K8T Master2-FAR : Corsair PC3200 issues
In the past 6 months I've RMA'd my Corsair TWINX1024RE-3200LLPT (2 x 512MB sticks) two consecutive times. I'm now convinced that the DDR is somehow being damaged while it is in my system, although I'm not sure exactly how. I currently have a 1GB set of DDR that was passing 2hrs+ on memtest86+ w/o any problems a couple weeks ago, but now under memtest86+ v1.27/v1.5 it is exhibiting problems - I'm also getting all sorts of random crashes in WinXP.
My signature should have the specs for my workstation, but just in case:
2 x AMD Opteron 248 (2.2GHz)
MSI K8T Master2-FAR / Award v1.1 BIOS
1 GB of TwinX1024RE-3200LLPT (2 x 512MB)
2 x 74GB WD Raptor (RAID striping array on VIA K8T800)
2 x 80GB WD800JB (compressed backup drives)
Sapphire Radeon 9700 128MB
Enermax EG651P-VEFM (550W 24-pin ATX PSU)
Lian-Li PC-69 Aluminum Mid-Tower Case
WinXP SP2 and all critical updates
I also have all of this running off of an APC 1000VA/600W Back-UPS (battery backup to protect from power surges and such) - the monitoring software for the APC shows the workstation drawing at most 230-260W of power at one time (the PSU/CPU is alone on the 1000VA, I have the monitor/router/modem on a separate UPS).
My PSU is rated for 36A on the 3.3V, 5V and 12V lines.
This DDR is not cheap (just shy of $400 for a 1GB set) - I would eventually like to have 2GB (4 sticks) active for my machine but I don't want to risk the $$$ until I know exactly what is causing good DDR to go bad in my system.
Could there be some kind of thermal issue that I'm not aware of? Is there any means of monitoring this via an onboard utility? I've tried CoreCenter but have had all sorts of problems with it...
I have the following thread on forum.msi.com.tw active on the question for further detail:
http://forum.msi.com.tw/thread.php?postid=513103#post513103
Any help would be much appreciated!!!
My signature should have the specs for my workstation, but just in case:
2 x AMD Opteron 248 (2.2GHz)
MSI K8T Master2-FAR / Award v1.1 BIOS
1 GB of TwinX1024RE-3200LLPT (2 x 512MB)
2 x 74GB WD Raptor (RAID striping array on VIA K8T800)
2 x 80GB WD800JB (compressed backup drives)
Sapphire Radeon 9700 128MB
Enermax EG651P-VEFM (550W 24-pin ATX PSU)
Lian-Li PC-69 Aluminum Mid-Tower Case
WinXP SP2 and all critical updates
I also have all of this running off of an APC 1000VA/600W Back-UPS (battery backup to protect from power surges and such) - the monitoring software for the APC shows the workstation drawing at most 230-260W of power at one time (the PSU/CPU is alone on the 1000VA, I have the monitor/router/modem on a separate UPS).
My PSU is rated for 36A on the 3.3V, 5V and 12V lines.
This DDR is not cheap (just shy of $400 for a 1GB set) - I would eventually like to have 2GB (4 sticks) active for my machine but I don't want to risk the $$$ until I know exactly what is causing good DDR to go bad in my system.
Could there be some kind of thermal issue that I'm not aware of? Is there any means of monitoring this via an onboard utility? I've tried CoreCenter but have had all sorts of problems with it...
I have the following thread on forum.msi.com.tw active on the question for further detail:
http://forum.msi.com.tw/thread.php?postid=513103#post513103
Any help would be much appreciated!!!
0
Comments
-drasnor
-drasnor
Drasnor : I actually read your thread when I first bought my board - I'd already increased the DDR volated in the BIOS to 2.65V. I tried adjusting this earlier on when the DDR issue initially came up to no effect. After I RMA'd my first set of DDR I was running fine for a few weeks (on the 2nd remaining gig - Memtest86+ tested fine for 2hrs+ w/ 1GB) then all of sudden my workstation starts flaking out on me again and Memtest86+ starts reporting errors.
MSI HQ is telling me to run my next test with the 2nd CPU removed to see if I might be overdrawing my PSU - yes they keep telling me to get an Enermax 465. I wish they'd told me this before I RMA'd my mobo (and PSU, and DDR) but I'll be sure to test everything (pain in my !@#!# ass) in stages when I get back to putting Humpty back together again hopefully later this week.
I'm worried that this is a thermal issue - I've ordered a set of new ThermalTake A1214 80mm fans that I'm going to add to my case - I may try and bracket a fan over the DDR but this is basically just a shot in the dark.
