Asus K8V-SE Deluxe and Corsair TwinX1024-3200LL

Hi, I am having a serious problem. Here are the stats on my computer: Windows XP Athlon64 3400+ Asus K8V-SE Deluxe Corsair TwinX1024-3200LL GeForce4 TI4600 Sound Blaster Audigy 2 Platinum Pioneer DVD Burner Samsung CD-RW Combo Drive Western Digital 80GB and 120GB HDD's 480W Thermaltake Silent Purepower PSU Thermaltake Xaser III V2000A case I'm not sure what else I should list. But, here is my problem... I am constantly getting: -"memory violation errors" -random program crashes (Where I load up a program and it will crash when loading, and will continue to crash everytime I try to load it until I reboot. Sometimes loaded programs crash too) -random reboots -sometimes when booting up Windows XP LSASS.EXE causes a memory violation error and I can't log into windows until I reboot -sometimes when booting up I get a message telling me my domain is not available and it won't let me log on (I am not on a domain, just a workgroup) -sometimes when booting up my video mode is stuck in 640x480 4-bit mode until I reboot. (It is not in safe mode) I am not sure if this is a ram issue or not, but due to the fact that I get a lot of "memory violation errors" I assume perhaps it could be. I have tried everything... my DIMMS are in slots 1-3. I tried changing the voltage settings and timings. I had everything on auto, then I tried changing the DDR voltage to 2.7 and 2.8... It solved nothing. I have tried manually setting the timings, and still I end up with all the same problems. I am not sure what else to try. I let memtest86 v3.1 run for about 40 minutes last night and it found no errors at all. Should I let it run overnight? Should I only test 1 DIMM at a time? Or could this be a defective motherboard or CPU perhaps? My computer is not overclocked either. Sometimes when playing games, the computer will freeze. More than likely it will just reboot instead. What should I do? Is this a compatibility issue?

Run memtest for at least 1 hour, that should be enough, if at the end of testing it still happens run it overnight. The 640X480 things sounds like safe mode to me, so ignore that right now, but this could also be down to a graphics driver so lets uninstall them and reinstall the latest ones. In windows in normal mode run Prime95 (search for it on google) torture test, the one for maximum heat, for at least 1 hour and see if it finds any errors. If none of this works run memtest again for 24 hours and see. Post back after doing all of this with results. Rob PS did your return key break during the really hard to read no gaps paragraph?

Sorry about that hard to read paragraph. I guess I should have read it over before submitting. I let memtest86 run for 2 1/2 hours and it has found no errors. Also, being stuck in 640x480 4-bit was not safe mode. The computer booted as normal. I will try your suggestions Robolf. Thanks for the reply.

Its ok mate, i just find it real hard to take anything in a paragraph like that, see uve changed it now :) Good luck and post back with your findings, were all here to help u. Rob

I just got Prime95. I ran the torture test "Blend", and after 30 seconds, my computer rebooted. I ran it again with the same torture test "Blend", and after 2 minutes I got this error: [Sun Apr 04 18:07:56 2004] FATAL ERROR: Rounding was 0.5, expected less than 0.4 Hardware failure detected, consult stress.txt file. I ran it again with torture test "In-place large FFTs", and after 4 minutes I got this error: [Sun Apr 04 18:15:26 2004] FATAL ERROR: Rounding was 0.5, expected less than 0.4 Hardware failure detected, consult stress.txt file. I ran it again with torture test "Small FFTs", but I stopped it after 10 minutes so I could post back here. It did not fail in those 10 minutes. I will let this test run for a while and post back here again.

Ok... I ran the "Small FFTs" torture test for 2 hours and 4 minutes with no errors. Since that test uses no ram at all, and uses my CPU's cache, therefore the issue must be the ram. I assume the ram must be defective... There is no other explanation for the problems I am having other than that. Should I RMA back to Corsair or the place that I bought it from?

Hmmm... I tested each DIMM seperately using Prime95 and torture test "blend." I only let each test run about 15-20 min per DIMM, and both have had no problems. Now, when both are in the motherboard together, the test fails within a few minutes. So, there is a problem with more than 1 stick of ram being used in my motherboard at a time. Could it still possibly be a problem with the ram? Or does it seem to be more of a motherboard issue? Or quite possibly a CPU issue? (Since the memory controller is built-in to the CPU). Or even a Windows XP issue? (Incorrect setting or something?).

