to h2o or no? q6600 on 780i can't pass 3.0

S

supaspoon

Limp Gawd
Joined
Nov 3, 2008
Messages
382
So, I've been having trouble trying to get this thing to go any higher. It runs great at 333x9, but i just haven't had any real success with anything higher lately. A few months ago I had it at 3.4 for a while (when system was brand new), but i pulled it back when i went to 8g of ram (yes i need and use it). After i added the 2nd 4gigs it crashed a couple times so i went to 3.2, then it still crashed a couple days later. So i'd settled at 3.0 as it seemed rock solid stable.

I've thought since then that the bottleneck was likely my nb/ram (4 dimms, ddr2 800)

Anyway, I've managed to resist pulling the trigger on wc gear all week, so last night i thought i'd try to push the system regardless of temps (to a point of course) just to see if it would run at all. No point in wc'ing if it just absolutely refuses to go higher regardless.

To my surprise, w/ a considerable voltage bump i got it to boot and actually do stuff at 400x8. But it still throws errors immediately on prime. But anyway i figured that this was evidence that my problem was indeed just the proc. and not issues w/ having 4 dimms <-> NB voltage etc. etc.

NEwhoo before i finish writing a book here's some before / after settings.

333x9 = 3.0 @ 1.38v (bios, approx .04 droop = 1.33/1.34ish)
ram at 667, 5-5-5-15, 1.925v (ram's specced for 1.8v, but its cheap so i don't mind bumping it)
NB @ 1.6875 (I think, maybe 1.65, will verify when i get home)

I had to bump the ram up from the 1.8 and the NB up a bit from stock, otherwise it would occasionally give me a bsod mem dump.

So for the attempt:
400x8 = 3.2 @ 1.425 bios (approx 1.38ish)
ram at 800, 5-6-6-18 (<- will verify), had to bump it to 2.1v, won't run any lower.
NB voltage unchanged.

So at this it would boot up fine and let me fart around. I even shot off a render from 3dmax and let that run for a while, it seems to run fine (this is usually and easy indication for me if there's any memory problems, it would bsod very quickly). However the cores error on prime blend test immediately. I've pushed the cpu bios voltage all the way to 1.5, approx 1.46ish actual, and other than pushing my temps it does nothing.

So i'm just wondering, is there possibly a different component that i need to be adjusting? fsb voltage/sb, etc. etc.? I'd really like to put it under water, but i just can't justify it if it's still not going to go any higher than 3.0 even if i volt the living hell out of it. I really don't mind even putting the whole chipset under as well if it might help get this thing over the hump. In a perfect world I'd like to be able to cool it enough on h2o to just volt the crap out of it to 3.6.

Sys specs.:

q6600 slacr
xfx 780i
2x gtx 260's sli
850w tt psu

and lastly, thanks to those who actually had the patience to read my rambling. ;)
 
If you're pushing 1.5V Vcore and upwards of 1.6V on the NB and having trouble getting to 3.2ghz, I'm not sure water cooling is the answer. I think those 2 voltages should get most people to at least 3.2ghz w/ Q6600/Q6700 cpus. 1.5V Vcore should get you to 3.6ghz ...I run 1.48V for 3.6 and I"ve seen alot of people who run alot less Vcore at the same speed. Those are high voltages to not be getting stability at a relatively moderate overclock. But, I've been using intel chipsets the last 2 mobos (p35 and now p45) so I'm not entirely sure what is considered a high NB voltage on a 780 based mobo. Have you tried searching for general overclocking on 780i boards ? Do people have good results with the boards?

You could always give an x48 based SLI board a shot and see if that makes a difference...I know it's an expensive suggestion, but honestly if you can get 3.4 or 3.5ghz on air w/ another board, it makes more sense than blowing $200 on a wc system when your cpu won't get past 3.0.

In fact looking at the some of the user reviews of the Asus Rampage board, it looks like a number of people switched from nVidia based boards to X48 based boards..

just brainstorming :)
 
i am in the same boat, i think you just have a bad chip! i gave up on mine after going as high as 1.6 @ 3.1 Ghz and NB at 1.3 i think. igot a dud and you probably do too!

now i did order a new Asus P5Q Deluxe mobo, right now iu have a DFI x38 board which is "so so" for handling quads so maybe a new board could be the cure...

if your temps arent high water wont help.
 
Thanks a lot for your input & suggestions. I have looked around a bit for info on oc'ing w/ the 780, and the general concensus i get from a quick google & forum hopping is that the board shouldn't have any problems at all pushing the 6600. So i'm kind of at a loss there as far as confirming the "it's the board" idea. Not to say that's not the case, just that I haven't been able to find anything difinitevly saying so. Switching to an x48 isn't really an expensive suggestion compared to going under water anyway. But I just don't want to jump totally blind w/ fingers crossed & end up w/ zero results, wether that jump be for a different board or wc.

