First time ACTUALLY tweaking my Timings

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
I have read the theory on tightening timings for performance, but my current experience with actually going through the tedium of adjusting/testing timings has left me with a few questions that some of the more experienced overclockers may be able help me out with:

First of all, I am using 2x2gb G.Skill 9-9-9-24 1600mhz DDR3 on my Gigabyte 890GPA-UD3H motherboard, with a c3 965BE CPU running the show. Here is the exact memory kit:
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231193&cm_re=Gskill_4gb_ddr3-_-20-231-193-_-Product

The CPU is currently running at 18x210 (C&Q won't run at higher than x18 CPU multi, and I like the power savings of C&Q) and thus, my memory is currently clocked at 1680mhz. I tested this speed for stability at stock 1.5v and it passed.

My methodology has been to adjust the timings (one value at a time now, started with big jumps) and then do 5 passes of memtest without errors, before moving onto P95 Blend for no less than 2 hours without errors. Once the settings have been adjusted, I move on to further tightening. Am I being diligent enough in my testing? Some report using 8 passes of Memtest... some input needed here, hah.

Current settings that passed my testing:
1680mhz 7-8-8-21 @ 1.56v Max recommended voltage for the kit is 1.6v. My MB uses .015v increments past 1.5v. I am willing to go as high as 1.605v in my quest (unless someone can reasonably dissuade me from exceeding the max volts by .005)

I am going to have a good chunk of time in the next few days to try tighten it further. Any suggestions?
 
Solution
haha yeah getting chores done usually feels good :p

Well if we're talking your final stability check, 10 at the most. I'd be happy with less even 5 should give you a very good idea of stability.

Is that RAM the eco version? Surprisingly low voltage too. Mine are up at 1.65V.
Sounds good to me. Pretty impressive you're at higher frequency and lower timings than they're rated. If you set them too tight and crash it's not a big deal... your process seems exceedingly slow to me lol. I'd do blend tests and maybe a few passes of Intel Burn Test which stresses RAM as well to save time. Once happy, I'd run Memtest86+ for a final check.
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960


Haha, thats exactly why I decided to ask around about it, this memory has taken everything I have thrown at it. Going to be trying 7-7-8-21 right away, will give Intel Burn Test a shot in place of my 2 hours of P95 blend. It might be excessive, but doing these long stability tests actually gets my ass in gear in doing household chores and the like, hah, nevermind the peace of mind of knowing my kit is surviving extended stress.

I'll report back soonish~!
 
haha yeah getting chores done usually feels good :p

Well if we're talking your final stability check, 10 at the most. I'd be happy with less even 5 should give you a very good idea of stability.

Is that RAM the eco version? Surprisingly low voltage too. Mine are up at 1.65V.
 
Solution

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
Nope, not the eco version, considered it (1.35v i think) but a bit of web sleuthing revealed that they can be extremely finicky to get performing above stock levels. I was surprised at the low voltage as well, pleasantly so, hah.

Instead of adjusting the timings, I decided to up my FSB to 213 for 1704mhz at the same mentioned timings (as well as the slight bump in CPU speed). So far so good, IBT is running right now, and provided it don't crash, will move onto memtest for a couple runs.

Memory.jpg
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
Yup, first time I have ran G.Skill memory, I'm a believer~ Thanks for the input, gonna see how far I can go this weekend. I would love to see it stable at 7-7-7-20, but would even be happy if this is as far as it gets, heh.