Corsair XMS DDR2 failures
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar
Corsair XMS DDR2 failures
Just thought you guys might want to know this info.
For the last year and a half my gaming computer has been troubled by random reboots and instability.
Now I pride myself in knowing how to put together and manage a PC, and this project was really shaking my believe that I knew what I was doing. About 6 months down the road, I ran Memtest 86. Low and behold, my corsair XMS twin-x DDR2 ram was throwing errors.
I bought 2 more Corsair XMS twin-x DDR from Newegg, again pc6400 like before. I figured I would swap those in and RMA the other ones. Once I got the RMA memory back, I threw a stick in along with the other two, since they are rated at the same speed. Meanwhile I'm experiencing all sorts of crashes. Well, I ran memtest 86 this morning before I went to school. When I came home I saw 30 more errors.
So the RMA ram I got back is ALSO DEFECTIVE, and I'm going to have to RMA to newegg again. Last time I did the RMA process it cost me $7 of my own money to ship the RAM back.
I'm never buying corsair memory again.
Also check newegg sometime, there are many people who have had to RMA three times until they got good memory.
Should you settle for less than perfection? Or test every stick of corsair until you finally get good ones?
For the last year and a half my gaming computer has been troubled by random reboots and instability.
Now I pride myself in knowing how to put together and manage a PC, and this project was really shaking my believe that I knew what I was doing. About 6 months down the road, I ran Memtest 86. Low and behold, my corsair XMS twin-x DDR2 ram was throwing errors.
I bought 2 more Corsair XMS twin-x DDR from Newegg, again pc6400 like before. I figured I would swap those in and RMA the other ones. Once I got the RMA memory back, I threw a stick in along with the other two, since they are rated at the same speed. Meanwhile I'm experiencing all sorts of crashes. Well, I ran memtest 86 this morning before I went to school. When I came home I saw 30 more errors.
So the RMA ram I got back is ALSO DEFECTIVE, and I'm going to have to RMA to newegg again. Last time I did the RMA process it cost me $7 of my own money to ship the RAM back.
I'm never buying corsair memory again.
Also check newegg sometime, there are many people who have had to RMA three times until they got good memory.
Should you settle for less than perfection? Or test every stick of corsair until you finally get good ones?
I've used Kingston Valueram and Corsair XMS (for value and performance builds, respectively) almost exclusively for as long as I can remember, and never had a problem with either. I'm not questioning your intelligence, but sometimes I skip out on the most basic stuff; Are you sure you have the right settings?
Although Corsair has certain memory parts rated at certain speeds, they still program the JEDEC standards, so unless you manually change them they won't be running at optimal settings, which may lead to instability. In my current build I have (2) sticks of 1GB DDR2-667 running at DDR2-750 4-4-4-12 @2v. If I just pop them in and let JEDEC run them, they run DDR2-667 5-5-5-15 @1.8v.
I've never had problems with Corsair, so that's why I'm curious.
Although Corsair has certain memory parts rated at certain speeds, they still program the JEDEC standards, so unless you manually change them they won't be running at optimal settings, which may lead to instability. In my current build I have (2) sticks of 1GB DDR2-667 running at DDR2-750 4-4-4-12 @2v. If I just pop them in and let JEDEC run them, they run DDR2-667 5-5-5-15 @1.8v.
I've never had problems with Corsair, so that's why I'm curious.
This is an interesting question. I was under the impression that RAM had a basic programmed function that set default speeds in BIOS. I believed this because a few years ago Corsair had "Plug and Frag" technology where you inserted the DIMMS and they set themselves automatically to optimal settings.
So I guess I have to check my settings when I get home. I'm not entirely sure that my Gigabyte bios even lists the voltage of the RAM, but you can raise in in 0.1V increments.
If this is the case I feel really stupid.
So I guess I have to check my settings when I get home. I'm not entirely sure that my Gigabyte bios even lists the voltage of the RAM, but you can raise in in 0.1V increments.
If this is the case I feel really stupid.
