Ram speed limiting my OC?

faxon

Platinum Member
May 23, 2008
2,109
1
81
So, i havent done a whole lot of tinkering with my OC passed 4GHz. i have tried a good few configurations of voltages on the NB, FSB, and CPU, but i havent really pushed it super hard since most of the time i end up coming back empty handed. now, i havent tinkered with it in a while, and im sick, with a second working computer, so i have a stupid amount of free time on my hands. my question is simple, is there anyone out there who runs a GA-EP45-UD3P with a Q9550 or Q9650 with 4x2GB of DDR2 (im using 1066 G.Skill), that found lowering the memory divider down a notch or 2 (it's currently set so im getting 1068 @ 2v 5-5-5-15 with 445x9FSB). my cpu is a Q9650 running at 445x9 FSB @ 1.325vcore (1.296vcore actual) with LLC enabled. cooling isnt what i would consider a limiting factor, though the CPU does run at about 73c under linpack, but i cant afford better cooling. is it possible that i might increase northbridge stability for higher overclocks with all 8GB intact. i use a ton of ram on my system, easily exceeding 4GB, so it's not worth loosing the ram to increase the OC. i would however benefit from the increase in clocks, due to the amount of audio encoding im doing right now.
 

n7

Elite Member
Jan 4, 2004
21,303
4
81
Drop to 2.00D/2.00B (1:1) so your RAM is the same frequency as your FSB.
At least this way you can test for max FSB/CPU without the RAM being a limiting factor.

I did not like the UD3P (prefer my P5Q-D) with my 8 GB Mushkin...i found it to be very flakey @ higher FSBs, whereas it was great with 2x2 GB.
It would not POST higher than 470ish regardless of voltages/settings.

But it did pass stability testing @ 467 w/ 8 GB IIRC.

If you get GTLs sorted out, you should be able to do similar, or likely better, as my luck seems worse than most.

I'd suggest running 1:1 & lowering your multi so the CPU is out of the picture while you find out how far you can run with a higher FSB stably.

So:
7x450+ (DDR2-900+), & so on.
You'll mainly need to work on VTT + GTLs & MCH + GTL if just increasing VTT + MCH doesn't work.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
i got mine to boot into windows @ 500FSB with 4GB of ram on auto voltage a while back, so im thinking it's probably the northbridge rather than the FSB. either way, i will test this out. thx for the advice
 

Udgnim

Diamond Member
Apr 16, 2008
3,662
104
106
using the 2.0D strap might help. it'll have looser timings than whatever strap you're currently using

with all your memory slots used up, increasing NB voltage is pretty much a given. I honestly let the mobo's auto setting determine the NB voltage in BIOS, then use Easy Tuner to see what it's set at while testing for stability. if everything seems stable, then you can start decreasing the NB voltage and continue testing for stability

unfortunately, I don't know what the "safe" max threshold NB voltage would be
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
i set memory strap to 2.0D, CPU multi to 9, FSB to 456, CPU to 1.35v (1.31-1.33 in CPUz), NB and FSB to 1.38v, and CPU PLL to 1.59v up from 1.5. running OCCT now. i retested using linpack large data sets up to 500FSB with these voltages and it passed, with cpu multi set to 7
 
May 13, 2009
12,333
612
126
What speed are you going for? Is this the speed your gonna run your rig or you just benchmarking?
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
im trying to see what my max air cooled OC is on this chip. whatever that is i will back it off for 24/7 stability. i had it 24/7 @ 4ghz for 2 months but at sub 1.3vcore, i did some cooling changes and its running a fair bit cooler so now im going for a higher OC. i probably wont see much difference in day to day app performance, but i have about 3TB of music and movies to encode from various formats to another (compression, transcoding for ipod use, ect), and once its all ready and on my hard drive im just going to do a big batch run on it. getting a couple hundred more MHz out of my CPU could be the difference between getting it done while im at work and getting it done before i go to bed, when i would rather be gaming before bed. and yes, i also plan on benchmarking it lol.

ed: been an hour now @ 4.1ghz under OCCT. i normally test longer but im gonna try it @ 4.2 now since i never got past 5 minutes before at the same FSB/CPU multi, and my temps are stable at 66, 64, 61, 60 across all 4 cores. its surprising how much of a difference just changing my case to a HAF 932 made on over all system temps even with the door off

ed: passed 2 hours OCCT @4.2ghz with 1.36vcore CPUz (1.4 bios) using the anandtech review overclock voltages for everything else.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
couldnt get it stable @ 4.3, kept asking for more and more vcore, core 1 kept failing OCCT inside 5 minutes. tuned it back down to my 4.2ghz settings. quite happy with how easy it was vs how much i struggled getting it to post @ 4.2 before. will run OCCT over night but it already passed my 2 hour pass/fail standard test. i dont test with linpack since im running my setup in a medium/low airflow setup for noise, so while i have the cooler to do it with, i dont have enough airflow to keep the cpu cool while i run linpack, and the system eventually reboots or hangs from high heat failsafes
 

Shmee

Memory & Storage, Graphics Cards Mod Elite Member
Super Moderator
Sep 13, 2008
7,382
2,418
146
Sounds cool, grats with 4.2 :D

Update ur sig btw! :p
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
lmao dude you should have reminded me when you were over before newb :p

ed: and stop selling me shoddy motherboards! :D lol j/k