I have a E8400 stepping E0 with a Gigabyte GA-EP45-UD3P
I can run the chip at 4.3Ghz prime95 stable (small FFT's for 12+ hours with no errors) at 1.36v… temps under full prime load are just hitting 60C at 4.3 (on air).

I have G.skill 2x2GB 533mhz 5-5-5-15 @2.1v ram and I have been using the multiplier that gives me a underclock on it. I have scaled back my FSB/multi to 9x460Mhz for 4.14Ghz in an attempt to get some stability and lower temps, but even going down to 3.8Ghz doesn’t seem to help my problem.

I ran this ram at the SPD settings at full DDR1066 using stock multiplier/FSB/ghz, and it ran memtest fine for the whole night (~8hrs). However when I have my FSB cranked up to the 400+ range for the OC, I start to get p95 large errors, even with the ram underclocked (but i am not noticing any in memtest... does this mean NB problem?). obviously I am using a different strap for the higher OC but the ram is only running at like DDR2-900 with the same timings for 1066. sometimes I cant even boot with certain straps.

the proc on its own is stable and sometimes the ram appears stable on its own, but when im playing a game (using HD4870) it will lock up on me... this is leading me to believe that the problem is a bios setting or timing somewhere but I cant figure out what, im trying everything.

I have my Vterm - 1.3, Vpll - 1.5v, and MCH (northbridge) at 1.3v and 0.9v for Dram term.
I have to figure out how to copy paste the whole bios settings but basically I have all the speed step and LLC turned off.

any suggested things to try or check? Should I bother messing with the skew timings? I don’t even know what they do…

 

[kalrith]

You need to single out each step of the overclock: FSB, CPU, and RAM. To single out your highest FSB overclock, you lower the CPU multiplier to minimum, put your RAM at the ratio that underclocks it the most, and put the RAM at loose timings. Then see if you still get the errors.

Once you determine (and write down) the highest FSB with no errors, move onto your CPU. Leave the RAM settings alone, lower the FSB to stock, and put the CPU multiplier at max. Keep testing with Prime and increasing the FSB until you find your max CPU overclock. Write it down.

Next drop the FSB to stock and drop the CPU multiplier to minimum. Put your RAM at the ratio that results in the fastest speed. Then slowly increase the FSB while checking for memtest errors (doing just test #5 should speed things up, and that's where errors almost always occur). It's up to you how much time you want to spend on the memory, but you can test it for max speed at loose timings and max speed at tight timings if you want. Write it down.

Then, use all those findings to figure out what you think the best overall overclock would be. Once you get that set up, then run prime again and see if it's stable.

You might also want to run these tests at stock voltage and with the voltage increased a bit. The voltage increase itself could be causing the overclock to fail.

It's a lengthy process, but there's no way of knowing which aspect of the overclock (FSB, CPU, RAM, or any of the voltages) is causing it to fail without singling out each individual aspect and testing it on its own.

 

[fixxxer0]

thats the problem... im able to run 500+mhz fsb stable alone, and cpu alone up to 4.3Ghz and ram fine at 1066... but when i start doing all of the things together im getting problems.




im pretty sure it has to do with the FSB/NB... i think when everything is running the NB might be getting overheated or hitting some limit. is 1.3v too much? too little?

 

[Zap]

okay so I have a E8400
...
I can run the chip at 4.3Ghz prime95 stable

I don't get it... are you asking for help or are you bragging?

Try putting a fan on the chipset and boosting chipset voltage.

 

[fixxxer0]

I don't get it... are you asking for help or are you bragging?

Try putting a fan on the chipset and boosting chipset voltage.

asking for help lol...


i guess i worded it wrong... the CPU is stable at 4.3 but the ram isnt... it runs prime95 SMALL FFT for 12+ hrs without problems... but if i do combined test with large fft over the ram and stuff it starts to give errors fairly quickly.

whats the max voltage for the nb??

 

[fixxxer0]

alright i figure i should reply with what i beleive to be the cause of all my problems.

the horrible ATI win7 x64 drivers (specifically the Catalyst Control Center i think)... i finally used the BlueScreenViewer program to check out all the minidumps generated when my comp was crashing. turns out most of them were related to a few of these ati***.sys drivers.


i threw in the towel when i was trying to update the drivers and i was getting continuous BSOD on boot, i could only get into safe mode. i just said F it and reinstalled windows, and installed the driver ONLY (no CCC) - as i have read on some other forums.

i couldnt help but notice that now i am able to run a higher clock - 475x9 for ~4.3Ghz, w/ ram@1:1, tighter timings (5-5-5-14) with LESS voltage on the CPU and NB. and its stable in OCCT, Prime, Memtest, and real world useage/games.