With further testing... I have discovered the problem is a timing issue. CMX512-3200LL timings is 2-3-2-6. With one stick of ram in my motherboard, 2-3-2-6 will work just fine. However, with both sticks of ram in my motherboard, 2-3-2-6 will cause a ton of problems. I have tested 3-3-3-6 and it works perfectly fine. I guess the motherboard cannot handle those timings with more than one stick of ram.

Yeah kind of thought the "in place large FFT's" would bring the problem up, its very strenuous on the memory and cpu, well at least youve found the possible solution. Sorry ive been sleeping over here in the UK since my last post. I Would have thought you could run 2-3-3-6 with 2 sticks on that board mind. To be honest it wouldnt be much slower at anything that 2-2-2-6. Ive also got a K8VSE Deluxe computer im testing but I havent yet got round to putting my two sticks of 512MB PC3200 C2 XMS in yet, so it will be interesting to see what timings i can run at. Just out of interest you have the "Performance Mode" set to either Auto or Standard in Jumperfree configuration, otherwise the HTT is overclocked to 227 and the multiplier droped down to 9, to give about 2050Mhz. Anyway hope that the timings fix your problem, use Prime to test. I would first try dropping CAS back to 2, if that sorts it, i think u wont see much more performance and unless u are prepared to spend hours testing your done :) Rob

Thanks for the replies guys :) My ram is in slots 1-3 and DDR voltage at 2.7. I have ran Prime95 torture test "blend" overnight with timings 2.5-3-2-6 and after 4 hour and 52 minutes, it has finally failed. Performance mode is "Auto." I will have to continue testing... I'm going to try testing 2-3-3-6 while I am at work.

[quote name='RAM GUY']I would suggest testing them one at a time and let's make sure one is not failing. Ad if you have XMS3200LL I would set the timings to the tested settings. Cass 2-3-2-6 with the Rass to Cass set to "3".[/QUOTE] I have tested both of them together with memtest86 v3.1a for over 2 hours with no errors. I was going to test them individually, until I discovered that the ram timings were the cause of my computer problems. How long should memtest86 run to determine if a stick of ram is ok? A couple hours? Overnight? Thanks for your response :)

Success! I ran Prime95 while I was at work using torture test "Blend." I set my timings to 2-3-3-6 (as suggested by Robolf), and it has been running for almost 10 hours with no errors! I have read up on other forums from other sites that AMD64 processors are hyper sensitive to ram and their timings. There is a timing issue in which RAS is accessed prematurely in which this was the case. I had to set my tRP (RAS Precharge) from 2 to 3. Thanks to everyone for their help! EDIT: Jumped the gun just a tad. "In-place Large FFTs" test failed. I forgot to try increasing my tRCP as suggested from an AMD forum. I am running this test successfully so far... I will see how it goes. :)

When uve fixed it dont forget to post back with the "fixed" timings, so it can help others in the dilema. I thought 2-3-3-6 would have done it but ive never used those modules just the normal 3200 C2. Keep going and let us know. Rob

I have been running the "In-place Large FFTs" test for 4 hours now without a problem. So I have come to a conclusion: With AMD64 CPUs, they are hyper sensitive to ram timings. There is an issue with it accessing RAS prematurely which will cause memory problems. You must adjust your timings manually. My ram is timed at 2-3-2-6... With one stick of ram, 2-3-2-6 will function just fine. However, two sticks of ram, I had to set my timings to 2-4-3-6 to function normally. I had to increase my tRCD (RAS to CAS Delay) by 1 clock cycle as well as my tRP (RAS Precharge) by 1 as well. If anyone is experiencing similar problems that I had, I highly suggest attempting to increase those two delays. Also, you should set your DDR voltage to 2.7, and if you have two sticks of ram, make sure they are in slots 1 & 3. Thank you everyone that has helped me. :) EDIT: I ran 2-4-3-6 just fine for over 4 hours with "In-place Large FFTs" but I tested it again and failed after 20 minutes. I am trying 2.5-4-3-6... Is this normal? I thought I had it... I guess almost :sigh!:

Here is the article on the tRCD problem that Athlon 64 processors have: [url]Not allowed see forum rules[/url] Scroll down until you see "Athlon 64 3000+ and higher."

