Ballistix Tracer Problems

[Hardcore]

Current System:

E4300 OC'D 3.2ghz water cooled @ 1.45v 400x8
P35C-DS3R
4x1gb Crucial Ballistix @ 800mhz 4-4-4-12 2.2v
8600gt
500gb sata wd
620w corsair power supply


Problem: I am getting blue screens and programs that have "stopped working"

Cause: Either bad ram or ram is overheating somehow...

Memtest: Generates hundreds of errors

Things I have tried:

1. Replace what I thought was the bad kit of ram with new tracer's.

2. Replace the power supply with corsair 620w

3. Put 120mm delta 90CFM fan blowing onto ram with side of case off. Ambient temperature in my basement is always below 18 celcius. Most often 15 celcius.

4. Tried running 5-5-5-15 timings with stock cpu and slowest possible mem speed and everthing in between. Memtest still generates errors as quickly.

5. Lowered ram voltage from 2.2v to 2.1v then to 2.0v. Board will NOT boot with 1.8v selected.


Basically a run down. Put the computer together myself. It ran fine for about a week, then I noticed I was generating errors in windows. I thought it was the power supply because I was getting hard drive errors as well. Replaced that and at the same time, I replaced what I thought was the faulty kit (tested with memtest).

New components the computer ran fine for about another 1.5 weeks or so until it started to generate errors again. I'm thinking it is a heat issue. I have check the heat spreaders on the ram and they are all fastened securely with no gaps. Like I said, the delta fan is blowing directly on the ram, so there is no possible way these ram chips could be cooled any further.

I leave my computer on for days on end without shutting down. Right now I'm shutting the computer down and will try running memtest from a cold boot to see if it is a heating issue.

Other thoughts on the problem would be greatly appreciated. Could it be my motherboard that is corrupting my memory over time?

Thanks in advance.

P.S. I will update when I have completed memtest from a cold boot.

 

[Hardcore]

Update:

I have check the ram heat spreaders again. 1 of the 4 chip's spreader is not secured perfectly. The tape doesn't seem to be sticky and won't hold the it securely enough. Looks like I might have to RMA yet another set of this stuff.

Lowered the voltage to 2.1 to see if I can keep them a little cooler.

 

[ashmedai]

Those are specced for 2.2V, right? (Pretty sure they can take lower, but need that for the rated speed.)

Try using Memtest on one stick at a time (or, removing the one that's visibly different first to see if it's the culprit). If removing a specific stick fixes it, you probably have bad memory. If you get bad results with only that stick in, and good results with a different stick in, it pretty much excludes the motherboard (bad socket or something like that can happen sometimes).

Loads of things can cause memory issues including (in rough order of decreasing probability) BIOS settings / heat (often ones that "go away" mysteriously), bad memory (fairly common, so always use memtest on new memory), bad motherboard parts (not so common, but sometimes you get one badly soldered memory socket or something like that, memory works fine one at a time but problems happen if a specific socket is populated), bad CPU (rarely just broken enough to cause memory issues but still turn on), power issues (again, rarely just right to cause just memory issues).

 

[Chilly]

When did you buy them?

I was comparing my older sticks to the ones I just bought and the new ones pcb is thinner, and the leds are slightly smaller, and the metal bracket that held the heatsink is different. looks theres been a revision, I'll take out my old sticks and see how well the new sticks clock, thou I threw them(new ones) in and system seemed to work and boot fine with 4 sticks at 400(800) at 3-3-3-7 at 2.2 volts,

I'm having some problems with the 4 sticks, but from what I see from google its a common issue with 4GB plus 8800 and higher on Vista x64 (people seem to think its caused by the 8800 because other cards work fine). perhaps this "bug" effects all 8x00 series cards? (you didn't mention what OS i don't think)

Perhaps this new revision is for the worse? Whats the batch number (orange sticker) on your ram?

 

[Chilly]

Update:

I have check the ram heat spreaders again. 1 of the 4 chip's spreader is not secured perfectly. The tape doesn't seem to be sticky and won't hold the it securely enough. Looks like I might have to RMA yet another set of this stuff.

Lowered the voltage to 2.1 to see if I can keep them a little cooler.

oh and on the tape thing, you need to hold. Secure from the bottom and do the whole bottom row at one, it will stick (same heatsink problem with my one of my new sticks)

 

[ashmedai]

#&(^#@*)!!!

Okay, I started playing around with a similar memory kit (okay, exact same thing) I picked up for a project that keeps getting stalled. Results: all four sticks test out fine individually. Any combination of two sticks tests out fine. Four sticks gives errors, leading to a massive scrolling page of errors after a prolonged period. After this happens, I remove the extra sticks (once I can touch them without getting burns) and try it with 1 or 2 again. At this point, it won't boot upon trying to reboot, the lights on the memory won't even turn on. Giving it a few minutes to cool makes it boot again, but with massive scrolling errors for 1 or 2 sticks. Letting it cool more makes 1 or 2 sticks go back to zero errors. The case I'm working with has ~4 120mm fans going, and slapping a floor fan next to the mobo doesn't prevent this from happening (probably slows it down, I wasn't timing it). Not enough space between the adjacent modules for the air flow to help, or too much heat being pumped into adjacent components, something like that. I'll have to hash this out with Crucial on Monday barring some miracle. But unless this is some glitch in the batch, the answer is probably "ha ha, no 4GB for you, suckah!"

Note: Having an error only with all slots populated can also be a motherboard issue. I'm excluding this due to the "memory won't light up and boot fails upon reboot attempt" symptom, which pretty clearly looks like a heat issue.

