el gappo

if you take a look around alot of phenom II's have trouble running 1600mhz ram at stock settings. sad but true

try a little extra ram voltage and +0.2 on your north-bridge with a 2t command rate. if thats stable work from there

mine wont even boot at stock setting whatsoever, but when i get it right i take them all the way up from 8-8-8-28-1t @1600 (stock) to 7-6-6-24-1t @1800mhz. you've just gotta put in a bit of effort

__________________
no P.O.S.T? i made a guide for you CPU-Z validation of am3 700e
Phenom II 965 4.532ghz CPU-Z validation AIR and 4.612ghz on the bong hwbot profile
from here on in i shall be known as GAPPO THE DESTROYER

morssolis

i tried it at auto at 1333 and the correct voltage and still only errors. the best it can do is 800.

if it were the imc why does 1 stick work fine and the other on the same settings or even lower settings does not work? i dont know a lot about this. i am sure you guys know a lot more than i do. can you explain why this would be please?

EDIT: let me say what i did to test these and you let me know what you think about it.

with both sticks in at the stock settings for the ram 8-8-8-24 1600 1.64 volts orthos crashes system on blend. memtest with these same settings is nothing but errors.

with both sticks in and timing on auto set at 1600 and 1.64 volts same thing as before.

with the good stick in everything passes no matter what i set it at. 1600 1.64v 8-8-8-24 timing no problem. orthos was fine memtest passed.

with the bad stick in i set to 1.64v auto timing 1600. memtest = errors.
at 1.64 1333 auto timing memtest = errors.
at 1.64v 1333 8-8-8-24 memtest = errors.
at 1.64v 800 auto timing memtest = pass.
at 1.64v 800 timing at 8-8-8-24 was fine also i think. i am pretty sure i tested this.
when i tested the bad stick in memtest i let it run for like 2 mins with nothing but errors. when it was at 800 i let it got through like 4 or 5 tests and had none.

i will try messing with the NB volts when i get home from work. i would love to get this to work. i just dont think its anything but the ram. with the little i know everything is telling me the stick is bad. i mostly think that because the other one has no problem and it will bot even run at 1333. the only variable i know of is the stick. all things equal besides stick 1 or 2.

thanks for the input

EDIT#2: i just tried real fast before i have to leave running just the bad stick at 1.7v auto timing 1333 and 1600 and nothing but errors in memtest.

BlackOmega

That stick is definitely bad. It's RMA time.

That does suck though. Although I did have one of my Corsair XMS sticks go bad after only 20 days. And I never even OC'd it. Took them almost 3 weeks to get another kit back to me. Wasn't too happy about that.

__________________

morssolis

what is a safe voltage to run at? when i get 2 good ones i am planing on overclocking a little. maybe 1600 with lower timings or 1800? i read somewhere g.skill says its safe to go .1 over stock. is this ture? would i get away with like 1.72 or 1.74? i also read ripjaw does not overclock well on am3. although i had this good stick running at the 8-8-8-24 1600 at 1.56 if i recall.

Thanks for the help

BlackOmega

Well the voltage is determined by the sticks. The sticks I run are rated for 1.9v @ 1866 with 8-8-8-24 timings. If I set them lower to lets say 1066 they'll run at 1.5v all day long without a hitch. But if I want to run some tighter timings like 6-5-5-15 I need to up the voltage. As for max voltage that once again depends on the sticks. I have taken mine all the way up to 2.1v without an issue (3.5 hours prime blend stable). So it's just trial and error. How much voltage you can run through them is also determined by the IC (integrated circuits) on the sticks. Some like volts and some hate it.

Also, remember what I said before, your memory overclock is determined by your CPU ultimately.

With that said when you are overclocking before you even boot into windows, run memtest for at least 4-5 passes (all tests not just 5 & 8). If you get no errors, boot in to windows and run Prime95 for a minimum of 4 hours on the "Blend" test. It'll test your overall system as to how the CPU and the memory work when they're both under load. I've found that my RAM can run @ 1600 with 6-5-5-15-20 timings all day in memtest without 1 error. But as soon as I Prime95 it, it would fail. I was only able to manage to get 3.5 hours stability out of them. So I decided to drop them to 1333 with the tight timings and up my northbridge. The actual speed difference in negligible, and the throughput is almost identical. So not much of a loss.

__________________