Opteron 148 Memory Overclocking with DFI SLI-D

Ok, guys I'm new to AMD, but so far I have my Opteron 148 at 3005, here are the settings:

CPU Mult: 11
HTT:273
LDT:4x
Memory at: 166
Memory settings: Auto on everything

The cpu does the small FFT (Max FPU stress) prime 95 test all day long, but when I got to blend (test lots of ram), it throws an error 10 mins in.

Here is the memory:
http://www.newegg.com/Product/Product.asp?Item=N82E16820146845

Can I assume the chip is stable with blend, or possible could the memory controller on the chip causing the problems? Just not sure what part is not stable from here.

What part is unstable, if the memory is runn at 333 and still the ram prime 95 test is failing?

Thanks
Indyxc

try LDT 3x and a tad more dram voltage

You need to drop LDT multi as Rattle said for preliminary testing.

I lowered the LDT to 3x, and it's still giving me trouble.

The problem is I can't get it to pass the Prime 95 blend test, but it passes the FFT test no problem.

With the Prime 95 test the max stable in the Blend prime 95 test.
HTT 250
LDT 3X
Memory 1:1

And then around 2800 it begins to throw errors with the memory at 1:1 in the Blend test in Prime 95. Yet, no problems with the FFT test.

Next I tried setting the memory to 100, so low speed on the memory, and I run the blend test with the CPU at 2800, but it fails after about 10mins. So I try the FFT MAX FPU stress test and it passes it all day long. It even passes the FPU test at 3000mhz.

So the situation is the chip passes the FPU test at any speed, but NOT the Blend Prime 95 test at any speed over HTT 255, NO matter what speed the memory is running at.

I'm still not sure what's unstable at this point. CPU, motherboard, Memory. I browsed around and no one else has had an odd problem like this.

I'm running without the IHS, and with a massive watercooling setup. The load temps are around 36C, with a max jump occasionaly to 38c.

Only things I can think off:

-I'm running x64 Windows, which might be causing some problems.
-Feed the chip more voltage
-I accidently chipped the corner of the CPU ever so slightly. It's not noticeable, until you look really close. I don't think this is causing problems, because it would probably not run if it was damage at all.

Where should I go from here?

Thanks
Indyxc

A64 CPUs are more susceptible to Prime95 type errors because the memory controller is part of the CPU itself. Just because the error occurs on a memory intensive test doesn't rule out the CPU like it would with an Intel setup. Your max HTT of 255 MHz is also most likely a CPU characteristic. I've had two 146s, similar steppings that run at 3 GHz in the same mobo with the same DDR and settings. However, both will only run stable at 10 x 300 MHz while only one of the two will also run at 9 x 333 MHz. Troubleshooting these types of issues is much more complex on an A64 setup.

Thanks for the Advice Dave.

It did end up being the cpu was unstable, so I got just the cpu stable, but am having other problems. Being working with it daily over the past week. These AMDS are a challenge.

Problem is I can't get the computer stable with the the following settings:

CPU: 3005 mhz
HTT: 273
LDT: 3x
Mem: 1:1

It fails the Prime 95 blend, 896k iteration 3rd Test every single time, 51 mins in. Never fails to fail. Oddly, if I run just the 896k blend test on it's own, it passes. Anyway to figure out which part is unstable, I isolate each component as follows:

So I isolate the CPU stability, and slow down the memory:

CPU: 3003
HTT: 273
Mem: 166 mhz (slowed down)
LDT: 3x

It passed 6 hours of the Blend test. I quit.

Next I isolate the Memory, and slow down the CPU.
CPU: 1000 mhz
HTT: 273
Mem: 1:1
LDT: 3x

It passed 3 hours of the blend, and then I quit.


Moral of the story, is individually they pass, but wont' pass together. LDT and chipset voltage are 1 bump above stock.

Load temps are 41c or so.

Any ideas on what I can do?

Thanks
Indyxc1