Anybody with experience mixing two different sticks of RAM in dual channel?

Enhanced Interrogator

[H]ard|Gawd
Joined
Mar 23, 2013
Messages
1,428
As long as they are the same speed, voltage, and capacity in each bank, then they should work, right?

I ask because I bought a quad channel kit of some fast Geil DDR2, but one turned out faulty. Could only get refund for the faulty one, so now I have an odd one out. I can't find a single module of the same model, but I can find other single modules that run at the same speed and voltage.
 
Just run them at the speed of the slowest stick and the higher specified voltage of the 2. I've never had a problem doing it that way.
 
Try it and see,
If it fails, do as tikiman suggests and work your way up from there.
I would get it working at slowest settings and make sure the system is rock stable for a few weeks, then start tweaking.
Otherwise you might put crashes down to the changes you just made when they might have occurred anyway.
 
Thanks guys, I found a single stick from the OCZ Reaper series, that should work well. My system's limited by the northbridge with 4 DIMMs anyway.
 
Well I got everything setup and my system passed memtest for 9 hours while overclocked 20%. But prime 95 and BF4 are crashing, even when I reduce the overclock by 4% or so. I'm guessing my northbridge/mch isn't agreeing with the mixed RAM.

I know it's not my CPU because I was running with this overclock before with two of the current Geil sticks and two of another (which I'm trying to sell because they're worth more, and my FSB stopped me from getting anywhere near their limit) . More expirementation to come.
 
I am not surprised that you are having issues with mixing RAM.

I have tried it in the past, and it is very hit or miss as to whether it will work properly or not.

There are things like sub-timings that are going to be different between the sticks. You are going to get very lucky if you find two different sticks that will work completely properly at the same voltage, main timings, and sub-timings.
 
The thing is that I'm not running them anywhere near their max speed/timings. I thought running them a little bit under spec would give them enough wiggle room.
 
In your last post you said you are running them overclocked 20%.
 
These are performance RAM that have to be overclocked to run at spec. So you have to give them an addition .3v to get them to run at the timings advertised. In my case DDR2 800 RAM that is made to run at CL4 @ 2.1v, which I'm running as DDR2 1,000 @ CL5 @2.1v, which is generally very safe as most articles I've read show this type of RAM running at 1,066 or higher with CL5 timings.
 
Last edited:
You may have to relax some timings further or reduce the overclock.
 
I've run two sets of memory together. Generally I just run at whatever the slower sticks stock speeds and voltages are and I haven't had a problem. If you try to overclock thats where you may start having issues.
 
You may have to relax some timings further or reduce the overclock.

Yeah, this is what it's looking like. I'm gonna play around more but I think the memroy controller already is at its limit just over 400FSB, so any small difference at that speed will probably destablilize it.

I hit a snag for a second there when I forgot to restore my BIOS to AHCI after resetting the CMOS, and windows did an automatic repair and put itself in IDE mode. Took me a little while to figure out why I couldn't make it to the desktop. Had to go into the registry to undo the "repair".
 
Are you populating all four DIMMs?

Yeah, and it wasn't a problem with my previous configuration with 2 and 2 instead 3 and 1. Right now I'm trying to put the odd one in different slots to see if it plays nice anywhere. Then I'm going to try to run the RAM at fast timings normal speed instead of normal timings and fast speed.
 
My experience in the past when overclocking with all 4 DIMM slots filled was giving the RAM more voltage than what they were rated for.
 
I didn't give it that much more volts, just enough to make it stable. I think the problem was the system just couldn't keep the voltage steady enough when overclocked. Giving it more volts allowed it to stay stable when the voltage dropped after being loaded. Just a theory, but it's the only reason why I can think that. Google it. I'm not the only one to have done it before.

What are your system specs?
 
Oh ok, I tried bumping voltage earlier in the day and it didn't seem to do much. I tried bumping MCH and FSB voltages and they seemed to keep it stable slightly longer but that may have just been due to chance.

My system specs are:
Core 2 Quad q9650
Gigabyte P35-DS3R Rev 2.1
3 Geil Black Dragon 2GB + 1 OCZ Reaper 2GB
XFX 7850 2GB
Windows 8.1
Audigy 2 that is disabled at the moment.

And now I have a new problem I've never seen before that I'm going to post in the motheboard section. So I won't be able to mess with my RAM timings and voltage for a little bit.
 
What are the voltages you are using?
What voltages are your memory rated up to?
 
When I'm overclocking, running 2.1v with CL5 1000mhz. That's what worked when I used 2 of the Geils and 2 other 2GB DIMMS I'm not currently using.

Today I've also tried running at 2.2v, as well different MCH and FSB voltages, which didn't work.

The Reaper and Geils are both rated for CL4 and 800mhz @ 2.1v.

I'm pretty sure this is an issue with my MCH or Northbridge having trouble managing the odd stick along side the 3 Geils.
 
Check inside the PC for heat sources.
Your face is pretty good for locating hot areas, then see what is getting hot with your fingers
Blow a fan on anything hot, this may help solve any instability.
 
Yeah, when I was overvolting MCH and FSB by .3v, I could only leave my finger on the northbridge for about 6 seconds. Knocked it down to +.2v and it was a much better temperature. Both scenarios crashed under prime 95 anyway, somwhere around the 20 minute mark.

I'll feel around though and see if any other chips are hot.
 
When something gets really hot, it can be permanently damaged and run hotter at its normal voltage.
This can make it less stable and less able to run at its default speed without more voltage and/or better cooling.
Of course more voltage can make the situation worse as you can have thermal runaway or exceed the basic voltage specs causing even more damage.
Always go careful when pushing the boundaries, make sure everything is running cool enough.
And check online how much voltage starts to cause damage from other peoples experience for the same kit/manufacturing process as that you are using.
 
And check online how much voltage starts to cause damage from other peoples experience for the same kit/manufacturing process as that you are using.

Yeah, I did that, and the best info I could find was an FAQ for my mobo on this site. Basically said 5 seconds on the finger test and you need to get air on it. So I just backed off on the voltage. That's not a super precise test though so I guess it's possible I pushed it too far. I was only running it for about half an hour like that though, so I'm hoping nothing bad happened.
 
Ok, got everything running, played a few long games of BF4, with CL4 timings at a 1:1 memory ratio at 800mhz. I got there by starting with one DIMM, booting up for a few minutes, shutting down, then adding another DIMM, and repeating, until all four were in. I'm not sure why that worked.


Before all this, I was running 5:6 with CL5 at 1000mhz, which gives me a 5% higher FSB and CPU clock. I couldn't really tell a difference in performance. I need to play Hainan resort in a 32v32, then I might have a better idea if it really is negligible. Maybe it's possible tighter timings are more important when bottlenecked by the CPU?
 
I've had memory sticks fail when using too loose timings as well as too tight.
Probably isnt what happened to you.
Perhaps a bad contact.
 
After running a couple days at 800mhz/CL4, moved it back to 1000mhz/CL5 and everything is running great. I think I get slightly better minimums with the latter in BF4, due to the slightly faster FSB and CPU speed I can get.

I don't understand why it works now, my memory controller just needed to be eased in to it? Guess it could have been a bad contact but everything in PC is squeaky clean.
 
Back
Top