Timing OC?

lemon07r

lemon07r

New Member
Can timings be OCed? Soon I will be getting a GeIL Ultra Plus 7-7-7-24 1600mhz 1.6v ram. It has a good heatsink and DBT so it should be able to handle heat. Would I be able to bump down the timings by increasing the voltage or something like that? And could some one give me a good estimation?
 
Yes you can run RAM at tighter timings than advertised, however as you mentioned you generally have to increase the voltage. There's no set plan for how to do it, you just need to adjust things until it works or you don't want to take it any further. Generally, I wouldn't expect you to get much better than 7-7-7 on DDR3-1600. You might be able to go to 6-6-6 but if antything I'd suggest OCing the frequency instead and either loosening the timings to get the RAM faster or see how high you can get it at 7-7-7.
 
Thanks for the info and quick reply. I'm thinking of bumping it from 1.6v to 1.7v to bump down the ''24'' in 7-7-7-*24*. How far would I be able to get. (an estimate would be nice cause I don't want to put it to low and damage something)
 
You shouldn't be able to break anything by putting the timings too low. At worst you will just need to clear the CMOS if it won't boot. You can probably take that one to 21 without blinking but it's also one of the less effective timings for performance increases. Have a read through http://www.techwarelabs.com/reviews/memory/memory_timings/index_3.shtml keeping in mind these tests are on RAM with a lot fewer clock cycles to burn on timing delays (PC100-166)
 
Ok, i see now, but the articles seem a little biased when it comes to ddr3, it'd be nice if the had an articles to show the boost from something like cl8 to cl7 or 1333mhz to 1400mhz.
 
It's not really biased, most of the measurements are speed neutral. All else being equal changing some timing from x to y creates z increase in performance. It shows that clock speed has the biggest effect on throughput by a long shot. Now, you are correct that raw throughput isn't the only factor in memory performance but the delay on CAS 2 DDR400 is 10 nanoseconds and the delay on CAS 7 DDR1600 is 8.75 nanoseconds. Plus new CPUs have better systems to hide RAM latency.

We know from the article that dropping the CAS by 1 equates to a 0%-0.002% increase. That's well within the tolerance for measurement error so lets call it 0. The speed increase from frequency is easily predicable. For 1333 to 1400 you are looking at a peak increase of 533MB/s
 
Last edited:
You must log in or register to reply here.
Back
Top