Another note: While this computer is intended to be running Vista 64 + an 8800 GTX, it's being tested while an older video card is installed, and using a bootable copy of Memtest86+. So neither is a factor, in my case at least.

 

[ashmedai]

Well crap, the modules still eventually overheat and stop working with only 2 slots populated. It takes longer, for what that's worth (supporting heat as the culprit).

 

[(cf)Eclipse]

After this happens, I remove the extra sticks (once I can touch them without getting burns)

i don't expect those sticks to last more than a month. drop the voltage, bump up the airflow. it's really the only answer in this case

for reference, my ddr2-1066 sticks only needed 1.85v to be stable at stock voltage. 2.2v definitely isn't needed, so just explore the low end of what works

also keep in mind that 4 sticks is also harder on the memory controller. bumping up the mch voltage or dropping the mhz down will definitely help some too

 

[Hardcore]

Heres another update. I think Eclipse is right on the money.

I lowered the voltage from 2.2 to 2.0v. The ram boots ups fine and so far no errors in memtest. I definitely looks like this is a heat issue. Which leads me to my next question....


How the heck are you supposed to be able to run this ram at the rated speed without it overheating? I can guarantee there's not many cooler basements than mine, running without the case side on with a 120mm deltra high cfm fan blowing directly on the ram.

Is it because the ram is spaced so close together?

Running this ram at 2.0v @ 4-4-4-12 @ 800mhz just doesn't seem to be doing it justice.

 

[Chilly]

Heres another update. I think Eclipse is right on the money.

I lowered the voltage from 2.2 to 2.0v. The ram boots ups fine and so far no errors in memtest. I definitely looks like this is a heat issue. Which leads me to my next question....


How the heck are you supposed to be able to run this ram at the rated speed without it overheating? I can guarantee there's not many cooler basements than mine, running without the case side on with a 120mm deltra high cfm fan blowing directly on the ram.

Is it because the ram is spaced so close together?

Running this ram at 2.0v @ 4-4-4-12 @ 800mhz just doesn't seem to be doing it justice.

I'm running 4 sticks at 3-3-3-7 @ 800MHz (400) at 2.2 Volts, and the sticks only get warm(well, REALLY warm, but not burning hot) to the touch. I have a 50CFM 92mm fan pointed at them from above (ie hidden in the top part of the case so that its not visable) so there not even hetting te full benifit of the fan. I don't know what to say other than, I was having crashing problems and etc as well, but once I reinstalled Vista x64 with the 4gb installed, everything has been fine.

To be on the safeside thou I will probbly drop them down a little bit (to 4-4-4-7 at 2.0 volts like you or something) since eclipse's cmment about only lasting amonth scares me .

 

[Chilly]

Heres another update. I think Eclipse is right on the money.

I lowered the voltage from 2.2 to 2.0v. The ram boots ups fine and so far no errors in memtest. I definitely looks like this is a heat issue. Which leads me to my next question....


How the heck are you supposed to be able to run this ram at the rated speed without it overheating? I can guarantee there's not many cooler basements than mine, running without the case side on with a 120mm deltra high cfm fan blowing directly on the ram.

Is it because the ram is spaced so close together?

Running this ram at 2.0v @ 4-4-4-12 @ 800mhz just doesn't seem to be doing it justice.

Okay, I just came back from some quick testing, I'm currently running at 1.85v 800MHz @ 4-4-4-7 no problems, all 4 sticks, 1.9v also works fine for 1066MHz @ 5-5-4-7, again I haven't tried anything else, and it was only quick tests (full cpu & ram strress prime 95 for like a minute) and I have nio clue what long term stablity would be like, BUT the ram runs much cooler now! now I'm gonna try and see what I can get out of the sticks at 5-5-4-7 at 2.0 volts

 

[(cf)Eclipse]

running without the case side on with a 120mm deltra high cfm fan blowing directly on the ram.

Is it because the ram is spaced so close together?

i have a 120mm delta at 7v for my ram. the ballistix get warm when i push it, and the slots are far apart on my board:

 

[ashmedai]

It seems we can rely on Crucial to both produce memory that can be significantly overclocked, and to overstate the necessary voltage.

Running 1.9V let me get 4x1GB through test 6 without errors (it had pretty consistently shown a few hundred there at 2.1 to 2.3). I'm still waiting to see if it can make it through two passes successfully, but that looks promising. I've been using 3Dmark to check if the crashing / "fails to reboot" issue is still there. If it passes Memtest, I'll be moving on to that.

The memory is still hot to the touch, but I think it's a bit cooler (subjective, going from how long I can touch it). Should I still worry, if it turns out to be otherwise okay?

Update 1: First pass in Memtest completed with no errors. Also, the forum is sloooow...hopefully the rampant server problems stop soon.

Update 2: It seems stable enough at 800 (passes multiple loops of Memtest, can be rebooted after running 3Dmark), although it still does this funny thing where it tries to power on several times (after hitting the power button once) before giving the happy beep that means it's decided to actually boot. Haven't seen that before. I can't get 1066 to work - it wants a bit more than 1.9 to get past the first boot screen, but that makes it overheat. Something's probably still getting RMAed...actually, I'm a little worried that the overheating memory may have damaged the motherboard itself somehow. There's an uncomfortably electric sound sometimes when it's failing to boot, which seems to be coming from around the memory.

 

[ashmedai]

Well what do you know, Crucial cross-ships.

Now to convince myself not to buy a Q6600 while I'm waiting...