And finally.... My system has arrived...

[DMB2000uk]

First off you should make sure that your CPU cooler is attached to the motherboard properly, it could be that it is slightly loose and causing the extra heat.

Did you find a proper cable from your PSU that has 8pins on the end? I presume you did, but just wanted to check that you hadn't forced a 8pin PCIe connector (or something) in there.

First culprit for crashes is the memory. Download and burn to a CD memtest86+. Boot to the CD and run the program. If there are errors that cause a system hang in 10minutes, they should show up in one pass (loop) of the program. That should take about 30 minutes to complete one loop. It will keep looping the tests forever, so after it tells you it has completed one loop you can exit the program and turn off your computer. If it errors, your RAM is bad, if it doesn't, post back here and we'll test something else.

Do you have access to a multimeter to test the voltage rails in the PSU?

Dan

[ibleet]

True enough! The ram is a likely culprit to crashes and blue screens. But that doesn't necessarily mean the ram is bad. There is also a chance that it may be undervolted, which would cause the same condition. Make sure your ram timings and voltage are both in range with mfg. specs (aka as advertised). After that, run memtest86+ overnight to check for potential errors.

My system kept having annoying unpredictable blue screen problems (at least twice a day), so I checked the ram. I was running the advertised timings and speed, but instead of 2.0v, the motherboard set it to 1.8v. I ran memtest and had 9 errors by morning. After a small voltage increase, memtest ran without errors for 8 hours and I haven't had a problem since...and that was several weeks ago.

As far as your CPU temps, what are the 4 core temps? Actually 38c isn't all that bad if you're using a stock cooler. For aftermarket air cooling, it should be more in the low 30's range, but depends on the case, fans, layout, cooling, etc... I would be more concerned with the core temps under load. What are those?

[DMB2000uk]

Good catch on the voltage levels, didn't think of that.

Dan

[martini161]

you have 4x1gb sticks of ram which is a configuration know to cause problems with many memory controllers (mobo chipsets) try adding a bit of voltage to the northbridge in the bios, and like the others said make sure its providing the right amount of volts to the memory. having all four banks filled puts more stress on the north bridge so it may need more volts

[vicaphit]

Sorry, I am not a system builder, so here is a free bump in the hopes that you get some help.

If the system posts then I assume that all of the parts are working. Try reinstalling the OS maybe.

[bandieramonte]

Okay, the help of all was really useful.

Actually, the memories are running undervolted and not at the mfg specs, which are 2.1 volts for this particular memories. So I will try to throttle their voltage levels to 2.1 and see if I still get crashes. If I still get crashes then what Dan say may be true; my memories could be damaged.

Now, since I'm so new this, I would appreciate some help on doing this tweak to the ram memory. I know how to access the BIOS, but what I don't exactly know is which particular variable must I change in order to get this memories to 2.1 V and 4-4-4-15 1T timings, because right now they are at 5-5-5-15 1T timings. My memories are this ones:
http://www.ocztechnology.com/products/m ... al_channel

So for a good start, I should do this as to improve system performance and stability.

Ibleet:
My Q6600 at full 100% load reaches temps average at 51 C. I really want to overclock to 3.2 Ghz at least. So I think that with these temps I could not achieve this. I will try to reinstall the GeminiII cooler. I just researched and I just put the thermal paste in the wrong way... and it was not a good one. I will buy the arctic silver 5 tomorrow. Many hardware things I still need to learn I see....

[DMB2000uk]

Your main problem could be that you are trying to run 4GB at T1. If it is indeed set at T1, then change that to T2. Even though the single kit is rated for T1, the northbridge will be straining to run two of those kits at T1, so they need to be run at T2..

You can however still run the timings at 4-4-4-15, so you need to change the CAS-TRCD-TRP-TRAS to 4-4-4-15 respecively.

Dan

[bandieramonte]

So, please correct me if I'm wrong.

All I need to do is to go to the BIOS, and change exactly these variables:
CAS , TRCD , TRP , TRAS to 4-4-4-15 2T and also the Memory Voltage variable (from 1.8V to 2.1V)

Do I have to leave the memories linked to CPU FSB? (I'm planning to also overclock the Q6600 to 3.2 Ghz or more)

[DMB2000uk]

Leave them linked for now, while we check that they are stable and working OK.

Dan

[Apoptosis]

So, please correct me if I'm wrong.

All I need to do is to go to the BIOS, and change exactly these variables:
CAS , TRCD , TRP , TRAS to 4-4-4-15 2T and also the Memory Voltage variable (from 1.8V to 2.1V)

Do I have to leave the memories linked to CPU FSB? (I'm planning to also overclock the Q6600 to 3.2 Ghz or more)

With a 680i you can unlink it and run it at what you want (ideal for overclocking). If you are running stock leave it linked, if you are running overclocked unlink it. With 4GB of memory stick to 2T as it will be stable. You can set the timings to 4-4-4-15 2T, but 4-4-4-12 2T should also be stable. You might want to drop the Trc setting also, but read this first - http://www.legitreviews.com/article/458/1/

[bandieramonte]

For now, since I'm more concerned with system stability, I will run the CPU at stock speeds and leave the memories linked. I will set timings to 4-4-4-15 2T and check for stability here. If it's stable, I should then try to overclock CPU.

The best way to check to see if the system is stable after each memory parameters' changes, is to run memtest86 and see if it has 0 errors?

