Computer Crashes?

[simple_inhibition]

run a couple instances of 3D Mak 05 and 06 and look for artifacting, or if the program crashes to the desktop, then you know youve gone too far. also if you want to stress out the GPU do a google search for rthdribl 1.2 and run that for a while

[gvblake22]

Yeah, rthdribl heats up the GPU faster than anything else I have seen; it's a good one

Have you tested your RAM with Memtest86+ to make sure it is stable? That might be a good thing to do if you haven't yet so you can knock out the possibility of the RAM being the overclocking bottleneck.

I am just going to go ahead and be honest with you, I think your RAM needs some tweaking. I think that's why your overclock is failing. You have some very specific/picky RAM sticks that will only perform well under very particular situations and settings. Can you give us some more information on your RAM timings? :thup:

[fps justin]

kk thanks. RAM timings are still stock. i belive that is 2.5-3-3-8. Uh changing the fsb made it slightly more unstable for prime 95. Crashed at 2 hours 15 minutes.

[gvblake22]

Yeah, your RAM is designed to run really tight timings but at the expense of extremely high speeds. Those aren't TCCD chips under the heatspreader on your OCZ VX
They are a type of IC that do not usually scale to high speeds very well, they are much better at running very tight (2-2-2-X) timings with high voltage (3.0V - 3.2V). You are going to have to run a pretty hefty divider on your RAM if you want a high overclock. Here's what I want you to try:

Drop your HTT speed to 200mhz and put the divider back on 1:1 so your RAM is also at 200mhz. Then set your RAM to 2-2-2-8 (1T) timings with 3.2v and run a couple loops of all tests of the lastest version Memtest86+. If it passes, then run Prime95 and/or some other stressful benchmarking applications (like any 3DMark, Aquamark, and/or rthdribl) and see how everything does. If all is stable, then start increasing the HTT speed in 5mhz increments (keeping the 1:1 divider). First test with Memtest86+ and then P95 or other. Make sure you run the full suite of testing at each setting change so you know if things are still stable. Just keep going until something becomes unstable, then report back and let us know what you found.

Also, are you changing your LDT multiplier to compensate for the increasing HTT frequency?

[fps justin]

yeah.

I have a problem i cna't drop the divider lower than 9/10. The next step is 5/6 is there sumthing i have to do inorder to drop the divider again? It doesn't load windows if i drop the divider further.

[gvblake22]

No, I think you are not understanding me clearly...
1:1 is the same as 10/10 or 200

[fps justin]

no, i get what your saying. I got my problem earlier. My 5//6 divider is like broken or something. At anyrate, for the memory test that i am about to do, do i leave the LDT multiplier to 5.0?

[Tyreal]

The LDT multiplier, when multiplied with your FSB should be around 1000 or lower

So choose the LDT multiplier from there. Keep your divider at 1:1.

[fps justin]

alright. Then i'll change back to 4 after 220 fsb. (if it ever gets that far.) This test takes AGESSSSS

EDIT: Could i just find someone with the similiar rig and same memory and just steal their information?

[gvblake22]

Originally Posted by fps justin

...for the memory test that i am about to do, do i leave the LDT multiplier to 5.0?

When you are testing at 200mhz HTT, then yes, you can leave it at 5x. Otherwise you want to set it to 4x as soon as you start overclocking for the reason Tyreal mentioned, you don't want to the HT bus to go faster than 1000mhz and be out of spec.

HTT X LDT = HT bus speed

So...

200 X 5 = 1000mhz HT bus speed (default)
205 X 5 = 1025mhz HT bus speed (bad because it is out of spec)
205 X 4 = 820mhz HT bus speed (good because it is not out of spec)

So you will need to keep the LDT multi at 4x until you reach 250mhz HTT frequency (250 X 4 = 1000mhz HT bus speed). After that, you will need to drop the LDT down to 3x.

Originally Posted by fps justin

EDIT: Could i just find someone with the similiar rig and same memory and just steal their information?

Maybe, but components are going to perform differently in every system. You can use someone else's overclocking experiences as more of a guide, but it is very possible that an overclock possible on one person's system will not be the same as another person's system; even if components are very similar.

You are just going to have to be patient and take your time when finding your highest overclock. Personally, I find myself tweaking and testing for months sometimes until I can find my best settings. So take your time, keep a little overclocking log of settings you have tried, and just go slow and test with every change of settings.