Corsair Dominator problem

[dgingeri]

I have 2 kits of Corsair Dominator DDR2-1066 (2X2GB) memory, but I have not been able to run them both in the same machine.

http://www.newegg.com/Product/Product.aspx?Item=N82E16820145214 (They cost a lot more than they used to. When I bought mine, they were running $69 per pair.)

I've tried downclocking them to 533, adjusted the timings down, and changed the command rate to 2T, but I have just not been able to get it to run stable. This is on a Gigabyte P35 board, a Gigabyte P43 board, and 2 different Asus P45 boards, so I know it's not the boards causing the issues.

Anyone else have trouble with this memory? Anyone know a solution?

 

[cisco]

You might try updating the bios to see if there are any updates referring to RAM. I have had problems in the past running 8 gigs of RAM. It is not likely a bad stick of ram, but worth mem testing just to be sure. I had more trouble with a board that had a max memory of 8GB than a board with a max of 16GB. The 16GB I had no problems from the get go with the exact same memory.

 

[dgingeri]

No luck. It's as unstable as ever, no matter the speed.

 

[flyinfinni]

Have you run memtest? And do both sets run properly when you run them individually? I would make sure to run memtest to make sure you don't have a bad stick. Other than that, nothing comes to mind other than bios updated, which you've already tried (and with 4 boards, you'd think one of them would work).

 

[dgingeri]

Each pair, even mixed between them (since they are all ver 3.1) work just fine. I only get instability when I install all 4. I'm thinking it's the memory itself. I have been unable to find anyone running these in a 4X2GB setup.

 

[flyinfinni]

Have you tried each stick by itself? Also- did Memtest give you any errors? What is the slowest clock and loosest timings you have tried running?

 

[dgingeri]

I tried them in pairs and never had an issue, so I figured testing them singly would be kind of redundant. Memtest passes with flying colors in pairs, but gives constant random errors with all 4 installed. no consistant memory addresses from one run to the next. It will also go 2-3 runs with no errors, then error 4-6 times on the 4th, then run fine for 2 more, etc.

I have my cpu fan (Ultra 120 extreme with a nice 120mm fan) positioned where it is drawing air right over the memory heatsinks, so I don't think it is overheating. I could be wrong on that.

On my current Asus board, increasing voltage from 2.0v to 2.1v to 2.2v only makes it more unstable. 2.0v seems to be the most stable. running the recommended 2.1v on one pair makes them unstable, but 2.0v works fine. 1.9v does not work at all at 1066, but 1.8 works fine at 800 and lower. I had to run it at 2.1v on my old Gigabyte boards to get 1066. That's just likely differences in board design.

 

[dgingeri]

I think I'm going to go ahead and get new memory. Do you know of any cheap, decent quality DDR2-800 that would work in 8GB configurations?

 

[dgingeri]

Actually, I just found another posting about this, and I guess increasing the NB voltage would stabilize it. Anyone else heard of this?

 

[flyinfinni]

I've not heard of that for your particular issue, but its worth a try before you RMA and get new RAM, especially since the Corsair Dominators are generally known as really top-notch parts.

 

[roonj]

Yes is the answer to NB issue, the voltage ratings are all made on the basis of just using 2 mem modules not 4. Sometimes boosting the NB 0.1v can help, another question do your modules have the same latencies at 800mhz? If not then set manually to match the weaker pair. Use cpu-z to see if the EPP profiles match on the chips. Usually corsair uses micron for these modules which shouldn't fail at higher voltage but I think I saw them switch to another lately for that chip. There is a version listing on the label which should tell you as well, like 1.1 or 1.2 and 2.1. If they don't match they're different chips on the board and may have different latencies.

 

[dgingeri]

yeah, they're 2 identical sets of the exact same memory, version 3.1. I tried it out last night and it seemed to work. I didn't do anything excessively memory intensive other than start up WoW and exit back out, but it stayed stable while I was doing my Excel homework. I'll test it further this weekend when I get the chance to game a bit, but tonight is for more Excel homework. (I hate that class.)

 

[XMSYellowbeard]

With 4 modules you will likely need to increase the voltage to the NB aka memory controller. Try testing with it in the 1.25v - 1.4v range and with 2.1v on the memory.

Keep in mind that 4 modules at 1066 is way beyond spec for that board so you may not get the full 1066.

 

[dgingeri]

I'm not worried about getting the full 1066 now, 800 will work fine. I just want to get them all stable.

With my NB, the base voltage is 1.1V. 1.4v sounds kinda dangerous. I started at 1.2v so far, but I wasn't able to test very far. Apparently, one stick is bad, and I put in an RMA request for that pair.

I went through and tested one at a time just as I got started because I got a bluescreen I hadn't had before. one of the ones that was sitting around in a static bag gave me some single bit errors, even running at lower speeds. I'll try again as soon as I get the replacement pair.