Admittedly, maybe the DDR is reporting bad simply because the mobo is overdrawing the PSU, but my last test (before RMAing my DDR the 2nd time) I switched out the DDR in sets and ran Memtest86+ - the result consistantly were that set A passed w/ no errors, and set B displayed errors at the same memory locations. This doesn't seem consistant with a power consumption problem, but there may very well be
2 quick questions :
What PSU do you use?
How many DIMMs do you have (all four slots or just two, if two, what slots are you using)?
I know you have custom cooling setup for your rig, I'm just using the stock solution myself, but I'd be interested to know.
Thanks!!!
I used to just use the first two DIMM slots (the ones closest to CPU0) but since I bought citrix's RAM from his FX box I use all four DIMM slots now (2GB, 4x512MB Corsair XMS PC3200RE-LL.
I'm using a Lian Li PC-7B and my cooling is passive on the RAM (no fans blow directly over them), 2x MCX462+ on the processors with Panaflo 80mm H1A's, 2x Panaflo 80mm H1A's up front on intake and 2x more on the rear outtake (yes, I had to cut a hole in my case to fit the second one). Stock sinks on motherboard chipset and graphics card. I'll get the temps for you later today (I'm on my laptop at the moment).
If you're extremely unlucky, remember that the RAM controller is on the CPU. It might be bad.
This sounds somewhat similar to my sister's old dual MP rig, which I hope isn't the case because I never did find out what was wrong with it. The power supply was good, but the motherboard had to be replaced a few times, the processors tested bad in a single processor machine a few times, and it went through RAM like nobody's business. I eventually gave up on it after I ran out of motherboard warranty and pulled everything for other machines, where they perform happily.
-drasnor
God Forbid. Would Memtest86+ register this - is there any way of checking this?
I've RMA'd the mobo and PSU (god bless Newegg), hopefully the replacements will arrive sometime this week. I can check the CPUs individually after that - that is, if there is a means verifying this problem.
Could a faulty memory controller damage good Corsair DDR?
Thanks!
Tex
As Tex said, ideally you'll want to test the CPU's individual in a single 940 board, but those don't exactly grow on trees so you'll have to be creative (friends with FX systems?)
I have no idea whether or not a faulty memory controller could damage the memory, but my intuition from dealing with other electronics makes me doubt it.
-drasnor
Tex
http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_871_9033,00.html
but I don't know if they will do much more than display info/id about the CPUs.
Gonna post this issue on the Opteron forums @ amd.com and see if I get any bites.
Quick question, from one of the responses my MSI HQ thread: "Corsair is arguably the worst pick for non ecc-reg on amd64."
Is this true? I went w/ Corsair since I've had pretty good experiences with the unbuffered PC3200 DDR (admittedly there is a certain degree of hype around their DDR as well)... is their registered DDR a different story?
Thanks!!!
Most are not a either or deal though. You either use unbufferred (non registered) or you use registered. And Corsair cut their nuts being famous for their regular non registered DDR. I have used corsair reg sdram but not their ddr.
Remember drasnor said all the memory controller ability is run through ONLY ONE OF THE CPU"s on that board so swapping the cpu's from one socket to another might fix the whole deal.
Tex
Tex
It could have been just a problem that occured only on AXPs where the memory controller was not on the CPU.
Tex : When I get my board back and start reassembling, I'll try testing the CPUs one at a time and will let you all know what I come up with. My real question at this point is if there is a specific way I can test for this memory controller issue aside from running w/ Memtest86+ (Drasnor seemed uncertain if Memtest86+ would reliably flag this issue) one each CPU individually - I'm pursing this on the AMD Opteron forums and will post anything I find out here as well.
Wish me luck!!!
Tex
Can anyone give me a temp to gauge what my DDR should be at? I installed an AerogateII thermal controller and currently my DDR is anywhere from 90F - 98F... does this sound reasonable?
Thanks!
Those temps are absolutely fine. Any ram should be able to take that. As far as i know, NO A64 cpu thus far supports DDR400 speed when more than 4 banks of memory are in use. For best stability, back the memory speed down to DDR333/166 with the divider unless ou have the mainboard at auto/spd timings, then it should do that automatically. At stock voltage, the memory should survive even without a casefan. At least quality ram. I dont recommend it, but it should, really.
When there is cpu trouble in memtest, especially it's memorycontroller, you'll see it in test 6. If memtest errors out in test 5 before test 6, then it is the memory that is fault.
that is really useful info mac, thanks!
Tyan Thunder K8W S2885ANRF
2 X 248 Opterons
LSI 1600 Elite
A Bunch of hdd's & a couple of dvd burners
nVidia 5950