WMV killed my hardware?? Strange issue...

[Xak]

Greetings all. I have fallen in love with my second DFI mobo, my SLI-DR. I quickly reached a sizable rock-stable OC, as described in my sig. The system gamed for days of a LAN, Primed for varying lengths of time from 8 - 22 hours numerous times, and I have been running the system 24x7 for over a month. Never a single error, BSOD, or glitch, ANYWHERE... until an hour ago when I got back from vacation.

 

After playing some THUG2, checking email, etc for an hour or two, I was downloading some videos and playing them back in Winamp via a playlist. Some of the files MPG, some AVI, some WMV. Some of the WMV files would cause the system to appear to lock up, even causing the mouse cursor to lag for several seconds, like a 100% CPU-usage real-time thread. I chalked it up to WMV being crappy, and simply waited for my system to catch back up (10 seconds or so) after sending the stop command to Winamp, and the system would return to normal. I deleted the offending file from the playlist and went on my way. This happened for several videos (all WMV), then the last one hard-locked for 10 seconds, then gave the famous BSOD IRQ_N_L_O_E. This is the FIRST BSOD or problem of any kind I have seen from running this OC for over 1 month 24x7! I hit the reset switch, and to my dismay, only got POST beeps and no POST, with 3 diag LEDs lighted. Memory, I thought. So first I tried pulling all power, clearing power from caps by pressing power button, then applying power, but same result. At this point, I pulled both DIMMs and re-seated them, same result. I then pulled out the module closest to the CPU socket (orange slots), leaving only 1 module in orange slot closest to edge of mobo. Finally, a POST. I enter the BIOS, and load my custom "stock" profile (timings manually set to stock, no OC whatsoever; 2.8v DIMM). POST'd again, ran Memtest for a few minutes, everything seemed ok. So I shut down and plugged the second DIMM back into the other orange slot, but this time, I didn't get any POST beeps, and the diag LEDs went from 4 to 3, hang for a second, then to 2, hang for another second, then back to 4, 3, hang... etc. I have subsequently tried optimized defaults (felt like a n00b when doing this :drool: ), and loosening timings manually from my "stock" profile, as well as my "Stable OC" profile, but they all provide the exact same frustrating result: I can NOT get the damn thing to POST with both DIMMs. I also tried the yellow slots, as well as varying DRAM volts from 2.6-3.0.

 

The really weird thing is, the system was running for 7 days while I was out of town (plus the time it had been up prior to that, probably another 5-7 days), and I was using it via terminal services while out of town, running flawlessly. The case is closed up, and NO hardware, BIOS, or software changes were made nor was it rebooted or shut down from the time I left town to the time I got back and encountered the detrimental BSOD.

 

The only POSSIBLE thing I can think of at this point is that I was messing with the PowerStream's adjustable rails before I read up on the proper way to do this (i.e. w/ voltmeter), but this was days before I left for vacation, then the system ran for all that time completely stable until I got home and tried to play those damn videos. Whatever the case, I don't have a voltmeter here, I will have to take it in to the shop tomorrow and set the rails all properly.

 

It certainly seems like faulty hardware at this point, possibly memory slot(s)/channel on mobo, CPU mem controller, or a module of memory (unlikely, as either module will POST, boot, and Memtest by itself). I know that the problem was *probably* not actually caused by a WMV file, but I try to blame M$ whenever I can

 

I'm going to bed, my head is fried and I drove 12 hours today, and I will tackle this tomorrow. Guess even though I have the day off I'll most likely be in the shop tomorrow (I'm a tech at a local computer store) trying to figure all this out.

 

Cheers and goodnight, I'll post my findings as they happen, and if anyone has anything they'd like to add I'd be interested to read it.

 

EDIT: Spellllling.

[Guest Dynamic]

So you ran this beast 24/7 eh? You crazy mofo, what for? So you said both ram posted fine individually and tested good in memtest? How did you proceed with the adjustable rails on the PSU the other day? The first thing i would do is try a pair of different ram and see what happens. I know you thought this because i can sense it, hehe. Hey don't feel so bad, you're hearing from a guy that tried modding for the first time and get no post, HAHAHHAHA! I hope you figure out if the memory is bad. Not sure if i understand the LED Diagnostics as of yet.

[Xak]

Well, I can't hardly believe it. I just got home with 2GB Corsair TwinX, popped them in and booted right up. Shut down, put the OCZ back in, and this time it did something I haven't seen before: The diag LEDs lit up repeatadly in sequence like a christmas tree!! Something definately very wrong with this RAM.

 

Good thing for OCZ's lifetime warranty, even though I haven't had them for more than about a month. I was running them at the OC in my sig at 3.0V, that's not too much voltage for these modules with active cooling, is it? Active cooling being a 120mm intake fan blown directly over the RAM and CPU (the PLUS accessory included with my Lian-Li).

 

Hopefully my boss will let me keep this set of TwinX on loan until my new OCZ arrives :nod:

 

Is this common for OCZ Platinum to go bad after a month of use?

[Reinvented]

Nope...not unless you get some cooling and stuff on it. Did you try running memtest on each of the sticks individually?

[zeimbo]

nice rig hehe. what are your video card temps with the zalman on there? i work at a computer store so i get everything at cost but im debating between the silencer 6 rev 3, the vf-700 or ordering the stasis cooler. maybe you were overvolting your ram? ocz is the best in my opinion, you can just feel from holding one of their products the quality and reliability of it.

