Mem error in Corsair TwinX1024-3200C2PT

I have a major mayday. I built a Shuttle 61G4 a couple months ago and use a G of the above named memory. Most of the time things are fine, but sometimes I get sudden, inexplicaple crashes, or just spontaneous reboots, or black screen memory dumps. Last time (yesterday) I sent the thing to Windows afte rthe reboot and it gave me a link to a memory test which I ran. Result was that on a test called STRIDE 6, twice with cache off and once with cache on, I got the following problem. at 26d3c0d8 i am supposed to have ffffffff and i actualy have fff7ffff. now, i have not much of a clue what this means except that there is a problem with the memory. Can you (a) explain what is happening and (b) tell me how to get Corsair to fix or replace the bad memory? thanks, silo

[quote name='RAM GUY']First this is not really the right part of the forum, so I am moving it to Compatibility Questions! Second, please try and set the Dim Voltage to 2.7 Volts. If you still have the problem, please test the modules one at a time with [url]www.memtest86.com[/url] and let’s make sure one is not failing! If it still will not post, please follow the link in my signature and we will be happy to replace it![/QUOTE] RG-- (a) How to set the voltage as you describe? (b) will run memtest86 as you suggest. (c) "post?" What means? (d) machine boots and runs fine most of the time, just get occasional spontaneous reboots, and similar. (e) I believe it is certainly that one of the modules "is failing," without doubt. (f) is the methofd for checking which simply pulling one and running the memtest, then pulling the other and reinserting the first and running the test? thanks, silo PS--Sorry to have put Q in wrong part of forum.

[quote name='RAM GUY'](a) How to set the voltage as you describe? A. I would check the user manual for the Dim Voltage setting or ask the MB or system maker! Sorry, I dont know! (b) will run memtest86 as you suggest. (c) "post?" What means? Pre Operating System Test! A. When the bios first loads and you see it counting the memory and such, this is the POST! (d) machine boots and runs fine most of the time, just get occasional spontaneous reboots, and similar. A. Please try what I suggested and we can make sure! (e) I believe it is certainly that one of the modules "is failing," without doubt. (f) is the methofd for checking which simply pulling one and running the memtest, then pulling the other and reinserting the first and running the test? A. Yes, more or less, but try to find the Dim Voltage! Usually it will be where you set the CPU speed and voltage![/QUOTE] Ram Guy-- I made a bootable CD with memtest86 (vers 3.0 i think) and ran it on my machine. Got lots of errors, mostly on the first pass, in different locations. Then I found VDIMM Voltage in my BIOS. I hope this is the right one. It was set on AUTO. I assumed it was what you wanted and reset it to 2.7 volts. Now I am going to run memtest86 all night or through several passes to see what happens. Will report soon. Thanks for the guidance. Silo

[URL=http://www.publicscrutiny.net/ramtest.htm]pictures of ram failure[/URL] Have similar result and have photos (640x480) of screen here Would like Corsair to send new memory sticks as soon as possible. Am tired of messing with this, if that's OK. fb

[quote name='RAM GUY']I would check the PSU with these type of errors. If possible can you test the modules in another system?[/QUOTE] PSU? THese are ram errors, are they not? Should I call somebody at Corsair? THis is a matched pair of 512 MB chips. It doesn't matter which one has the bad ram, BOTH must be replaced. You are supposed to have a Lifetime Warranty and I am asking for replacement under that warrany. I _could_ look at every possible cause, but it seems to me these are ram errors, as demonstrated by the mem test, and therefore the ram should be replaced. If I am wrong about this, I welcome being set right. If I am NOT wrong, I welcome the replacement parts ASAP. I would appreciate a telephone number or e-mail address so I can take this issue to the next level. Thanks.