I guess it could just be a shitty chip too though, dunno. I read somewhere that some of the L8's are a crappy batch, & after hunting down my box for the serial # it is indeed an L8 :mad:. My workstation at work looks to have a 6700 in it, so if i get desperate i may sneak into the office this weekend & borrow it for some testing.

My temps are fine when i'm running at my current 333x9 setup. But w/ these higher voltages you bet your ass it gets hot quick. I'm just trying to see if i can get anything relatively stable with these higher voltages regardless (to a point of course, i'm not trying to instantly assplode anything) of temps in order to decide wether or not wc'ing would be a complete waste of money. I mean if the damn thing still wont give at 1.6v or something retarded anyway then there's just not point.

Anyway, I'll play around w/ it some more tonight & verify/add some of my settings (i'm at work so all these #'s are just from my head). It just seems odd to me that the entire range from 1.425v up to 1.5v looks to yeild the exact same results. Just seems like there's got to be something i'm missing.
 
What's your vid on that cpu? Like others said, watercooling isn't the answer...get yourself a cpu that will oc better.
 
Yeah it's kind of a tricky thing to try and figure out...I was going to suggest maybe it was the cpu. If you can "borrow" the work machine's Q6700 and get better results then you could just have a bad chip. You're right though..it's weird that you can't get anymore out of it from 1.425+.

You shouldn't have to micromanage the bios settings until 3.5ghz+. You said that you were running @ 3.4ghz with 2 sticks of ram right? How does the nvidia chipset handle 4 sticks?

My overclock managed to stay stable when I went from 4gb to 8gb..but again , diff chipset.
 
Vid is 1.325v

ok, i've gotten back home & been playing around w/ it for a couple hours now, & think i'm maybe getting somewhere. It seems like my ram refuses to run at funny numbers. it's spec'd to for 533/667/800. I had it in sync mode & at the 333x9 the ram was at 667, fine. Now I've unlinked it, & as long as the bus i set land the mem right on top of one of it's spec speed it seems workable, but otherwise i get freeze's, bsods, and non-boots. I don't know if that's common and I just happened to miss a key line of text in one of the n00b oc guides, or not. /shrug

Right now I just ran prime for about 20 minutes at 3.3ghz and the temps peaked at 68-69. Now that still a bit too high, as that's the threshold of my comfort level and i know rendering from 3dmax usually pushes temps about 5c higher than prime.

Anyway, here's the settings.

1471 - puts the mem at 667mhz & fsb @ 367x9 = 3.3gz
cpu - 1.4875v bios = actual 1.39/1.43 load/idle
mem - 2.1v
nb - 1.65v

mem's at 11:10 ratio, 5-5-5-15

Am i a total nublet not knowing the mem speed has to sit right on top of spec, or did i just get fussy sticks or something?

Anyway, yeah i was running it at 3.4 prior to adding the additional 4gigs. i just bumped my nb up to 1.65 and mem to 1.925 (right now at 3.3 i bumped the mem up to 2.1 but will see if i can get away w/ backing it off.).
 
Well, i've basically got it now at the above clock & have the cpu voltage in bios at 1.475, and backed the mem voltage back down to 2.0. Its running a fair bit hotter than i'd like for continuous use, but at least i know now that it can actually do it.

It's just weird though. was playing around w/ nudging it just a smidge higher, and i can run 1482 w/ the mem at 667, but at 1477 w/ mem at 800 it won't even boot.(nudged mem back to 2.2v for 800). :rolleyes:

Am i taking crazy pills? This being my first build i'm still a n00b at this & just not sure if it's normal, or if i'm just making this shit up in my head or what.
 
Well, I really got on a roll last night and i need some sleep. This being the first time I've had any bit of success w/ this thing in months I've just wanted to keep going & going. Unless I find reason to believe otherwise I've come to the conclusion that I'd just been suffering the consequences of buying shitty ram. It's seems to be a non-issue as long as i keep it around the 667 mark.

Anyway, I lapped my cpu/HS & remounted (jesus that took a long ass time). I pulled the board out to make damn sure I got the hs mounted nice and snug (it's a real pita to mount in the case). I also pencil modded my board, & the droop is now minimal over 1.45 and non-existant below.

So far I'm extremely impressed with the results from lapping. I thought it might give me 2-5c drop or so, but i'm now running 3.42ghz 380x9 and my temps are still a few c lower than they were when i was running 3.0 @ 333!!

And here's the current results.
3.42ghz @ 380x9, 1.456v load / 1.448v idle (this was the first setting that showed any fluctuation after doing the volt mod)
my hottest cores peaked at 62 and hover around 60-61.

oc.jpg
 
You must log in or register to reply here.
Back
Top