Jump to content
Corsair Community

Asus K8V SE Deluxe and 2Gb TWINX2048-3200PRO - locks up occasionally


charliek

Recommended Posts

I've been running 1Gb of Corsair TwinX1024-3200LL for a while now, with no problems at all, but the time came to upgrade - so I went for 2Gb of TwinX2048-3200PRO (labelled XMS3200 v.3.2).

 

I installed them in slots 1 and 3 and let the motherboard detect them and run them at SPD, which was fine - BIOS reported 3.3.3.8 - my previous ram liked 2.7v which I left as it was.

 

I ran MemTest86+ v.1.6 overnight - 14 cycles with no errors.

 

In actual use, however, I am getting random lock-ups: WinXP Pro freezes completely, no mouse response, NumLock on the keyboard is unresponsive.

 

This happens maybe once a day or so, and at no particularly obvious time... running SiSoft Sandra burn in, Battlefield 2, Adobe Premiere, etc without a problem, then overnight it freezes while doing nothing-in-particular. This afternoon it froze with one instance of FireFox open.

 

I've tried fiddling with APM settings, re-seating the ram, removing an idle SCSI card, installing Cool'n'Quiet, upgrading every driver I can think of, even reinstalling Windows from scratch, but still hitting random lockups every so often.

 

Yesterday morning I put the old RAM back in, and have had no lock-ups since.

 

Clearly this isn't definitively pointing at the RAM (especially since it benchmarks with no problems), but it's currently my chief suspect.

 

Any ideas anyone?

Link to comment
Share on other sites

  • Corsair Employees

Please install the modules in slots 1-3 or 2-3 and set the Dim Voltage to 2.7X volts and then set the timings manually to the tested settings for the specific module you have Cas 3-3-3-8, and then test the module/modules one at a time with www.memtest.org! If you still get errors, please follow the link in my signature “I think I have a bad part!” and we will be happy to replace them or it! However, if you get errors with both modules that would suggest some other problem and I would test them in another system or MB to be sure.

.

Link to comment
Share on other sites

Please install the modules in slots 1-3 or 2-3 and set the Dim Voltage to 2.7X volts and then set the timings manually to the tested settings for the specific module you have Cas 3-3-3-8, and then test the module/modules one at a time with www.memtest.org! If you still get errors, please follow the link in my signature ...

 

 

Thanks for the reply RAM GUY :)

 

The modules were in slot 1 and 3 (I tried 2-3 also), the voltage was set to 2.7, and there were no errors in memtest, never have been - only lockups in XP

 

The SPD automatically sets 3-3-3-8, will manually setting the same values make a difference? If the modules pass memtest as a pair, will they likely show errors singly? I'll give it a shot anyhow.

 

The only reason I suspect the RAM is that the problem only manifests itself with these modules, and goes away when I put the other ones in.

Link to comment
Share on other sites

  • Corsair Employees

You might try them one module at a time and see if you can isolate one of the modules causing this. But I have no problem replacing them if you like.

Please follow the link in my signature “I think I have a bad part!” and we will be happy to replace them or it!

Link to comment
Share on other sites

You might try them one module at a time and see if you can isolate one of the modules causing this. But I have no problem replacing them if you like.

 

I shall have a go at that... getting replacements here in France can be tricky, and potentially expensive in postage (I went through the exercise here).

 

I'm testing the modules one by one at the moment, and will let you know how I get on.

 

Incidentally, I have taped my case's temp probes to the module's heat spreader, and I'm getting surface temperatures of 45-50 degrees C - is that in spec? Would the 2048-3000PROs be more sensitive to heat than the 1024-3200LLs? Having said that, if heat were the problem, I'd expect the lockups to occur during benchmarking, or playing games, not on a largely idle desktop, where the lockups appear to be happening at the moment.

Link to comment
Share on other sites

I shall have a go at that... and will let you know how I get on.

 

 

Just an update... I put one of the CMX1024-3200PROs in and, after a couple of hours, got a freeze. I've taken it out, and put the other one in and, after about ten hours have yet to get a freeze. I'll leave it in for a while, though, before coming to any conclusions.

 

This stick is running at the same temps as the other one, so I'm concluding that 45 degrees is peachy.

 

Do you have any idea what kind of a fault might cause a module to lock up the system after a while, but yet put in a clean 12hrs on MemTest?

Link to comment
Share on other sites

Let's get them both replaced. Please follow the link in my signature “I think I have a bad part!” and we will be happy to replace them or it!

 

Sounds like a good plan - the other one froze too.

 

I'll see if I can get them replaced by the local supplier, as I live in France - posting RAM to and from the US is likely to be costly and slow :)

 

I'll let you know how I get on.

Link to comment
Share on other sites

Please do let me know!

 

Well here's a thing... :)

 

The supplier here (materiel.net) sent me a Return Number without hesitation... Corsair seem to have chosen good suppliers here in France, as I had similarly good support from another of them, 1000ordi.fr

 

So... I packed up the TwinX2048-3200PRO, and put the original TwinX1024-3200LL back in. I packed up the RAM with a view to posting it in the morning. However, in the morning the PC had frozen, in the same way as before, with the previously-fine original RAM in! "Oh arse", as we brits say.

 

So, it would seem that my problem is elsewhere.

 

In the absence of any better ideas, I took the whole PC to bits, put it all back together again, and reinstalled XP from scratch - that was 48hrs ago, and I've had no freezes since.

 

The only significant change that I have made is to change the cooling on the VIA chipset (northbridge). Seeing as I had the thing in bits, I decided to re-seat the GPU and CPU waterblocks - while I had the cooling loop in bits, I decided to change the passive northbridge heatsink for a chipset waterblock that I had hanging around waiting for the next time I had the hood up.

