[Solved] Help: getting RAM to 2000 stable (no more BSOD pls)
-
Yesterday I felt like overclocking the CPU (i7 960) a little and I managed to leave it at 4.2ghz (21x200) and it lasted 6h of Prime95. The problem I have is with the RAM (OCZ platinum 3x2Gb 1866 9-9-9-28), I had it for quite a while at 2000 but I don't remember the configuration and now I can't find it to leave it stable (6-7 minutes in prime95 and BSOD). The configuration that so far lasts the longest is 10-10-10-30 2T at 1.66v and with the QPI/DRAM at 1.55v. Let's see if someone knows how to leave the system stable. Thanks in advance.
-
Hello and welcome.
What is the reference of those modules? -
Hello! It is one of the first triple channel models that there were when the first i7s came out. Anyway, when I get home from work I will take a picture of the module and put it here, although I have searched for them on the OCZ page and never managed to find them.
Goodbye!
-
Here is the photo of the RAM with all the data:
With the CPU at 4.2Ghz I pass all the tests (prime95, intel burn test, occt) but the moment I touch the RAM sometimes it crashes and sometimes it doesn't…
The truth is I don't know what I'm doing wrong, I don't know if I'll have to touch some other voltage or something
Thanks and goodbye!
-
"…
With the CPU at 4.2Ghz I pass all the tests (prime95, intel burn test, occt) but the moment I touch the RAM sometimes it crashes and other times it doesn't...
..."In short, it's not stable in this case.
Relaxing the latencies to 10 doesn't seem to be enough, you'll have to try with more relaxed latencies or increase the voltage of the modules a little, but keep in mind that they are already increased compared to the JEDEC standard which is 1.5v.
It also depends a lot on the chips that these modules have, try to find out to have more information.
And keep in mind that RAM is more delicate to touch than the micro, be careful with the excess voltage, your modules already use a voltage +10% over the standard and the highest I've seen in series is +16% and they heated up noticeably more.Salu2.
-
Let's go, it's not stable for the case.
Relaxing the latencies to 10 doesn't seem to be enough, you'll have to try with more relaxed latencies or increase the voltage of the modules a little, but keep in mind that they are already increased compared to the JEDEC standard of 1.5v.
It also depends a lot on the chips that these modules have, try to find out to have more information.
And keep in mind that ram is more delicate to touch than the micro, be careful with the excess voltage, your modules already use a voltage +10% over the standard and the highest I've seen in series is +16% and they were heating up noticeably more.Salu2.
I'm going to try to relax the latencies then. About the voltage, since these are the ones that come at 1.65, I increased them a little thinking it wouldn't be much. If that, I'll put them at 1.65 which are the recommended ones and continue from there.
Thanks and regards! -
By the way, I advise you to use a performance test for the ram, to check what each of the speed and latency configurations really yields.
For example, if you get 2000Mhz with cl11, you would have to see if you are not really losing net performance compared to 1866 cl9.Salu2.
-
After many tests and tweaks, it seems that I have managed to leave it stable at 9-9-9-24 2N@2000Mhz.
I was advised to touch some voltages (QPI, IOH, ICH and CPU PLL apart from increasing the RAM voltage a little more to take into account the Vdrop that my motherboard has) and it works perfectly.
12 hours of prime95 and more than one memtest later, the computer is working perfectly. At 4.2, after several hours of prime, it does not exceed 75º.By the way, having it stable, I tried to increase the latencies to check the performance with different latencies at 2000Mhz and the truth is that there was hardly any difference in synthetic tests, so I imagine that in everyday use it won't even be noticeable.
Best regards and thanks!!!
