Q9450 OC / Temps / Settings

[LUCI5R]

More NB volts is my suggestion. You have 4GB of ram and are trying to o/c a quad.

Yeah, I have been suggested (over in the P5Q thread) to up CPU Voltage to 1.400v and NB Voltage to 1.28v, so I'll give that a shot later today.

What's the max I should go to on NB?

Also, I'm wondering if 1.98v is enough for DRAM Voltage, or if I need to up that.

Lastly, I've been using a 1:1 divider. I wonder if I need to use a different divider.

[LUCI5R]

How far did you get with the FSB Term and NB voltages on their lowest settings?

I started FSB Term @ 1.30v and NB @ 1.20v and I could get 3.6Ghz stable with that (Didn't prime for 24hrs though).

[jas420221]

I would say your memory is fine as you arent even running it at its stock speeds yet (500Mhz). So if 2.0v is stock voltage, Im sure you are fine on that.

[rdv]

For the test run that accompanied the pictures here were the settings:
CPU Voltage: BIOS:1.35v Real: 1.344v
CPU GTL Voltage Reference (??): BIOS: AUTO
CPU GTL Voltage Reference (??): BIOS: AUTO
CPU PLL Voltage: BIOS: AUTO BIOS Hardware Monitor:1.648v
FSB Termination Voltage: BIOS: 1.34v Real:1.28 to 1.30
DRAM Voltage: BIOS: 2.1v Real: 2.14 to 2.16
NB Voltage: BIOS: 1.34v Real: 1.52 to 1.55
NB GTL Reference: BIOS: AUTO Real: ??
SBridge Voltage: BIOS: AUTO Real: 1.07
PCIE SATA Voltage: BIOS: ?? (I will assume AUTO)

All "Real" quoted are under prime95 load. I saw ranges of voltages in some cases during the 24hr test.

The pattern I used to find stability was to raise the vNB as high as was safe and my cooling allowed, then step by step upped my Vcore voltage until 24hr prime95 passed. Then I went back and low balled the vNB step by step upping it until the sweet spot was found for 24hr stable.

I did that for 3.6GHz and now trying to finish 3.7Ghz. One other thing is I would have never reach a stable "safe" 3.7Ghz without the Pencil Vdroop mod. My Vcore would have been way too high, I know because I tried before applying that mod. Here is the link, it is only for some ASUS boards: Pencil Vdroop Mod - (I used a 3B pencil) http://forums.techpowerup.com/showthread.php?t=60341

You have to have allot of patience as I have had prime95 failures after 7, 12, 13 and even 20.5 hours. The 20+ almost killed me but I want to stick to the plan.

Assuming my latest stability test passes the 24hr mark the only "unsafe" for 24/7 voltage in my opinion is my vPLL (1.648) which is +0.048 over the guidelines I captured from this thread and some others.

I have queried on this thread for some advise but so far have not received any replies.

[bigslappy]

;p sub'ed

[LUCI5R]

Some of the voltages you have are 'slightly' higher then what I'm using, but well within guidelines & safe limits. I think I'm going to give your settings a shot "If" I'm unable to attain stability using some other recommendations I've recieved in the P5Q thread (basically lower volts then some of yours).

The only one I myself would have pointed out - which you already did - is the vPLL, which is slightly higher then recommended. I'm not sure for 24/7 usage I'll cross any guideline - but if temperatures keep within limits, I'll certainly won't mind hitting the max "safe".

Yes, patience is definately the 'key' with Priming ... so is 'retries'. I'll give these few things a shot later today and hopefully tomorrow I'll have something better to report.

This is really helpful. Thanks!!

Edit: Do you keep Load Line Calibration ENABLED or DISABLED?

For the test run that accompanied the pictures here were the settings:
CPU Voltage: BIOS:1.35v Real: 1.344v
CPU GTL Voltage Reference (??): BIOS: AUTO
CPU GTL Voltage Reference (??): BIOS: AUTO
CPU PLL Voltage: BIOS: AUTO BIOS Hardware Monitor:1.648v
FSB Termination Voltage: BIOS: 1.34v Real:1.28 to 1.30
DRAM Voltage: BIOS: 2.1v Real: 2.14 to 2.16
NB Voltage: BIOS: 1.34v Real: 1.52 to 1.55
NB GTL Reference: BIOS: AUTO Real: ??
SBridge Voltage: BIOS: AUTO Real: 1.07
PCIE SATA Voltage: BIOS: ?? (I will assume AUTO)

All "Real" quoted are under prime95 load. I saw ranges of voltages in some cases during the 24hr test.

The pattern I used to find stability was to raise the vNB as high as was safe and my cooling allowed, then step by step upped my Vcore voltage until 24hr prime95 passed. Then I went back and low balled the vNB step by step upping it until the sweet spot was found for 24hr stable.

I did that for 3.6GHz and now trying to finish 3.7Ghz. One other thing is I would have never reach a stable "safe" 3.7Ghz without the Pencil Vdroop mod. My Vcore would have been way too high, I know because I tried before applying that mod. Here is the link, it is only for some ASUS boards: Pencil Vdroop Mod - (I used a 3B pencil) http://forums.techpowerup.com/showthread.php?t=60341

You have to have allot of patience as I have had prime95 failures after 7, 12, 13 and even 20.5 hours. The 20+ almost killed me but I want to stick to the plan.

Assuming my latest stability test passes the 24hr mark the only "unsafe" for 24/7 voltage in my opinion is my vPLL (1.648) which is +0.048 over the guidelines I captured from this thread and some others.

