New Server Build - Memory Module Issues?

[rezilient1]

Hi there.

I have a brand new server build with the following:

GIGABYTE GA-EP35C-DS3R Mobo
INTEL Core 2 Duo E8400
CRUCIAL Ballistix 2GB PC2-8500 (DDR2-1066)
EVGA e-GeForce 8600 GTS 256MB GDDR3
WD Raptor 150 GB SATA (WD1500ADFD)

Right from the start I had problems installing Windows Server 2003, but I chalked that up to either a failing DVD-ROM drive (its old) or bad media. Eventually I got Windows installed and started setting everything up, installing the drivers, etc. Things were going smoothly for a couple hours. Then when I tried opening IE6 my system crashed, and just POWERED DOWN. I checked eventvwr and it showed a lot of these:

Faulting application iexplore.exe, version 6.0.2900.2180, faulting module unknown, version 0.0.0.0, fault address 0×058f10db.

Faulting application explorer.exe, version 6.0.2900.2180, faulting module unknown, version 0.0.0.0, fault address 0xe8a10e2e.

etc.

So since its a new build and all Google searches about the above errors yield info about possible Spyware/Malware/Virus I figured it was probably hardware.

Ran MemTest86, it hung on Test 2 and showed tons of memory errors for Test 2. Tried each memory module one at a time, same thing, fails and hangs on Test 2.

Ran WinDiag, it is still running Pass 6 as I type this but its showing FAILED on nearly every test.

So my question is, what is the likelyhood that BOTH memory modules are bad? I would think pretty slim. I am thinking its the MOBO but any chance its the CPU? Are there any other tests I can run to isolate or pinpoint what parts I need to return? Does software like SiSoft Sandra do any error logging?

Thanks.

Reza

 

[rezilient1]

Also worth mentioning, in the Award BIOS all the memory and CPU settings are left at Default/Auto so I haven't touched any of the timing or overclocked anything yet. Maybe I need to tweak that?

 

[RadiclDreamer]

Unlikely that both are bad. Have you tried giving a slight bump to the ram voltages? It could also be that the memory modules are incompatible with this particular board.

 

[rezilient1]

I have not touched the voltage yet, need to try that, still left at default settings. I will try that first and adjust the timings, then if no success I will try updating the BIOS too. This memory doesn't show up in the official supported list, but I've seen other people using it with this board.

Thanks, I will let you know.

 

[mechBgon]

In the big picture, when building a home server of any sort, I'd lean towards high-quality "vanilla" RAM (e.g. Crucial's regular modules) rather than high-strung stuff like Ballistix. Stability trumps memory bandwidth any day in a server.

 

[rezilient1]

But I can always underclock the module to 800mhz right?

 

[mechBgon]

Originally posted by: rezilient1
But I can always underclock the module to 800mhz right?

Yes, insofar as that goes. It might be a good idea to run some Memtest86+ testing to check for error-free operation, just to be sure it's happy.

 

[bwatson283]

mech you are so awsome.................i bow to you :notworthy:

 

[mechBgon]

Originally posted by: bwatson283
mech you are so awsome.................i bow to you :notworthy:

Never mind the bowing, I want pizza!

 

[rezilient1]

Ok so I tried this last night.

I bumped up the voltage to +2 (2.1v) and then +3 (2.2v) and then +4 (2.3v) and ran memtest86 again each time, was still getting the same amount of errors. Voltage reading is from the Award BIOS "Health" reading. Then tried to adjust the memory frequency down to 800mhz (was "AUTO" set to 1066mhz) and was still getting the errors. Also messed with the timing, no luck.

Downloaded the latest BIOS from Gigabyte called "F2" (or ga-ep35c-ds3r_f2) and kept the settings at 800mhz, auto timings, and auto voltage. SUCCESS! It ran Windiag and Memtest86 clean, with zero errors.

So if anyone is having this same problem, seems like there are issues with this RAM and the F1 BIOS, so try upgrading to the latest BIOS.

Now a follow up question.. My memory timings are currently 5-5-5-15, t2 @ 2.2v and I am running the memory at 1000mhz, 2:3. Does that sound fairly conservative? I want the thing to be stable so not trying to push it, but I also don't want to hurt performance either. So I'm looking for a good "middle of the road" setting. Also what about all these "advanced" timing settings in my BIOS? Should I just leave these at AUTO?

ACT to ACT Delay (tRRD)
Rank Write to Read Delay
Write to Precharge Delay
Refresh to ACT Delay
Read to Precharge Delay
Static tRead Value
Static tRead Phase Adjust

I ran Prime95 last night for 30 minutes, allowed my CPU cores to reach about 70 degrees (is that too high for these 45nm procs?) and it was rock solid stable.

Thanks!

 

[RadiclDreamer]

For a server I would leave the timings on auto.

As for the temps, is this 70F or 70C? If F then they are fine, if C then they are pretty high

 

[rezilient1]

That is celcius. On idle I am seeing about 45 degress celius at the cores, but 70 was pushing it with Prime95 for several minutes.

 

[RadiclDreamer]

That is a bit high, especially since its a 45nm unit. Try removing the thermal compound and cleaning the heat sink and reapply.

 

[ThePiston]

I have same Ram, mobo and basically same CPU (Xeon e3110). I upgraded to the F2 BIOS before I even thought about testing - luckily. That ram is awesome. What bus speed r u using?

 

[rezilient1]

I'm just lost here...

So now when I run memtest86 and Windiag I don't see the failures after 1-3 passes. Except after the computer has been running for a while I start getting BSOD and application faults again. So I left Windiag running all night with both chips. Passed the first few passes, then failed from like Pass 7. Tried one chip at a time for several passes, no problems. Tried again with both chips in different banks, works for a while, but again when I leave it for several hours it failed.

When the memtest was left over night, in the morning when I see the failures I hit "x" to exit and then I see a screen that says "Recovering BIOS from hard drive" and "Restoring BIOS" or something to that extent. And then it reboots. Does that recoving thing again. Reboots. And then finally boots up, back to the default BIOS settings which I have to tweak the memory voltage and freq again.

CPU is idle at 45-50 celcius, mobo is in the 40's. Stressed the machine using Everest for 2 hours, was sitting at 67 celcius most of the time but stable. The BSOD only seems to occur when I leave the machine running for a bit, or when opening up several applications.

What is the deal here, am I looking at bad ram? mobo? cpu? How can I isolate this?

Thanks for any help.