Weird and frustrating problem with overclock

[bububear]

Hello, I am getting quite frustrated trying to overclock my CPU again. I have 1 stick of OCZ 512 MB EL DDR PC-3200 Platinum Revision 2 RAM which I can overclock to (2.5-3-3-10 @ 2.85v) 281 x 9 = 2529 Mhz at stock CPU voltage and also stable at this setting. I then bought the OCZ EL DDR PC-3200 2x512 Platinum Revision 2 Dual Channel Kit, and my problems starts here. Before I continue, here is my system specs

MSI K8N Neo2 Platinum (Socket 939)
AMD Athlon 64 3200+ Winchester
Powercolor Radeon 9550
Samsung 160 GB HD
Seagate 200 GB HD
Hitachi 250 GB (on SATA 3 since it's locked)
Enermax 350 Watt PSU
Lite-on DVD +/- RW

I install the new RAM in DC mode (which is dimm 1 and 2 on this motherboard) and I run memtest86 v3.2 to find the max mhz which turns out to be 274mhz at 2.5-4-3-10, which is lower than my 1 stick at 281. Anyways, I know for sure my CPU can do 2529 Mhz at stock voltage, so I set it to 266 x 9.5 = 2527 Mhz. I turn on my computer, it boots into Windows XP with no problems, and I run SuperPi, and it fails. I run Prime95, and it also fails. I ran Windows memtest, and I get an error, which is really strange since I didn't get one in dos memtest. I use clockgen and set it to 240 x 10 = 2400 Mhz, and it still fails. I set it to 230 x 10 = 2300 Mhz, and it's able to pass SuperPi at 36s and doesn't get errors in Prime95. I tested each RAM individually in dos memtest86 and each RAM can do 281+ mhz in single channel but that's not the issue here. I tested my Dual Channel ram at 274 mhz at 2.85v and it passed without any errors in dos memtest. But when I run Windows memtest, I get errors almost instantly. And for some strange reason, when I have my new RAM (DC Kits) running memtest, I hear a really quiet squeaking noise coming from my RAM. I decide to pop my old 1 stick of 512 MB Platinum Rev. 2 RAM into my computer, and everything runs fine again on my original setting (281 x 9). So now I'm wondering, is my new RAM defective or something? Am I suppose to be hearing a really strange squeaking noise coming from my RAM?

I also tried running this in dimms 3 and 4 and I still get the same error, and also the strange squeaking noise.

 

[Acanthus]

someone else actually complained about sounds coming from the memory on load a couple days ago, i thought they were on crack, but this is the 2nd time

Running 2 sticks of memory can reduce your memory OC, its just the way the memory controllers operate.

I would try putting a fan on the memory and further increasing the voltage, have you done full tests at lower speeds to try to identify the problem?

You may fail at stock speeds if the memory is defective.

 

[bububear]

My voltage is set at max at 2.85v. I ran the tests down to 230 Mhz, and it seemed stable at this setting since it passed Prime95 and SuperPi. I just find it extremely strange that this can't even do 2400 Mhz (240 x 10) usind the DC kit. I have my LDT at the right settings to, so I know it has nothing to do with that.

 

[exar333]

you need a much better PSU if you even want to think about overclocking...350 wont cut it

 

[stevty2889]

I don't agree with you on the PSU in this case ExarKun, I have a 350w enermax, it has 26A on the 12v rail, I used it to overclock a power hungry 2.8Ghz prescott to 3.5Ghz, with my 6800GT and 2x36gb raptors..and it was 100% stable. Also using a 350w enermax with my 3.06ghz northwood, which is running at 3.45ghz.

bububear, haveyou tried running superPI and prime95 with your overclock with a single stick? It is possible that the memory controller is limiting your OC. Also, have you tried using a divider for the ram? If you check out most of the A64 OC posts, you'll see that running dividers on the memory has very very little performance impact with the A64, so you don't neccessarily need to run it at such high speeds...you could run it at lowers speeds with tighter timings..

 

[bububear]

I think I found the problem while running Windows Memtest; one of the sticks tops out around 267 mhz. I tested each stick out, and this one could not pass SuperPi at 267 mhz while the other stick does 281+ mhz with no problems. Funny thing is, dos memtest cannot detect this for some strange reason.