[Xak]

what are your video card temps with the zalman on there?

Not sure exactly, I can't stand Speedfan taking 30 seconds to scan the BUS to pick up the video card, so I haven't seen the temps since I had the same cooler on my 6800GT. I ran it at above Ultra speeds with a 1.4V BIOS voltmod, and the temp peaked at 60C, which is not bad at all, considering that was my old behemoth of a steel case with poor airflow. Temps on the 7800GT in my sig from what I've seen is about 60C after exiting a benchmark as measured by the nVIDIA temp panel. I've seen the Silencers and the VF700-Cu, and I think the Zalman provides more heat dissapation. The stock copper sink on my eVGA card (the N517 model, 470/1100 stock) didn't cool it enough to even be 100% stable!

 

ocz is the best in my opinion, you can just feel from holding one of their products the quality and reliability of it.

I agree with regard to OCZ. My rig was not very stable at all until I upgraded my PSU and memory to OCZ's. The PowerStream is incredibly heavy, well-made, and I have done a lot of reading about how good they are. My opinion was the same on the RAM until I had this problem. I expect my new pair to OC as well as my last pair and not give another bad thought to OCZ. Bad parts are a fact of the industry. So far my RMA process has been a breeze. They put a $100 hold on my CC and are going to overnight me a new pair on Tuesday (due to the holiday). I expect a positive outcome of this whole incident.

 

 

A question for you zeimbo: I assume you are running your RAID-5 on the Sil 3114 chip. I really want a RAID-5 array, and hope to pick up 3 more 300GB Maxtor's, but have heard some really bad things about the CPU-based RAID on the SLI-DR boards. What is your experience? I am not expecting large performance, but I am attracted the storage space ratio and data reliability of RAID-5, as long as the performance to at least as good as a single drive. I figure that paired with a RAID-0 of Raptors for boot/usage drive will pair nicely with this large redundant array. Thoughts?

 

Cheers from a guy on a single-channel 1GB system running stock! =(

[Xak]

First let me apologize for the lengthy posts before... I can get carried away sometimes and want to outline every detail so people don't have to ask questions... :shake:

 

I received a brand new pair from OCZ's advance replacement. The folks there were very nice and overnighted the new modules to me to make up for the delay due to the holidays.

 

Before installing the new pair, I wanted to rule out the voltage rail issue that possibly killed the first ones. I bought a DMM and set all my rails to be spot on 3.3, 5.0 and 12. On a side note, they do not waver 0.01V from BIOS to idle to full load (Prime, 3DMark) :nod: OCZ PSU's are so pimp WHY AM I HAVING THIS TROUBLE WITH THEIR RAM? :confused:

 

What is the consensus on rail adjusting, to be spot on or a little over-volt? It also seems that my VCore is exactly what is set in BIOS (1.58) and VDIMM is 0.06 over according to my DMM. These do not waver at all during load either.

 

The new modules appear to suck quite badly. They will not OC to 205 MHz at CAS2, it has to be raised to CAS2.5. They also produced 2 errors in Memtest running stock the first time around, but after that first loop they do not produce errors (ran 3 more loops). This did not inspire me when I saw this. Does RAM need to burn-in???

 

I loaded my previous OC and raised to CAS2.5 figuring if they run stable here, I would be satisfied. Ran Memtest at this speed (exact OC in sig, just CAS2.5) several loops, no errors. Booted and ran Prime but it errored out after 3 hours.

 

Could it be that this new pair just sucks and I got lucky with the first ones (while they lasted)? Or should I RMA them AGAIN?

 

I will do more testing and call OCZ and see what they have to say about the matter. Any suggestions from those on the board are appreciated as well. Oh, and does anyone know what the max. supported voltage OCZ will warranty this RAM for?

 

DOH another long post... sorry! Hope everyone is having a good 2006 so far!

[Xak]

Just wanted to make a final post with how this all ended. I RMA'd the first set to OCZ and they sent me new ones, which would not overclock 5MHz at CAS2, and come to find out they would fail Prime at stock 200MHz after 4 hours. I contacted OCZ again, and not only did they replace them, they upgraded me to Gold PC4000 (actually a little cheaper on the Egg, but whatever) free of charge. The OCZ tech actually Prime'd them for me overnight to ensure they were good, overnighted them to me the next day, and included a paid shipping label to send the second set back.

 

The new modules run great. They are rock solid at DDR500 at stock timings, and are currently in testing at DDR533 3-4-4-8-13 1T. At these settings I get 7125/7050 MB/sec in Sandra memory benchmark! I know I know, bandwidth is useless in real world, but I'm enjoying pushing the OC spectrum at the bandwidth end, I was previously set on getting CAS2 as high as possible. Everything is still blistering fast either way :cool:

 

Many thanks to OCZ for their EXCELLENT support, and ven though first 2 sets failed I will gladly buy OCZ again. Another big thanks goes out to everyone here at DFI Street, you guys rock!

 

Happily running stable again...Cheers! =)

[nealh]

Many thanks to OCZ for their EXCELLENT support, and ven though first 2 sets failed I will gladly buy OCZ again. Another big thanks goes out to everyone here at DFI Street, you guys rock!

 

That is what I call support..go OCZ