As I see a few P5N-T threads appearing and am having difficulty with my own board I thought I'd make this thread and try and turn it into a general P5N-T overclocking experience thread… or pit of depression, whichever you'd like!
The idea being that everyone share their current specs, settings and clocks so hopefully we can all get a bit more from our setups, particularily with Quads and high FSBs, or at least confirm its limitations…..
So, I have a Q9550 and the P5N-T (BIOS 1303 I think), 4 sticks of 1066 RAM and 2 8800GTs. Try as I might I cannot get this board remotely stable past 1685 QFSB. At 1650-1685 QFSB (3.5-3.58Ghz) it will complete a 3dmark run but at current settings is not prime95 stable. I did manage to get windows loaded with 1700Mhz QFSB but it crashed soon after. Here are my results so far:
(All other voltages set to auto, all C1E etc turned off. Memory timings are set to 5-5-5-15@1066 or 4-4-4-12 when between 800-900Mhz.)
It seems the N/b volts have to go pretty high on this chipset, in Jdocs' response to my first thread he mentions using more than 1.7v on his northbridge which seems huge.
At my 3.4Ghz clock it would boot with lower n/b volts than 1.48 but I actually found a several hundred point increase in 3dmark06 score when going from 1.42v-1.48v (but none from going any higher) so that's where I left it.
I am not a patient overclocker either and I haven't properly tested for stability at lower volts, so it may be that I could use less on the CPU and the northbridge at the 3.4Ghz clock.
I would really like to get somewhere up near my target speed of 3.6-3.8Ghz as I know the Q9550 should easily be capable of it on good air (TRUE). I shall be trying over this weekend and hopefully will be able to post something positive.
If you have an Asus 780i board then please post your experiences!
I run my Q9550 @ 3.4 on a gigabyte DQ6 motherboard.
To be honest i don’t see any purpose to push it any harder. I am happy with the gains i have, i don’t need to push the processor to destruction… @ 3.4 the processor runs stable, and temps are mid/high 40’s, It will run like this without reducing the life expectancy.
@ grieve... Ok, well thanks for that, useful comment. I personally want to squeeze a bit more out of my system to try and smooth out my frames for full HD gaming, so I do want to overclock past 3.4! Anyway..
Using a slightly more methodical system of testing I made a little progress last night. I firstly tried to get the lowest stable vcore on the 3.4Ghz clock which I found to be 1.2875v (in the BIOS), only one setting lower than I had it before.
Then I managed to stabilise 3.51Ghz (1650 QFSB) with 1.3125v vcore and 1.56v on the Northbridge, RAM at 825Mhz (linked-sync) with 2.11V and 4-4-4-12 timings. (When I say stable I was able to complete a 3dmark06 run whilst running 4 core small FFTs in prime95 - I thought this was a good place to start for short-term testing!)
With that setup 'stable' I changed the RAM divider to 'Auto' which set the RAM at 990Mhz, I kept the 4-4-4-12 timings which booted fine and I re-ran 3dmark and played a few levels on Dirt without a hitch before hitting the hay.
During all this I monitored my temps and all cores were 61c or below (prime) and the n/b didn't go above 45c, hovering around the 42-43 mark under load.
I'm hopeful now that I will get to 1700mhz QFSB. I notice there is a new BIOS available from Asus, v 14xx, so I think I will try that as I'm on 1303. I think I will also look into the VTT option as many have mentioned this as key to a stable overclock.
Yesterday I updated the BIOS to 1402 and initially still couldn't get past 3.51Ghz. Looking again at my core temps I thought 61c (hottest core) was a little high. Also ran Everest and that was reporting my MCH temp as 61c, when Speedfan's temps 2&3 (of which I assumed one was the n/b) were both around 43c. Alarm bells rang and I whipped the case open and got the TRUE off. I sanded the base of the TRUE (as per lapping guide) which took ages to get even remotely flat using p400 then p800 paper. By the time I'd finished (given up!) there were still small strips of nickel plating on each side that hadn't completely gone, even though I'd been sanding swathes of copper off the middle. Shows just how convex these TRUEs are when you get them! Thermalright should clearly be machining the bases after they solder that bottom plate on.
Anyway, replaced the TRUE and manged to dig out a 10mm x 60mm fan which would squeeze under the TRUE and cable tied that to the n/b heatsink. Hottest core was now down by 5c since the lap (even with fresh goo) and n/b down to 55c when under prime.
Got 1685 QFSB (3.57Ghz) stable with all volts except RAM set to auto, then this morning with auto volts it cold booted at 1700 straight into Windows and ran for about 5 mins before freezing. 3.6Ghz stable by the end of the week? Fingers crossed!