I've twice seen problems with Corsair RAM - caused by BIOS having no support for specific RAM module. In one case helped copying all RAM timing settings from another PC (where same modules worked flawlessy); in another case BIOS upgrade helped.
DRAM modules have SPD function (tables), where's written info about timings and other paramters, unfortunately SPD format and attributes seem not fully standardized and often BIOSes have problems with specific RAM modules (especially faster ones).
--
Some (if not most) Gigabyte BIOSes allow fine tuning of RAM parameters by pressing Ctrl+F1 (try on different BIOS screens).
DRAM modules have SPD function (tables), where's written info about timings and other paramters, unfortunately SPD format and attributes seem not fully standardized and often BIOSes have problems with specific RAM modules (especially faster ones).
--
Some (if not most) Gigabyte BIOSes allow fine tuning of RAM parameters by pressing Ctrl+F1 (try on different BIOS screens).
Another thing to try no doubt. I have 4 sticks @ 1GB each, I will begin testing each individual stick in the same slot, and then after I have confirmed which ones work, will try moving around slots.clalor wrote:Have you tried the RAM in different slots? I went through a few sticks of Crucial memory on an older computer before I realized that the first slot on the motherboard was causing the issue. Moving the RAM to the next available slot solved the problem.
man, who thought messing with computers could be so fun!
Unfortunately, it takes a long time to confirm with memtest86. Maybe it will be faster w/ 1 lone stick.
-
- *Lifetime Patron*
- Posts: 541
- Joined: Sun Jan 04, 2004 11:35 am
- Location: Chicago, Ill., USA
- Contact:
I've bought many memory brands over the years and have RMA'ed virtually all of them at some point.
I've also noticed that memory that was once fine can go bad. I've seen this with two sets of G.Skill "HZ" memory. I thought this stuff was the best as it's supposedly the same stuff as the really high-end memory but at a lower price. I have a hunch now that they're over selling it. At any rate, my first 2x1 GB matched pair memtest86'ed okay when new, ran for a couple months, then I started having mysterious problems. Later memtest86 would sometimes show an error, but not always. I RMA'ed the memory anyway, and haven't had a stability problem since.
Now I've got a 2x512 MB set; one stick has gone bad. The computer won't boot with it in any slot. What's really weird is that I got it running for a while; shut down the PC to install a new DVD drive, and it wouldn't boot. It's going to be RMA'ed shortly.
A few years ago I had some Corsair XMS go bad.
I recently purchased a 2x1 GB Crucial Ballistix set where one stick was bad straight out of the box. RMA'ed it from newegg, no problems since.
I'm starting to form the opinion that memory is like hard drives---no one brand is better than the other; you'll have people who swear by certain brands and hate others, but if you sample enough people, I bet you'd find that the failure rates are fairly similar for all brands.
I just deal with it, look at it as a fact of life.
I've also noticed that memory that was once fine can go bad. I've seen this with two sets of G.Skill "HZ" memory. I thought this stuff was the best as it's supposedly the same stuff as the really high-end memory but at a lower price. I have a hunch now that they're over selling it. At any rate, my first 2x1 GB matched pair memtest86'ed okay when new, ran for a couple months, then I started having mysterious problems. Later memtest86 would sometimes show an error, but not always. I RMA'ed the memory anyway, and haven't had a stability problem since.
Now I've got a 2x512 MB set; one stick has gone bad. The computer won't boot with it in any slot. What's really weird is that I got it running for a while; shut down the PC to install a new DVD drive, and it wouldn't boot. It's going to be RMA'ed shortly.
A few years ago I had some Corsair XMS go bad.
I recently purchased a 2x1 GB Crucial Ballistix set where one stick was bad straight out of the box. RMA'ed it from newegg, no problems since.
I'm starting to form the opinion that memory is like hard drives---no one brand is better than the other; you'll have people who swear by certain brands and hate others, but if you sample enough people, I bet you'd find that the failure rates are fairly similar for all brands.
I just deal with it, look at it as a fact of life.
