Can a motherboard kill RAM?

[xenolith]

I have a weird problem. I built an all new system based on an ABIT KX7-333R 4 month ago. When I first got it, I put in a new stick of Mushkin 512MB PC2700 DDR. The RAM failed right away at the end (@ 98%) of test #5 of Memtest86 with no o/c. No problem, had it exchanged in 3 days with a good stick, thank you Mushkin. The new stick passed all tests even with a slight o/c, and thought nothing more of it.

4 months later...

I'm racing along at Daytona in Nascar Racing 2003 (great game BTW) and it locks up hard, I hit the reset button. I thought no problem... shiznit happens. A little later I'm racing down-under... Australia... in F1 2002... short time in the race system locks up hard again. OK, the system never even hiccuped for 4 months and all the sudden this...?

I checked the obvious, dead fans, hot parts, loose cards, etc... then Memtest86. Yup, the RAM failed at the exact same spot of the test... the end of #5. And when I mean failed, I mean over 200 thousand errors!!! But again, only at 98% of test #5. Weird.

Could the motherboard be ultimately responsible for the RAM going bad?

 

[nemo160]

yes
i had a motherboard actually burn traces off a stick of ram

 

[xenolith]

Just curious, did you jack up the volts on that stick nemo? I'm using the default 2.55 volts.

 

[lobadobadingdong]

I've seen lighting fry just the powersupply and the ram before several times, and once or thrice seen the line jumps from lightning be just enough to fry the ram alone, that's why you don't carry them over carpet without anit-static bags. Very sensitive stuff.

 

[LiLxJohnny]

Ditto, my motherboard had a problem with providing "clean" voltages/power to the cpu and memory and ultimately fried both of them and not too long later the motherboard died too. I later learned that the problem may have been the capacitors b/c Abit used some faulty capacitors on their boards but since mine was passed it warranty period, I couldn't have it RMA'ed. I could only RMA the CPU (retail athlon) and Memory (Lifetime warranty).

 

[nemo160]

that was at default voltage, i think it might have been the gf4 i was testing for a friend though

 

[xenolith]

Update:

Boy, I'm finding Memtest86 is a great diagnostics tool...

I found it's not the RAM. On a whim, I reluctantly increased my T-bred's voltage from 1.5 to 1.65 and the errors went from over 200K to 4. I increased it to 1.7v and I get 0 errors. If it was bad RAM, increasing the CPU voltage would have no effect in the test's outcome. Could it be a bad CPU?

Ah, But...

Should i rule out the motherboard? I went back and set everything back to default speeds, and sure enough, I get over 200K errors at the end of memtest's test #5 again. What would happen if I increase the memory's voltage just a little to 1.65v even when the CPU is at default speed and voltage? It's hard to imagine, but increasing DDR voltage made it worse... I escaped out of memtest after the errors past 500K!

I don't think DDR voltage would have any bearing on the CPU's L1 or L2 cache, or would it? If not, then I can confidently rule out the CPU and finally conclude it's a bad motherboard... or am I way off...

 

[Johneverd]

What brand of power supply are you using? What is the wattage?

 

[xenolith]

Dual fan TTGI 420W PSU, 210 watts on the +3v & +5v rails. More than enough for only an AMD 2000+, GF4, NIC, 1 DDR, 1 HDD, 1 CDRW, and 3 case fans.

 

[xenolith]

Winbond HW Doctor sensors read:

CPU Vcore: 1.71v
+3.3V: 3.49v
+5V: 4.87v
+12V: 12.05v
DDR V: 2.58v
Aux V: 1.5v
-5V: -5.10V
-12V: not available
Standby V: 4.85v
Battery V: 2.99v

The +3.3V line appears a little high while the +5V line is a little low. :disgust: But, the BIOS only has a +3.3V setting of 3.5 or 3.6 to select from, no 3.3 or 3.4. Why would a mb not have a 3.3 setting for the 3.3V line? I also read somewhere that most PSUs have the +3V and +5V lines on the same rail, and if you increase one line you will lower the other line. So, because the +3.3V has to be 3.5v, it may be contributing to system instability by causing the +5v line to be low.

Regardless, the system appears to be stable now since I increased the Vcore to 1.7v w/CPU at default speed as stated earlier. I'm just thinking of ditching this mb soon and getting either an ASUS or Epox NF2 or KT400A mb if they prove to be better.