[quote name='RAM GUY']The TRCD or AKA known as Rass to Cass should be set to "3" with these modules Please see my previous post's![/QUOTE] I had it set to 3 and it fails. I have tried everything except test each stick individually which is what I'll try next. Here is part of the article from that link that states there is an issue with tRCD with Athlon 64 CPUs: [QUOTE]A very trivial explanation is that the controller expects that the memory is as fast as it itself but we don't have memory yet running at a 2 GHz and beyond clock rate. What it comes down to is that, depending on the BIOS tuning (by the manufacturer), the read command is issued a bit too early and violates the tRCD. The workaround is to add one additional cycle to the actual memory latency, that is, a 2:2:2-rated DIMM will run stable at 2:3:2; a 2:3:2 rated module will, most likely, need to be set to 2:4:2 in order to run stable (where the latencies are CAS:tRCD:tRP). Bottom line is that when shopping for memory for the Athlon64, it is highly advisable and in most cases mandatory to only consider memory with a tRCD of 2 or, to make it simple, memory that is rated at 2:2:2 a the designated speed. Alternatively, registered DIMMs can be used as well without problems, even though they are slightly more expensive and will give lower performance.[/QUOTE] I'm not sure why they would state registered DIMMs can be used though, as only unbuffered is supported by Athlon64 (Non-FX). I will be testing each stick of ram individually with memtest86 and will post the results here. Thanks. EDIT: Sorry about the link, I did not realize it was to an online store until just now.

I have the latest bios. My motherboard is a K8V-SE and there are no bios releases except for the first initial release. I have tested one stick of ram in my computer last night, and the other stick in my wife's. Her machine is an Athlon XP 2000+ machine on an Abit KG7-RAID motherboard. The max speed her motherboard can go up to for ram is DDR266. The stick of ram I tested in my computer was tested for over 6 hours and had 38 passes with no errors. I didn't think it would make too much of a difference testing it on hers, but the other stick of ram that was tested in hers had gotten 1066 errors before one pass was complete. After 6 hours, there were over 14000 errors. Is that ram stick actually bad or is it because of the machine it is in? I am currently testing that same ram stick in my machine while I am at work.

Errors can be down to timings but maybe this stick is the cause of your problems, really i would advise testing it in the machine you intend to run it in for these reasons. A good 1 hour or 2 memtest if you can. Rob

I just read an article on tomshardware.com that compared many different motherboards with ram. The article is called "Athlon64 Boards Found Lagging in Memory Support." The article clearly shows that there is an issue with using more than one stick of ram. With one stick of ram, I can operate perfectly fine at the rated timings of 2-3-2-6. The article stated they used Corsair ram that was timed at 3-4-4-7. On the Asus K8V board, they had to set the timings to 2.5-5-5-8 for the board to function correctly with the ram. It appears that they have added 1 clock cycle to each of the last 3 timings, and decreased CAS to 2.5. So, in conclusion, I don't have defective hardware after all. My board just lacks proper memory support to handle more than one stick of ram effectively. When I get home from work, I will raise each of the last 3 timings up by 1 (from 2-3-2-6 to 2-4-3-7) and see what kind of outcome I get. I will post the results here. My wife called me at work earlier, and that other stick of ram has been testing for 4 hours now, and has 0 errors. Both DIMMs have been tested individually and are ok according to memtest86. Ram Guy: I guess your motherboard would allow you to have two CMX512-3200LL DIMMs at 2-3-2-6, but mine clearly does not. :( EDIT: The only two boards that were able to run two sticks of ram at their rated timings are Biostar K8VHA Pro and Chaintech ZNF3-150. Also, the Corsair memory used were two Corsair CMX512-4000Pro. They are rated at 3-4-4-8, but they showed CPU-Z reporting it as 3-4-4-7.