Just for testing - try with Prime95, torture test, 2 instances (or how many cores your CPU has) - for RAM unstability issues this test usually fails within 15 minutes. I've seen numerous cases, when memtest86+ runs perfect, but Prime95 fails.djkest wrote:Unfortunately, it takes a long time to confirm with memtest86. Maybe it will be faster w/ 1 lone stick.
Start with 4 RAM sticks - if Prime95 doesn't fail, then it will run with 1 or 2 sticks either. 4 sticks is most unstable combination anyway.
Okay okay.
Origionally had slots 2, 3, 4 occupied with 1GB sticks.
I pulled the #2 DIMM leaving only 3,4 occupied.
On the DIMM it says 5-5-5-12 1.9V (currently running 5-5-5-18 b/c unable to change timings in bios)
The BIOS doesn't have good RAM adjustments. Only thing I was able to verify is that it was running RAM @ 1.8V, I bumped it to 1.9V using the "+.1V" option.
I just completed 1 pass of Memtest86+, 0 errors! I'm going to run it all night just to be sure. If this passes, I will pull those 2 DIMMS and throw in the other 2 and test them for at least 2 hours.
Another interesting thing (to me anyway) is that while in the BIOS, I turned off all IDE devices, turned of Parallel port, serial port, raid, and floppy. I disabled everything I didn't use. Now, my computer gets from on to windows load in 7 sec, down from about 20. Makes me want to dive into the other desktop and see what I can accomplish.
Origionally had slots 2, 3, 4 occupied with 1GB sticks.
I pulled the #2 DIMM leaving only 3,4 occupied.
On the DIMM it says 5-5-5-12 1.9V (currently running 5-5-5-18 b/c unable to change timings in bios)
The BIOS doesn't have good RAM adjustments. Only thing I was able to verify is that it was running RAM @ 1.8V, I bumped it to 1.9V using the "+.1V" option.
I just completed 1 pass of Memtest86+, 0 errors! I'm going to run it all night just to be sure. If this passes, I will pull those 2 DIMMS and throw in the other 2 and test them for at least 2 hours.
Another interesting thing (to me anyway) is that while in the BIOS, I turned off all IDE devices, turned of Parallel port, serial port, raid, and floppy. I disabled everything I didn't use. Now, my computer gets from on to windows load in 7 sec, down from about 20. Makes me want to dive into the other desktop and see what I can accomplish.
Okay, the tally is in.
@ +0.1 V, using only slots 3 and 4, I have made 8 passes with memtest for all of my DIMMS with 0 errors. I also have ran 3DMARK06 3 times in a row and played Dawn of War for 1 hour.
So basically, it looks like I have a bad slot on my motherboard, and that my RAM requires more voltage.
I also updated the BIOS to the most current version, which was scary. Overall it was a learning experience, and hopefully with this solved and my smoking northbridge tamed this computer will finally run stable.
@ +0.1 V, using only slots 3 and 4, I have made 8 passes with memtest for all of my DIMMS with 0 errors. I also have ran 3DMARK06 3 times in a row and played Dawn of War for 1 hour.
So basically, it looks like I have a bad slot on my motherboard, and that my RAM requires more voltage.
I also updated the BIOS to the most current version, which was scary. Overall it was a learning experience, and hopefully with this solved and my smoking northbridge tamed this computer will finally run stable.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
I have to throw my 2 cents in since I also purchased Corsair XMS 667 CAS 4 (2 sticks) because it was really cheap and it OCs well. One was clearly DOA, as the system ran fine with the other one at all voltages and the DOA one didn't run at any voltage.
However,
Both mobo's and RAM are blamed for errors beyond the average person's knwledge. Somebody's standards somewhere call for RAM to work at 1.8 volts and for mobo's to default to that. There are problems with that:
1. The BIOS setting of 1.8 volts may actually deliver slightly less than that, which is all it takes for the RAM to not work.
2. 2 sticks makes the voltage droop a small amount, enough so you can't get to the BIOS setup screen, ie the RAM fails.
