Corsair 2GB DDR XMS3500LL Pro memory timings

Maxx1984

Maxx1984

Maxx1984

Associate
Joined
18 Oct 2005
Posts
38
Hi can anybody here recommend me some decent memory timings for running an Opty 146 at 2.8ghz on a DFi SLI D as im getting a bit stuck (beeing a stupid noob :/) any help appreciated :)
 
Try running a 166 divider and use 2-3-2-6 timings and go for the fastest you can manage at those timings (230-235mhz). That should get you to about 2.8ghz with that opty. If that works try going to 2-3-2-5, i found that it was just as accessable as 2-3-2-6. You'll probably need 2.9-3.0v but you might get away with 2.7 or 2.8v.
 
Ok thanks - will give it a try...

OK its booted into windows on 2.3ghz so far, haven't got much time to do testing yet. The ddr is running at 333, is this going to be a problem, cause a decrease in performance? or is keeping it at ddr 400 better??
 
Last edited:
Well you'll have to either drop down another divider or loosen the timings on the ram to something like 2.5-3-3-6 (maybe 2.5-3-2-6). Personally i'd loosen the timings. You will also be at the mercy of your CPU and whatever heatsink solution you've got. If you are running on air you almost certainly won't get to 3ghz stabley with any ram at any timings as the CPU will simply be too hot. 2.8 though it probably achievable with a decent heat sink like a Big Typhoon or one of the Thermalrights.
 
That RAM won't do 300Mhz. 250Mhz max. Infineon chips run good timings so use a divider that keeps the RAM running below 250Mhz. At over 218Mhz most people need to loosen the CAS to 2.5. That should be good to 240. 2.5,3,3,7 will get 245-250Mhz average. Voltage 2.7v throughout. The Infineon doesn't respond any better above that.

3.1Ghz is around the max for high end air. I reach 3.06Ghz on a virtually silent HSF.
 
Last edited:
Hmm....there seems to be quite a gap between what you and I dirk have been getting from these sticks. Maybe it is the DFI/Corsair issue. In which case Maxx might well have the same limitations.

Also I think you're being optimistic about what an opty 146 can do on air. I am using the same heatsink as you and though i can get to 2.7 stabley i'll be well into the mid 40's under load and if i go to 2.8 (which isn't so stable) it'd probably be 50 odd.
 
50c is fine. The cores on our processors are basically the same so in theory the two different CPUs should be capable of similar results...but thats rarely the case. :rolleyes: It's only my high case air-flow that enabled me to push past 3.0Ghz. YSTech fans really help ;)

With a Scythe/Vantec Tornado 3.1+ is very possible.
 
You are quite right that the cores come from the same design, but FX chips are choosen for their robustness. There are some CABNE 0540s that might be the same grade, but most optys are somewhere between 3700+ cores and FX-55s and they don't fair so well under heat.

I'm also not quite as confident a toaster as you. Even if the core could go to 3.1ghz I wouldn't consider it stable or safe if it was running at 50C+ (they ain't XP-Ms). Though I suppose that is very much a matter of perspective. I guess i am just too scared to run a chip that hot for long periods. To me it'd just be a matter of time before it burns out, not a matter of luck. But that is a subjective and indeed emotive position.
 
Ok thanks for the feedback. With the actual overclocking I am just hoping to get to these high speeds of 2.8/3.0 not really expecting it too much. I am happy with the chip even running at stock speeds compared to my old cpu. Atm I am running at 250 on the memory so cpu clocked at 2.5ghz now :) yet to run any tests but got into windows fine. Idle cpu temp is pretty much the same as stock aswell (ok maybe a degree or so more)

Mem timing so far are:

2.0-3-6-2

Frq: 208.4mhz - according to cpu-z
 
Last edited:
Ive dropped the 6 to a five like you said, put the freq upto 216.7mhz (htt 260), boots into windows but failed the OCCT test after 55 secs :/ Is it worth trying to use a lower divider??
 
No, you shouldn't need a divider. Try changing the DDR voltage. This ram isn't very responsive to higher volts, but i have found that it does help a bit. Try going up at 0.1v at a time and retesting. If that doesn't work go back to 2-3-2-6. The difference between 5 and 6 is very very little. In fact the difference between 5 and 10 is probably less than the difference between CAS2 and CAS2.5.
 
Safe is very much a matter of perspective. However with this ram, voltage really is less important and as a result i've not heard of anyone overcooking and melting them (though it's probably possible). Some people seem to be getting slightly better stability at lower voltages despite the fact that you might be able to push them further at higher. I've only had these a month or so, and most of the time i've run them at my boards max which is 3 volts. I strongly suspect it's perfectly safe to run them at 3 volts 24/7 without ruining them (otherwise i wouldn't have done it), but as i've been able to get stability at lower volts, why take the chance?

As far as absolutes go, i'd say you probably can't melt any ram at 2.8v or even 2.9v (except maybe that dodgey batch of Ballistix), but if you can only get stability at more than 3.....then maybe it's not worth it. To be honest though you'll probably find that additional volts with this ram makes very little difference indeed (not nothing, but not much) and if you find something running fine at 2.9v then try running it again at 2.8v then 2.7v.

I haven't actually decided where I want to leave my final overclock, but I suspect that once all is said and done I won't be running my ram so hard that i need more than 2.7 or so.
 
Ok looks like a sorted out the problem I was having by increasing the cpu volts. I Still need to finish the OCCT memory test then super pi for tonight. Do you thinkI may be able to get more than 2.6ghz with this memory or have I pretty much reached the memory's limit??
 
Well you can certainly get more with a divider, but I'd be surprised if you couldn't get up to at least 2.7-2.8ghz at 1:1. I think i managed 270 at 3-3-3-7 and 285 at 3-4-4-8 using my x9 multi. Neither was perfectly stable, but there will be a happy medium somewhere in the 275-280 bracket i think with loose timings. I just haven't experimented too much there as i doubt i can use that speed vs timings ratio in my final overclock.
 
im on 2.7/2.8 now, its booting into windows now but failing the OCCT 30 min stability test, I have upped the voltages, changed the timings but still in the same situation so im probably going to put it back to 2.6ghz, where it previously passed the test and super pi. Also CPU-z is not displaying the ram frequency and FSB Dram in the memory section :/
 
Have you lowered your http multiplier to x3? You should be able to get more out of both CPU and memory.

If i were in your position i'd try running a divider, 5:4 should do, and seeing how fast you can get your CPU stabily, then you'll know if it's the ram that is destabilising it or the core.
 
You must log in or register to reply here.
Back
Top Bottom