Overclock Suggestions

[tonyfreak215]

I have a E0 Q9550 on a P5Q-E motherboard with 8 gigs of 1066 GSkill Memory. I just received my NH-U12P cooler and decided to overclock to 3.4ghz. I decided on a moderate OC because I run this computer 24/7 and would like it to last 3-4 years.

My Settings:
400mhz FSB
400FSB Strap to NB
PCI-E 100MHZ
Ram Frequency 400mhz
5-5-5-15 timings

Voltages
CPU Voltage: 1.225
Both GTLs:.63
CPU PLL: 1.52
FSB Term :1.24
Ram:2.1
NB:1.24
NB GTL:Auto
SB:1.2
Sata:1.6

How do these sound? Are any of them too high to cause any problems?

The NH-U12P is a monster of a cooler. It idles at 33c and only reaches 60C while running the Intel Burn Test. (running a mid tower with an open side (won't fit on with the cooler)

Once I get my new case (Xclio windtunnel) I will post some pics.

 

[n7]

More than safe levels other than the RAM.

Something to note: basically all P5Q series tested have shown to overvolt vDIMM by about 0.08v.
You're running basically 2.18v into your RAM right now when it could be doing DDR2-800 5-5-5 at 1.8v BIOS or lower.

You can likely hit nearly 3.8 GHz range without increasing vCORE much, & with vDIMM in the < 2.02v range, you should be able to do your rated DDR2-1066 or higher.

 

[Phunk0ne]

I'm not sure if I find it odd, but an E0 9550 @ 3.4 with 1.225VCore, I find that rather odd, since my C1 9550 did just that with same VCore.shouldn't the E0 be able to hit that speed @ a lower VCore?

Correct me if I am wrong . . .
I know very well that each cpu differs from eachother just like twins do, but in overal, I assumed the E0 would hit same speeds @ much lower volts.

 

[n7]

Not necessarily.

On average, sure, an E0 will be better.

Also, assuming he's not using LLC, he's getting less than 1.2v under load due to vdrop/droop, likely around 1.18v...

 

[polarbear6]

well i hit a 3.6ghz with vcore at 1.2v(not 1.225) without vdrop on my e8400 with 100% stability
on a c0 stepping

 

[tonyfreak215]

Thanks for the replys.

This set up is Rock stable at 20 passes for the Intel Burn Test and my Max temps were only 64C around the 18th pass or so.

@ n7: I did not know about the ram over volt. I'll try dropping the voltage down to around 1.94. I have G. Skill PC2-8500 so it should be good with 2v.

I do have LLC enabled. All it does is prevent vdroop, right? Could it do any harm?
*edit* Did some more research on LLC. Anandtech suggest you turn it off as there is a risk to overvolting your processor.

I had the ram set to 1066 speeds, but then I had a 12:15 ratio. Isn't 1:1 generally what you shoot for?

For now I am happy with my 3.4ghz, now I just need to to work on lowering the voltages.

Do you think 5 passes with IBT be enough for the initial voltage test?



*Update*

I have lowered my RAM voltage to 1.94 and my CPU voltage to 1.2 and disabled LLC.

This gives me a CPUZ voltages:
1.168 at Idle with Temps at 31C.
1.120 at Full load (IBT) with max temps of 56/53/56/56 (Real temp set at 100 TJMax)

It was fully stable over 5 passes of IBT.

 

[n7]

I leave LLC on as i don't like having to run idle voltages so high (to compensate for vdroop under load).
Obviously, not everyone likes using LLC, but i don't believe it's an issue when used with reasonable voltages personally.

1:1 would be the last ratio i'd shoot for, as on the the P5Qs, that means higher tRD & the worst performance you can get with RAM.
If you can run 5:6 or 4:5 or 3:4 or higher, there's no reason not to.

You can switch to 3:4 (DDR2-1066) if you stay at your present OC of 8.5x400.

1.12v stable is amazing.

As i said earlier, you should really go higher
Lower RAM to 1:1 just for testing now (then increase ratio later if possible), & try to 8.5x425-450.
You should have 3.8 GHz easily with only a small amount more vCore.