[Methious]

No Memtest will only test the memory, to test total stability you need to run Orthos Stress Prime or Prime 95 for a length of time. I generally start out with an hour of Orthos, if your running a quad you'll need to run two copies of Orthos with the core affinities set to core 0 and core 1 on one copy and core 2 and core 3 on the second copy. if that runs for an hour I move on to Prime 95 for a couple of hours.

Big debate on how long to run prime 95 so I'll leave that up to others.

[bandieramonte]

Actually, before tweaking the memories I ran Prime95 and my system only resisted for 2 seconds... lol
I will try to run prime95 after doing the tweaks and I will run it during overnight. If by dawn my system is up, I think I can consider it stable enough and proceed to CPU overclocking. If not, I will post back here and see what can be done.

[ibleet]

Okay, the help of all was really useful.

Actually, the memories are running undervolted and not at the mfg specs, which are 2.1 volts for this particular memories. So I will try to throttle their voltage levels to 2.1 and see if I still get crashes. If I still get crashes then what Dan say may be true; my memories could be damaged.

Now, since I'm so new this, I would appreciate some help on doing this tweak to the ram memory. I know how to access the BIOS, but what I don't exactly know is which particular variable must I change in order to get this memories to 2.1 V and 4-4-4-15 1T timings, because right now they are at 5-5-5-15 1T timings. My memories are this ones:
http://www.ocztechnology.com/products/m ... al_channel

So for a good start, I should do this as to improve system performance and stability.

Ibleet:
My Q6600 at full 100% load reaches temps average at 51 C. I really want to overclock to 3.2 Ghz at least. So I think that with these temps I could not achieve this. I will try to reinstall the GeminiII cooler. I just researched and I just put the thermal paste in the wrong way... and it was not a good one. I will buy the arctic silver 5 tomorrow. Many hardware things I still need to learn I see....

Ok, 51c full load isn't bad. You probably have your cpu voltage set to auto, in which case, its getting much more voltage than it needs to run stable at stock settings. The Q6600 Go will most likely run stock stable at 1.15v, thus reducing your heat by a good bit. It will overclock to around 3Ghz at ~1.25v and run up to 3.6Ghz at ~1.35v. Your temps will still be fine with your aftermarket cooler.

Don't worry about keeping your ram at 5-5-5-15 for now. Just set it to 2T, and 2.1v, no more than 800Mhz for now. The looser ram timings help eliminate the ram as a problem, and you can tighten up those timings and overclock the ram after the cpu overclock is stable.

One step at a time or you will just frustrate yourself trying to find your problems.

[bandieramonte]

So, I should first of all set the ram timings to 5-5-5-15 2T and 2.1V to check if it's stable. If so, then I would proceed to overclock CPU, check if it's stable and finally I would then tighten the RAM timings?

In other words, I should overclock CPU before overclocking RAM timings to 4-4-4-15 2T?

[ibleet]

To be honest, I'm pretty sure your current problem is under-volted ram, so feel free to go right to 4-4-4-15, 2T...

Run Memtest86+ overnight if you're concerned.

[bandieramonte]

I have just tightened the RAM memories' timings to 4-4-4-15 2T 2.1V, and I have grown even more frustrated since then... The only variables that I changed from the BIOS were CAS , TRCD , TRP , TRAS and the memory voltage (2.1V).

I get more freezes now, the system freezes while window starts, when I click anywhere on the screen, sometimes I get blue crash windows and restarts, and sometimes the system resists about an hour.

To help you help me, I post this picture:

Dibujo.jpg

(208.43 KiB) Downloaded 19 times

I don't know what I'm doing wrong here. I have removed from my rig one of the GPU's, the sound card and every USB, fireware and IEEE1394 cables from the motherboard just to simplify the problem. I have double checked the correct installation of the rest of the components.

And about the CPU, the picture says that VID is 1.2874v while the core voltage is ranging from 1.12V to 1.26V, depending on CPU usage. When CPU is at 0% usage the core temp tends to 1.12V while temps reach is low to an average of 35C. The opposite holds true. I thought that core voltage was supposed to remain static. Are the CPU temps high?

I will really appreciate help here, particularly with my system instability.
Thanks.

[DMB2000uk]

Changing core voltage (and clock speed) is normal as this is part of intels speedstep technology to reduce heat and power when the CPU is idling.

What is the TRC setting in the BIOS? If it is at 23 like CPUz reports, try changing it to 30.

If that doesn't work then what I need you to do is take out one of the 2GB kits of RAM, leave the settings like they are supposed to be (i.e. what they are at now) and run memtest for a couple of loops (like an hour).

If that kit passes, swap them out for the other kit and repeat the tests again with the second kit.

Basically at this point we need to know if it is actually faulty RAM, or your northbridge not being able to handle 4GB (for whatever reason), or if the issue is something else entirely.

Dan

[martini161]

you have 4x1gb sticks of ram which is a configuration know to cause problems with many memory controllers (mobo chipsets) try adding a bit of voltage to the northbridge in the bios, and like the others said make sure its providing the right amount of volts to the memory. having all four banks filled puts more stress on the north bridge so it may need more volts

this may fix your problem, im just to lazy to type it all out again

[bandieramonte]

When you say a "bit" more of voltage, exactly how much more? I don't want to add more than necessary screwing up or maybe frying the northbridge.