 

The Via chip has a kind of design which looks thermally questionable: the circular metal contact surface (heat spreader) seems to be very slighly recessed in the square black chip package - meaning that the thermal interface material (I use Arctic Silver 5) has to bridge that gap. Removing the old heatsink revealed a fairly crap contact - blobs of TIM, rather than a uniform thin layer.

 

I'm wondering if that might have caused these symptoms? It would make sense particularly if 2Gb RAM places more strain on the northbridge than 1Gb of RAM would, thereby accentuating an existing weakness. It might also explain why the problematic 2Gb showed no errors in memtest (which makes no calls to HDDs, GPU or other hardware, and would probably leave the northbridge idle)

 

Would that make sense, or am I talking out of my hat (another thing we Brits say)?

Link to comment
Share on other sites

Since the memory controller is in the CPU, the extra strain usually felt by more memory is placed on the CPU as well, however, the lack of calls to any other peripheral (except a low dose of graphics) through the chipset could explain why Memtest didn't crash the system.

 

Sounds like you may have found the issue, it's a shame you didn't think to see just how hot it may have gotten, but probably not worth it to take off the block just to test how quickly you can let out the 'magic smoke' from your chipset! :D

 

I'd run a couple of burn-in wizards or PCMark or 3DMark from Futuremark.com and see if that crashes the system. Those heat the whole thing up pretty doggon well.

 

PLUR

CK

Link to comment
Share on other sites

Since the memory controller is in the CPU, the extra strain usually felt by more memory is placed on the CPU as well, however, the lack of calls to any other peripheral (except a low dose of graphics) through the chipset could explain why Memtest didn't crash the system.

 

The fact that the AMD 64 has an integral memory controller is one of the reasons I didn't look at the chipset before, and one of the reasons that I'm not wholly sure that what I suggested earlier really is the cure, as opposed to a charming coincidence :)

 

Having said that, the CPU is very efficiently chilled by my watercooling, such that it rarely hits 50 deg C even in burn-in/benchmarking, so I would not have expected any southbridge issues.

 

My limited understanding of PC architecure is that while the memory controller is now offloaded to onboard the CPU, the Northbridge still handles a deal of the communications between periperhals and RAM and, as such, would likely have to work harder (and get hotter) in an environment where peripheral/RAM communications was rich (i.e. WinXP daily use, but not memtest). Does that sound feasible?

 

Interestingly, my previous freezes were exclusively in 'mundane' windows use, web browsing or idle desktop, and were not provoked by my attempts to stress the system, benchmarking and battlefield 2.

 

I'm wondering if that is because both 3DMark and Battlefield will do the bulk of their work over the hypertransport and AGP buses, and will actually not stress the Northbridge much, and if perhaps the Northbridge is now chiefly occupied by mundane general usage: I/O, HDDs, USB, PCI peripherals, etc.

 

Again, I'm fully prepared to be wrong!

 

Sounds like you may have found the issue, it's a shame you didn't think to see just how hot it may have gotten, but probably not worth it to take off the block just to test how quickly you can let out the 'magic smoke' from your chipset! :D

 

I'd run a couple of burn-in wizards or PCMark or 3DMark from Futuremark.com and see if that crashes the system. Those heat the whole thing up pretty doggon well.

 

I did think of checking the northbridge temps, but was stuck for how to do it reliably - there is no on-chip diode (I don't think), and while my case has two temperature probes, there was nowhere sensible to stick one - short of taking the heatsink off and taping it to the chip. I felt that would be asking for trouble - my watercooled case has, as a consequence, fairly low airflow... as you said, a puff of magic smoke is not worth it, even to be proven right.

 

I'll run some benches, just out of interest, as you suggest - even though I am now well past the previous freeze interval, and benches weren't provoking it in the past.

 

I suspect I'll have to put it down to 'Just One Of Those Things', but many years as an IT Manager has left me with a pathological reluctance to 'just let it lie'. :)

 

Thanks for the feedback, mind you!

 

Charlie

Link to comment
Share on other sites

  • Corsair Employees

Please let us know how you make out, but it sounds like you may have it running at 100%. However I would just add that from past experience with WinXX from time to time during many thousands of installs I have ran into issues where I was not able to explain the problem, but upon re-installing all of the problems were no longer present. The only possible explanation would have to be some file or files were corrupted during the install of the O.S. and or driver and re-installing corrected it. With WinXP this should not be an issue from what I understand because it will check it self and automatically repair/repair files as needed. But considering that you have to do almost 100 MB or more of online updates anything could happen.

 

But bottom line if you have a problem with our modules even if we don’t think the memory is the problem we would not have a problem replacing them. For the life of your modules.

Link to comment
Share on other sites

Please let us know how you make out, but it sounds like you may have it running at 100%. However I would just add that from past experience with WinXX from time to time during many thousands of installs I have ran into issues where I was not able to explain the problem, but upon re-installing all of the problems were no longer present. The only possible explanation would have to be some file or files were corrupted during the install of the O.S. and or driver and re-installing corrected it. With WinXP this should not be an issue from what I understand because it will check it self and automatically repair/repair files as needed. But considering that you have to do almost 100 MB or more of online updates anything could happen.

 

But bottom line if you have a problem with our modules even if we don’t think the memory is the problem we would not have a problem replacing them. For the life of your modules.

 

Thanks folks... still no freezes, so I guess (touch wood) that the problem is resolved.

 

Thanks again for the support. Hopefully the next update you'll get on these sticks is when I'm looking for advice about replacing them with 12Gbs of QUADXtreme12288-38400Ultra for my new AMD256bit QuadCore Lightboard. Or something.

 

Cheers

 

Charlie

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...