[quote name='RAM GUY']I would check the PSU with these type of errors. If possible can you test the modules in another system?[/QUOTE] I'm curious why you think these are power supply issues? This is a brand new Shuttle ST61G4. Shuttle SilentX 250, the world's most advance SFF power supply. First introduced in the XPC ST61G4, the SilentX provides increased output, lighter weight, and much quieter operation than previous power supplies. It is also available as an accessory to upgrade previous generation XPCs, compatible with 13 models (Beginning with SS40G)! Developed under secrecy for over a year, the SilentX has proprietary high-efficiency design, providing more power with less heat output. Airflow and intake/exhaust vents have been optimized, resulting in less resistance and reduced turbulence in the airflow. Also, the power supply chassis has an all-aluminum design, significantly reducing weight. Specs [URL=http://us.shuttle.com/specs2.asp?pro_id=432]HERE.[/URL] I don't mean to be belligerent here, but shouldn't your ram work in most stock computers without excessive fiddling? That's why I'd prefer to have you send me replacement ram, and I will then send you the ram in my system that failed last night as indicated in the photos I linked to above. I cannot pull the ram and send it to you first as I have to have the computer functioning at all times for my work. If this is not possible, I suppose I could go to Circuit City or someplace similar and buy new ram, but then would Corsair assure me that they will refund my cost on this ram (I'd have to check it but I think I paid $208 at ZipZoomFly in January 2004)? Thanks again for you patience. Silo

I get these msgs on Sudhian: -------------------------------------- 1f2f I agree that it might be his power supply but i think it's most likely the ram. You should be able to run memtest without getting a single error. If you get even one error, it means that every now and again, as in if you get 1 error in 100 cycles, once every year or less your ram would fail, causing a reboot. open up the box, look on the side of your memory for an exact model # or whatever, and go online to Corsair's website. Find the exact recommended timings (CAS latency, TRas, etc) for YOUR Ram It should be in the form 2.5 4 4 8 or somehting like that. Go into your bios and change your timings to exactly match this. You have to select EXPERT in a couple of places to change this. Then run memtest86. If you still get errors, then i suggest you borrow some memory from a friend, test it out in your computer, run some Prime95 or folding to try to get it to reboot. If it does, then it wasn't only your ram that was causing it, even though you should get new ram anyway. Think about getting Sisoft Sandra to benchmark your memory too. I bet it'll fail the test and reboot right there. Get new memory, your stuff is obviously trashed. -------------------------------------- you shouldn't have to go through all that for a chip to work on basic settings. the chip is obviously bad. leave the volts alone. RMA the chip. try to get Corsair to upgrade you to an overclocking chip so that you can run at basic settings. all DDR is no created equal. many hardware sites have uncovered the "DDR scandal." lots of chips are made to run at one speed, yet can be pushed higher. many unscrupulous memory corporations market them by their overclocked spec, rather than their true spec. unfortunately when it comes to DDR you've got to exceed your boards max DDR speed to ensure that you can certainly run at the max. ----------------------------------------- To Silo: Your Memtest86 output shows a relatively small defect in your memory (81 errors in 29 passes means that only a few bits are faulty). If bumping to 2.7V doesn't resolve the issue and you tried the other suggestions, ask Corsair for an RMA. My first Corsair 1 GB 3200LL set had one bit error and got replaced without any hassle. ---------------------------------------- Frankly, I don't know what do here. I don't want to get new memory and have the same problem. You want me to pull one of the modules and rerun the memtest? Will try.

The first stick showed 1 error after a 7 hr run. [URL=http://www.publicscrutiny.net/ramtest-2.htm]Here is the stick and the result[/URL] The second stick showed no error after a 3 hr run. What next?

Ran the 2nd 512MB stick overnight on memtest and it ran for 9 hrs without an error. Thus it appears that the 1st stick is the problem. Can it be replaced by itself? Or must I replace both? I cannot do without the computer so must keep 1 stick here to run with. thanks

RG-- I have no friends. I do have 2 questions, however: (1) Only 1 of the sticks produced errors when run alone. However, these were a matched pair of 512s. Do I have to RMA BOTH of the sticks or can I do just the one that's bad? Would the new 512 ram be matched to my existing stick? (2) if I need to do 2 sticks, can you ship me the new ram BEFORE i return the ram i've got? As I said, I need to keep the machine running and can't afford to pull the ram and shut it down entirely while i wait for this ram to get back to you, and the new ram to get back here. thanks.

hi, do this modules working ok while used separately ?

Ram Guy-- back after an extended leave. Still have the problem as above. You replaced 1 stick several months ago and it never worked. In fact, I didn't put it back into the machine until recently. Then I started having the same problem (periodic spontaenous reboot) again. I think I am going to buy another 512 G stick soon and pull both of these and send these two back to you for RMA. Is that OK? Here is the memory support list for my machine: [URL=http://tinyurl.com/2ks5u]Shuttle ST61G4 Mem Support List[/URL] pls advise. rick b. (aka silo)