Help getting 3.8ghz on Q9450 stable

[Dopekitten]

Alright so i've got my new Q9450, it's stable at 3.6ghz (prime 8 hours) with 1.2625 V in BIOS without LLC enabled. In that case, the only settings i changed were the CPU voltage and the FSB speeds, nothing else. (e.g. no VTT/GTL/NB Voltage)

So now i'm trying to get it stable at 3.8ghz. It boots to windows fine with 1.33 V in BIOS (CPU-Z reports 1.288 V). However, the first 2 cores fail Prime Small FFt's immediately (the others don't).

Now, i tried tweaking the GTL/VTT/NB voltage settings, however, anything above stock NB voltage (1.10 V) causes my system not to boot (no post even). This is the same with the VTT voltage, anything above 1.10 V and it doesn't post.

Any ideas why, and how to get my system stable?

Other information: Realtemp reports average of 60 C under prime full load at 3.6ghz, and my room temp is currently 23 C

System Specs:
-Asus P5Q-E
-Q9450
-4gb 1066mhz Corsair Dominators
-Nvidia Geforce 6800 (Sucky i know, ima get a 4870 soon enough)
-Xigmatek S1283 (with bracket and AS5)
-Antec 1200 Case (Fan's are all on max)

Help is appreciated, thanks

 

[aigomorla]

want my honest advice?

Be happy with 3.6

Maybe go for 3.7

3.8 is asking for too much to chew. If you were able to go this high, and not solve it yourself, its cuz you had a good chip.

Otherwise, theres too many things that can go wrong when you jump into bleeding edge territory.

 

[Dopekitten]

Alright then =P, how bout we work on 3.7ghz then.

I'll show you what i've set in BIOS

FSB: 463-with x8 multi for 3704mhz
NB to FSB strap: 333 (400 seems to cause more instability)
CPU Voltage: 1.3625
VTT: 1.20 (anything above (not going above 1.35) 1.20 does nothing for stability)
GTL 0/2: .635
GTL 1/3: .680
NB Voltage: 1.20 V

LLC: Disabled
(The rest is auto, except C1E and the likes)

So, i can boot (obviously), and for the short time Prime runs on all cores, temps are fine (using realtemp), however, after about 1 min in prime, the Core 1 fails. This can be fixed by fiddling with the GTL voltages (normally lowering the GTL 0/2 to .630 (original) does the trick), however then if this is done, then Core 0 will fail (after about a minute). The other core plod happily along with no problem. I've tried numerous things with GTL/VTT, but no luck.

Also, my VID is 1.25 V

 

[ShadowFlareX]

It could just be that your Q9450 won't go much over 450MHz FSB, perhaps that's your Q9450 FSB limitation. I know mine can't even boot to windows at 463MHz so I just backed down to 450MHz. To even having it running at 3.6GHz with 1.2625V without LLC is wow. Mine won't take anything lower than 1.4V in BIOS without LLC.

Try lowering the multi to 7 and FSB 463MHz, see if you can run Prime and stuff. This is just to test if your Q9450 can really do 463MHz FSB. And you're right about that setting NB to FSB strap 400 giving you instabilities, same here. I just set mine to Auto even.

 

[Dopekitten]

I'm not exactly sure what the NB to FSB strap is supposed to do. Is it the base FSB then? I don't know, and i don't see the advantage of increasing it...

Edit: Update, at 3.6ghz, its stable all day long, but then if i stop prime, wait an hour, then start it again, it fails evreything. Clearing CMOS/rebooting and resetting the same settings in BIOS fixes the problem. Any ideas on why this happens?