What am I missing? Q9550 (C1) at 3.6 unstable

KingstonU

Golden Member
Dec 26, 2006
1,405
16
81
Q9550 (C1 stepping) with P5Q Pro, rest of Rig in Sig. Was stable at 424 FBS (3.6) previously now fails Prime 95 after 20 seconds. But seems to be stable at 410 FSB . Recent changes: XP SP3 32-bit to Windows 7 64-bit, OCZ Vertex 120GB.

Below are my settings. What should I change?

AI Overclock tuner : Manual
CPU Ratio Setting : Auto
FSB Frequency : 424
PCI-E Frequency : 100
FSB Strap to North Bridge : 333
DRAM Frequency : DDR2-848
DRAM CLK Skew on Channel A1 : AUTO
DRAM CLK Skew on Channel A2 : AUTO
DRAM CLK Skew on Channel B1 : AUTO
DRAM CLK Skew on Channel B2 : AUTO
DRAM Timing Control : Auto

1st Information : (5-5-5-15-3-52-6-3)
2nd Information : (8-3-5-4-6-4-6)
3rd Information : (14-5-1-6-6)

CPU Voltage : 1.27500
CPU GTL Voltage Reference : 0.65
CPU PLL Voltage : 1.54
FSB Termination Voltage : 1.36
DRAM Voltage : 2.02
NB Voltage : 1.40
NB GTL Reference : Auto
SBridge Voltage : 1.20
PCIE SATA Voltage : Auto

Load Line Calibration : Enabled
CPU Spread Spectrum : Disabled
PCIE Spread Spectrum : Disabled
CPU Clock Skew : AUTO
NB Clock Skew : AUTO
CPU Margin Enhancement : Optimized

SpeedStep : Disabled
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
Try upping your CPU voltage a bit. 1.27 is pretty good but it's a C1 stepping which didn't OC as well as the E0 chips did. Was it stable at 3.5 at that voltage?
 

KingstonU

Golden Member
Dec 26, 2006
1,405
16
81
Yah it's stable so far at 3.5GHz, but my previous OC of 3.6 only needed 1.26 V CPU, now I need 1.275 for 3.5 to be stable, did 64bit Windows 7 cause that?

I am at 1.3 volts right now and am going to run Prime95, will post the result in a bit.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
sounds like it might just be a slight variance of the OC over time. given that you're doing 3.5 at what would be "stock voltage" more or less i wouldnt really complain. it's possible that the 64bit OS caused this, but given the other benefits of going to a 64bit OS i would hardly complain. that chip could probably do 4GHz with not much more boost in voltage to boot, so long as the board can handle it lol

ed: for some perspective, the Q9650 (E0 of course) was able to do 3.8GHz @ 1.25v (what i set as my starting voltage) without issue, linpack stable. to get to 4GHz i needed 1.296vcore, and to get 4.2GHz i needed to give it 1.376vcore. it's possible that over time the chip may have lost a tiny bit of its new luster, and it's even more likely that the board is simply wearing out slowly over time. either way, i wouldnt be to concerned unless it no longer passes @ stock
 
Last edited:

KingstonU

Golden Member
Dec 26, 2006
1,405
16
81
Wasn't able to do stress testing today cause I needed the PC to work on some other stuff, but will for sure tomorrow. If I can get 3.6 at 1.3volts I maybe not keep it since you say my board and/or chip are wearing down and I plan on keeping this system for another 1-2 years most likely. but I'll still benchmark to see how much of a benefit this Vertex is.
 
Dec 30, 2004
12,553
2
76
You've got a C stepping, that's the problem.

The HiK dielectrics break down over time too as electrons teleport into the middle of it and get wedged/stuck there. They are thinner than SiO2 dielectrics-->fewer atoms-->more likely to

nevermind it's too late for that.
 

KingstonU

Golden Member
Dec 26, 2006
1,405
16
81
Ok got it to 3.54 GHz at 1.3125 volts. Doing some benchmarking to see how much the Vertex and 7 create improvement over my old setup of 3.6GHz with XP and standard HDD.
 

KingstonU

Golden Member
Dec 26, 2006
1,405
16
81
My temps reach 69'C after 10 minutes of Prime 95 though! LOL And that's with the OCZ Vendetta 2 cooler with Freeze TIM. Maybe I'll try zip tying an additional fan to the back of the Vendetta 2.
 

dajeepster

Golden Member
Apr 15, 2001
1,974
16
81
power supply starting to go bad.... had it happen on two comps... new psu and i was off and running again.
 

Dopekitten

Member
Jul 11, 2008
67
0
0
Mess around with your GTL settings. Also, that crap about C1 stepping isn't really true. I've got a Q9450 stable at 3.8ghz. Here's what I have in Bios.

Vcore: 1.32 in bios, 1.296 actual idle, 1.312 actual load (i have LLC enabled)
VTT: 1.28
PLL: 1.50
GTL 0/2: .635
GTL 1/3: .67
vNB:1.28
vDRAM:2.1
 

Rifter

Lifer
Oct 9, 1999
11,522
751
126
Its the 64bit OS, you can get a higher OC running 32bit, usualy about 200Mhz higher from what i have seen, i tested this myself with system in sig with 32 bit and 64bit vista. Had to lower OC to 3.6 When i upgraded to 64bit.