3. Some sticks want at least 2 volts.
I always start a new pc with just one stick, and go straight to the BIOS to set it to 1.9 volts. Then I put the 2nd stick in. This method works well for many people, as seen of the feedback at NewEgg, which is where I learned about it. Mobo makers have gotten smarter (they hate returns that work) by making the default voltage 1.85 volts, which allows more RAM to work without much of a downside.
Biostar is really smart, having a 2.2 volt jumper on their T-Force series, most likely to be OC'd and therefore have faster memory that requires more voltage. I'm not sure about their lower mobo's as I have not purchased any of those.
As I learned from NewEgg feedback my memory didn't OC well until I upped the voltage. It's around 2.1volts now, and slightly warm. I was surprised how thick the metal of the heatsinks are on the XMS series. My DDR2-667 is running as DDR2-800 CAS 4 1T without any problems.
The restocking fee bothers me with DOA parts, these parts are not going back into inventory, they are being returned to Corsair, I hope.
I can't explain why so many have had DOA sticks. One person noted a high degree of static, the blister package plus the styrofoam peanuts being shakened in transit. Needless to say I went totally anal-retentive with static prevention the second time. I can't say for sure it was a factor. But if it was why wasn't other RAM affected as well?
However,
Both mobo's and RAM are blamed for errors beyond the average person's knwledge. Somebody's standards somewhere call for RAM to work at 1.8 volts and for mobo's to default to that. There are problems with that:
1. The BIOS setting of 1.8 volts may actually deliver slightly less than that, which is all it takes for the RAM to not work.
2. 2 sticks makes the voltage droop a small amount, enough so you can't get to the BIOS setup screen, ie the RAM fails.
3. Some sticks want at least 2 volts.
I always start a new pc with just one stick, and go straight to the BIOS to set it to 1.9 volts. Then I put the 2nd stick in. This method works well for many people, as seen of the feedback at NewEgg, which is where I learned about it. Mobo makers have gotten smarter (they hate returns that work) by making the default voltage 1.85 volts, which allows more RAM to work without much of a downside.
Biostar is really smart, having a 2.2 volt jumper on their T-Force series, most likely to be OC'd and therefore have faster memory that requires more voltage. I'm not sure about their lower mobo's as I have not purchased any of those.
As I learned from NewEgg feedback my memory didn't OC well until I upped the voltage. It's around 2.1volts now, and slightly warm. I was surprised how thick the metal of the heatsinks are on the XMS series. My DDR2-667 is running as DDR2-800 CAS 4 1T without any problems.
The restocking fee bothers me with DOA parts, these parts are not going back into inventory, they are being returned to Corsair, I hope.
I can't explain why so many have had DOA sticks. One person noted a high degree of static, the blister package plus the styrofoam peanuts being shakened in transit. Needless to say I went totally anal-retentive with static prevention the second time. I can't say for sure it was a factor. But if it was why wasn't other RAM affected as well?
On the one hand, I felt like an idiot for not knowing about the voltage, but I was under the impression that it was normal RAM and would run under normal circumstances. That is hardly the case.
You think the average user is going to know that they need to increase the voltage on their RAM to make their computer run well?
Also, I got 2 more sticks of the same Corsair DDR2, and this time there was no voltage rating on them at all. These are running okay in a system at 1.8V, but it makes me wonder if they will crap out on me when put under heavy strain.
You think the average user is going to know that they need to increase the voltage on their RAM to make their computer run well?
Also, I got 2 more sticks of the same Corsair DDR2, and this time there was no voltage rating on them at all. These are running okay in a system at 1.8V, but it makes me wonder if they will crap out on me when put under heavy strain.
-
- *Lifetime Patron*
- Posts: 1809
- Joined: Sat Apr 24, 2004 1:45 am
- Location: At Home
So here I sit, forced to consider the source of 372 errors in Memtest86+ v. 1.70 on my new 4GB of Corsair XMS2 memory. And I'm only 18% into the first pass.
I bought this 2x2GB, PC6400, DD2-800 kit (CORS TWIN2X4096-6400C5) because of the $40 MIR @ NewEgg. It's paired with my equally-new Abit AN-M2 mobo. To avoid the Ninja, it's running in dual-channel mode on the two outer-most (color-coded) DIMM banks.
