• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Can't hit 4.0 but 3.8 no problem on q9550. Need help

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

OBLIVIONLORD

Member
Joined
Dec 11, 2003
These are my bios settings...

BIOS Version: 1406
CPU Multiplier: 8.5
FSB Frequency : 471
FSB Strap to North Bridge : 333mhz
PCI-E Frequency: 100mhz
DRAM Frequency: 1:1 @ 942mhz
DRAM Command Rate : auto
DRAM Timing Control: Manual
CAS# Latency : 5
RAS# to CAS# Delay : 5
RAS# Precharge : 5
RAS# ActivateTime : 15
DRAM Static Read Control: auto
Ai Clock Twister : Lighter
Transaction Booster : 10

CPU Voltage : 1.35
CPU PLL Voltage : 1.52
North Bridge Voltage : 1.30
DRAM Voltage : 2.10
FSB Termination Voltage : 1.32
South Bridge Voltage : Auto
Loadline Calibration : Enabled
CPU GTL Reference : Auto
North Bridge GTL Reference : Auto
Both CPU and Dram Spectrums: Disabled
CPU Margin Enhancement: Performance

I am 100% ram stable in Memtest at both 3.8 and 4.0 so that isnt an issue for me. The problem is this... I have adjusted the cpu voltage from 1.30-1.40 and prime95 fails within 1-6 minutes. Not even a slight hint that even at 1.4 there would be more stability than 1.3 to say the least. I have also done the same with the NB and FSB voltages from 1.3-1.4... still the same 1-6min fails in Prime95. I have set Transaction Booster anywhere from 10-15 and they all fail Prime95 with the 1-6 minutes just the same. I am absolutely lost at this point as to why that 20 fsb of a difference means a whole lot to 10 hours of Prime95 stability which is what I can achieve at 3.8ghz but not at 4.0ghz.
 
Lower your multi, and see how high you can get the FSB. Make small changes to the following:



CPU PLL Voltage : 1.52
North Bridge Voltage : 1.30
FSB Termination Voltage : 1.32
CPU GTL Reference : Auto
North Bridge GTL Reference : Auto
CPU Margin Enhancement: Performance




Go ahead and set this as follows:
DRAM Static Read Control: disabled

My guess is you need to play w/ the GTLs some. Do you have Clock Skew control? That might help as well.
 
i cant seem to figure out how to slow the ram down do a divider which will do that with this board. It appears that FSB strap is all i have to work with and im guessing that 400mhz is the way to go however it really doesnt slow the ram down compared to 333mhz.

Anyway I just did a 3.9ghz 460fsb and ran prime95 for 11 hours. It is perfectly stable. Now it's just that damn determining 100mhz to hit the ultimate 4.0 damn it!
 
Update.. at 3.95ghz, the first core to fail is within 23 minutes. Second core is at 7 hours.

I've tried messing around with the GTL values but it doesnt seem to do anything. The cpu, fsb, nb voltages do nothing to help. The margin enhancement does nothing when at optimized. Dram static control being at disabled does nothing to help.

Only thing i havent played with is the NB gtl values but i doubt that will do anything if increasing its voltage directly didn't do anything.
 
NB GTL would have been the 1st thing I would have tried.

More NB voltage doesn't always help. It's how that voltage is read that's more important. GTLs affect how the signal is read.
 
Last edited:
ill give that a shot

Just a question here... why are you at 3.91? Why not shoot for 4?
 
Believe me, I've tried. I can actually pass P95 small fft at over 4GHz, but this P35 NB is not stable. OCCT2.0 RAM test fails almost instantly at 471FSB. I can pass every test at 464FSB, but after a few BSOD I decided to give a little margin for error, and am at 460 now.

I want to get a P45 eventually, and when I do I hope to be able to hop over this FSB wall. I'd like to get to 4.2GHz+.

Edit: Glad you got it working! Congrats on the big 4.0!
 
One of the major things that limited my overclock was the mobos ram communication 'skew' if thats what its refered as.

I ws playing Supreme Commander for a few hours while at 3.8ghz and it closed on me which got me man. Another day it closed after 5 minutes. I then thought that I should try to figure out what the stability issue was. It wasnt the game itself but my system because Prime95 would fail after a few minutes when I was at 3.8ghz. I was able to pass IntelBurnTest 20 passes supprisingly which meant that it wasnt the processor. So I ran memtest and it failed within a few minutes. I then tested out 1 stick in chan A and it ran flawlessly. Same went with the other stick. However when I ran both sticks in Chan B slot 1, it errored within 1-6 minutes. After doing a bit of reading, I came across 'DRAM Skew' and thought I would advance its timing to see if that would help. Suprisingly it did after I changed it to '150ps'. I then ran the memtest at 3.8 and got gutsy at 4.0 and it pasted for 2 hours flawlessly.

Altering the NB frequency didnt change a thing to help me in memtest when at 460 and 470fsb.
 
I run 4x1GB RAM, and having all the slots populated seems to make a big difference in the 465+FSB range. I don't have any DRAM skew, just CPU, and NB...it helped a little, but nothing worthwhile.

It sounds like you have your RAM kind of tight. When going for 4GHz did you have your RAM timings (including tRD, and tRFC) loosened? I'm sure those 4GB sticks don't have the best timings.
 
When i stressed the ram itself through memtest it would fail at 6-6-6-18 while the fsb was at 450 - 470. The only thing that helped was advancing the Dram Skew on Channel B slot one which is what was occupied. If I didnt have that option then i wouldnt even been able to go to 3.8 stable.
 
When i stressed the ram itself through memtest it would fail at 6-6-6-18 while the fsb was at 450 - 470. The only thing that helped was advancing the Dram Skew on Channel B slot one which is what was occupied. If I didnt have that option then i wouldnt even been able to go to 3.8 stable.

Your set up is pretty close to mine, however i got all four slots occupy with that OCZ Reaper 2gig module. I relaxed the timing to 5-8-8-18 give it a 2.1v and plenty of cooling.

Never give it more than 2.1v coz the board will overvolt the ram alot.

did u enable the loadline calibration??
Did u enable the overvolting jumpers on the board??

I enabled both the jumpers and loadline calibration.

Idle Vcore is 1.33
prime95 Loaded vcore goes up to 1.35.

The vcore returns to 1.33 once it return to idle.

It keeps my 4.00ghz stable for 20hours prime95 small ftt.
 
Back