Overclock.net banner

1 - 11 of 11 Posts

·
Banned
Joined
·
4,703 Posts
Discussion Starter #1
I'm starting to lose my temper, badly.

I reseated my heatsink, added some fresh OCZ Freeze to that and my GPU's, now my stupid memory won't boot at the correct settings.

Currently running memtest86, but I'm struggling to figure out why the hell my memory has decided to do this to me? I haven't moved it at all today (and the PC was just fine yesterday)

EVERY time I boot the memory either resets ot 667MHz, which is incorrect, or if I up the voltage to 2.1v and try go for 1066MHz, it has a fit and either resets to 667MHz, or won't boot (black screen upon boot)

I'm resetting my CMOS left and right and the main thing that is pissing me off is everything was just fine last night. I haven't even touched the memory today. MY GPU and CPU temps are great for the addition of the OCZ, but now this.

Help!
 

·
Loosing Vision
Joined
·
1,885 Posts
You may need to slow a bit. Last night I changed my Ram setting to get them to where they are suppose to be and I had several fails before I took it off of extreme and put mine to standard. I not sure if turbo on the ram settings will work yet for me.
But one thing can cause the others to change you config up a bit.

Good luck.
 

·
Registered
Joined
·
865 Posts
Your FSB frequency (according to your sig) looks like it is at 333 MHz. When your memory reverts back to 667, it's probably b/c your memory is strapped at 2.00C multiplier. The C designates 333 Mhz as the frequency strap.

Try this out to get your memory at 1066 Mhz.

You leave your FSB frequency at 333 Mhz. Set the memory strap to 333 Mhz if it isn't already. Set your memory multiplier to 3.20C (333*3.20=1066). Leave the voltage at 1.8 V and manually set your memory timings (5-5-5-15 or something like that).

That should let your memory work at 1066 Mhz but with a FSB to DRAM ratio of 5:8 I think.
 

·
Banned
Joined
·
4,703 Posts
Discussion Starter #4
Quote:


Originally Posted by Meta-Prometheus
View Post

Your FSB frequency (according to your sig) looks like it is at 333 MHz. When your memory reverts back to 667, it's probably b/c your memory is strapped at 2.00C multiplier. The C designates 333 Mhz as the frequency strap.

There are two ways for you to get your memory to run at 1066 Mhz.

1. You leave your FSB frequency at 333 Mhz. Set the memory strap to 333 Mhz if it isn't already. Set your memory multiplier to 3.20C (333*3.20=1066). Leave the voltage at 1.8 V and manually set your memory timings (5-5-5-15 or something like that).

That should let your memory work at 1066 Mhz but with a FSB
RAM of 5:8 I think.

I'm on a 780i motherboard, so I don't have access to my FSB Strap directly. I can set the FSB of the CPU and memory independantly, or link them (should be 5:4 for 1066mhz memory and 2.66ghz RAM @ 8x multi)

Originally, I had a similar problem and OCZ told me to up the voltages to 2.2v and knock the NB voltage to 1.4v and try setting the RAM that way. At the moment, it doesn't want to post if I do this.

Two sticks of paired RAM are currently 80% or so through the memory test without any errors. I have a feeling the other set will bring about some errors, hopefully not.

I could try going slow, but for some reason these sticks only ever liked posting and booting at 1066MHz, refusing to boot @ 800Mhz, sometimes. So it's 1066MHz, or nothing, it seems.

HELP!
 

·
Registered
Joined
·
865 Posts
Well the only other way to do it is the following:

1. Set your FSB to 533 Mhz.
2. Set your memory multiplier to 2.00D (D meaning 400 Mhz or higher)
3. If you can, set your FSB strap to 400 Mhz. If you can't I guess you can't. I'm not very familiar with the nForce boards.
4. Set your memory voltage to 2.1 V and some loose timings (5-5-5-15 for example).
5. Set your NB voltage higher. Set your cpu voltage higher. Set your PLL's higher.

That is the only other way I know of getting 1066 Mhz exactly without messing with the multipliers.

BTW, I'm not responsible should you try this for any damage whatsoever that may occur to your system.
 

·
Banned
Joined
·
4,703 Posts
Discussion Starter #6
Well, I'm memory testing all 4 sticks @800MHz at the moment.

Like I said, I can't access the FSB strap or memory multiplier. But CAS 5 is what they're rated for, so Im only using 5-5-5-15-2T and that's all I ever have used. I usually use 1.4v on the NB and 2.2v on the memory.

If I can get this to pass on 800MHz, I'm going to take it and run.
 

·
Banned
Joined
·
4,703 Posts
Discussion Starter #7
Ok, I freaked out when my CPU was at 54C and took the PC back into the dining room to work on. I had strapped some fans onto my TRUE and onto the case suspending them above my GPU's. This was clearly pulling my TRUE sideways and off my CPU a little, if that's possible. My temperatures are now quite normal and my PC decided to boot @ 1067MHz.

However, I'm now getting random blue screen errors. NPFS.SYS, SYSTEM SERVICE EXCEPTION, PAGE_FAULT etc etc.

Just performing a checkdisk scan has just returned error free.

memtest86 returned no errors. Anyway of checking the L2 cache on my CPU?
 

·
Banned
Joined
·
4,703 Posts
Discussion Starter #8
All the progams I'm running are crashing out. Firefox, everything.

This is so frustrating.

EDIT:

Apologies for the triple posting. After reading around, it could be file system errors? I'm going to boot into Linux and make sure it isn't a hard drive failure.
 

·
Banned
Joined
·
4,703 Posts
Discussion Starter #10
Running Prime95 with both GPU's in, cores 2 and 4 threw up this message:

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.


There is no stress.txt file. But all other cores work fine. With only 1 GPU in (the one I suspect is fine) only one core errors. What's going on? Do I need more voltage?

Very strange, I'm at stock settings and this seems to be an overclocking error. At least there are no BSOD's
 

·
Banned
Joined
·
5 Posts
I figured out what this error was. Basically, my CPU was running tests with known results and giving results that were wrong, thus a hardware error.

Very strangely, everything is fine now. Just passed 13 hours with no errors. Don't think my girlfriend will allow any more testing on her electric bill!
 
1 - 11 of 11 Posts
Top