Like the above poster, the sticker on the DIMMs says 1.9v. My mobo had the default voltage set to 1.85, so I bumped it up effortlessly to 1.9. The mobo then measured an actual DD2 voltage of 1.92 on the next bootup. Is this a problem?
The sticker on the DIMMs also lists the timings as 5-5-5-12. Memtest, however, is reporting that the CAS is 5-5-5-18. BIOS confirms that when DRAM is set to auto, it detects 5-5-5-18. Why? The first thing I will try is bumping it down manually from 18 to 12.
I'll also be performing the aforementioned suggestions. At this point, I'm probably screwed. I stupidly sent in the rebates today for both the mobo and RAM before reading the fine print: no returns (neither refunds nor replacement) for items with MIR.
I bought this 2x2GB, PC6400, DD2-800 kit (CORS TWIN2X4096-6400C5) because of the $40 MIR @ NewEgg. It's paired with my equally-new Abit AN-M2 mobo. To avoid the Ninja, it's running in dual-channel mode on the two outer-most (color-coded) DIMM banks.
Like the above poster, the sticker on the DIMMs says 1.9v. My mobo had the default voltage set to 1.85, so I bumped it up effortlessly to 1.9. The mobo then measured an actual DD2 voltage of 1.92 on the next bootup. Is this a problem?
The sticker on the DIMMs also lists the timings as 5-5-5-12. Memtest, however, is reporting that the CAS is 5-5-5-18. BIOS confirms that when DRAM is set to auto, it detects 5-5-5-18. Why? The first thing I will try is bumping it down manually from 18 to 12.
I'll also be performing the aforementioned suggestions. At this point, I'm probably screwed. I stupidly sent in the rebates today for both the mobo and RAM before reading the fine print: no returns (neither refunds nor replacement) for items with MIR.
just wanted to add my experience.
i bought a pair of corsair xms ddr2 (2x2gb), random crashes, ran memtest and saw errors.
rma'ed it with ~$6 shipping.
the pair failed to boot/and failed memtest when it did run.
rma'ed it again, ~$6 shipping.
once again, it failed memtest.
another rma, ~$6 shipping.
finally this pair passes memtest,
and i can only pray it stays that way.
i have bought ~8 pairs of memory from corsair and only with this last pair did i experience issues.
during the process, i have tried increasing the voltage, up to 1.9v, and lowering the speed from ddr800 to ddr667, it didnt help.
i dont think i would stop buying corsair memory, i just hope i dont run into issues like this again.
i bought a pair of corsair xms ddr2 (2x2gb), random crashes, ran memtest and saw errors.
rma'ed it with ~$6 shipping.
the pair failed to boot/and failed memtest when it did run.
rma'ed it again, ~$6 shipping.
once again, it failed memtest.
another rma, ~$6 shipping.
finally this pair passes memtest,
and i can only pray it stays that way.
i have bought ~8 pairs of memory from corsair and only with this last pair did i experience issues.
during the process, i have tried increasing the voltage, up to 1.9v, and lowering the speed from ddr800 to ddr667, it didnt help.
i dont think i would stop buying corsair memory, i just hope i dont run into issues like this again.
Ive had bad experience with geil ram and crucial ballistix pc8500 1GB sticks, I've a total of 6 of those, and 3 are broken. It seems that asus boards are very picky on ram, and vdimm voltage isnt always what you think it is, my Maximus Formula SE board will overvolt dimms badly, while other boards will undervolt etc, so its not an exact science either. As they say: YMMV. I've had the least problems with corsair, apacer and kingston to date, but i dont overclock ram either.
Bump of the Year Nomineebbzidane wrote:just wanted to add my experience.
In the end I gave up trying to run my 1066Mhz sticks at that speed.
Am running 4x1gb OCZ at 333mhz (667 effective) at 5-5-5-15 1T on Asus striker.
It's very stable at that speed and the games I play all run fast so I don't mind really.