I have queried on this thread for some advise but so far have not received any replies.

[rdv]

My Load Line Calibration is ENABLED. From what I read once the Pencil Vdroop mod is performed that setting does not matter much.

As for lower voltages on some of my setttings I tried many combinations. So far I have tried 42 test runs and 20th distinct set of BIOS settings. That excludes any settings/tests that could would not post or pass 10 minutes of prime95 and there were several of those failures.

Good luck

Edit: Last test just failed after 19hr 45min. Oh well have to up the vNB again and retest. At least vNB are still within safe limits.

[bigslappy]

I've got 2 Rampage Formula boards now & enuff parts to do another build soo I'm getting a Q9450 tomorrow for the 2nd build
got a QX9770 on the other
gonna water them both all over
Qx9770 is on the test bed now with a DD tdx water cpu block
I have had it up to 4 Ghz soo far

[LUCI5R]

My Load Line Calibration is ENABLED. From what I read once the Pencil Vdroop mod is performed that setting does not matter much.

As for lower voltages on some of my setttings I tried many combinations. So far I have tried 42 test runs and 20th distinct set of BIOS settings. That excludes any settings/tests that could would not post or pass 10 minutes of prime95 and there were several of those failures.

Good luck

Wow, hats off to you. You're definately a patient man.

Two things that I've wondered about - one is of course Load Line Calibration. I've tried both ENABLED and DISABLED, and I really don't know if it made any difference either way, so I still don't know which one I should use.

Second is divider. I've been using 1:1, and I'm wondering if there's any reason I should deviate from it, and give, say, 5:6 a try! Not sure.

[jas420221]

Wow, hats off to you. You're definately a patient man.

Two things that I've wondered about - one is of course Load Line Calibration. I've tried both ENABLED and DISABLED, and I really don't know if it made any difference either way, so I still don't know which one I should use.

Second is divider. I've been using 1:1, and I'm wondering if there's any reason I should deviate from it, and give, say, 5:6 a try! Not sure.

LLC is for voltage droop on load. You should be using CPUz to figure out the size of your droop.

I already answered the divider question for you. You arent even close to 500Mhz which is your rated speed on those sticks. 1:1.

[LUCI5R]

LLC is for voltage droop on load. You should be using CPUz to figure out the size of your droop.

How can I do that? Sorry, probably a n00b question - but I really haven't dwelled deeper into droop.

I already answered the divider question for you. You arent even close to 500Mhz which is your rated speed on those sticks. 1:1.

So basically I should stick to a 1:1 divider till reach at least 500Mhz or above?

But I guess then my question is - I can use my sticks @ 500Mhz or above, but then the divider won't be 1:1 ... so which one should I opt for? Running my sticks @ 500Mhz & above .. .Or. .. running a 1:1 divider?

[jas420221]

1:1 is always best. Running a higher divider, will net you results in benchmarks, but no real performance gains.

DL and install CPUz, on the first page, it shows the voltage, then fire up P95 on all 4 cores and watch the voltage droop, thats vdroop and LLC attempts to eliminate that.

[LUCI5R]

1:1 is always best. Running a higher divider, will net you results in benchmarks, but no real performance gains.

Sounds good - I'm not looking for benchmarks - so I'll stick to 1:1

DL and install CPUz, on the first page, it shows the voltage, then fire up P95 on all 4 cores and watch the voltage droop, thats vdroop and LLC attempts to eliminate that.

I have CPU-Z, I never realized you could measure droop this way. If there is a considerable droop - I should keep LLC ENABLED, right? Is there a measure of what should be considered a considerable droop?

[jas420221]

Optimal is none, but to say .08 or .03 use it or not...I cant say. Its up to you, I would leave it enabled personally.

[LUCI5R]

Optimal is none, but to say .08 or .03 use it or not...I cant say. Its up to you, I would leave it enabled personally.

I'll try both ways and see where I'm getting at with stability.

Thanks!

:P

[rdv]

... Second is divider. I've been using 1:1, and I'm wondering if there's any reason I should deviate from it, and give, say, 5:6 a try! Not sure.

The guys in this forum are giving you better advise than I could, because the memory 5:6 divider value came from AUTO BIOS settings except for the vDRAM which is set to 2.1v. I had tried the recommended 1066Mhz settings from the Corsair memory forum but they would not post with a "DET DRAM" message when I started to OC. The recommended settings were no problem on stock settings. In fact my first successes with OCing was when I returned my memory settings back to AUTO except for the vDRAM. I will admit that I could try tweaking my memory settings but unless the benchmark testing shows issues I will leave them alone.

I have started my next stability test with a slightly higher vNB. I will see if these settings are the stable ones.

[LUCI5R]

Instability! Can't get past it.
I used some of the advice I got from you guys last evening and I was able to push the FSB up a bit. Right now I'm sitting @ 3664Mhz (458 x 8). The thing is, it's still not 24hr Prime95 stable - what's happening is that at least 1 core gives me "worker thread stopped" while the others continue without failing. One of the 4 cores always gives me (and they rotate) "worker thread stopped".

Also, I attempted a video encode and early on got the BSOD (which is a first for me during an overclock attempt) and I'm assuming that's from a failed overclock.

This is where my settings are at right now:
(I've bold the ones I upped/changed last evening)

CPU Ratio Setting: 8.0
FSB Strap to North Bridge: AUTO
FSB Frequency: 458
PCI-E Frequency: 101
DRAM Frequency: DDR2-917 (1:1)

CAS# Latency: 5
DRAM RAS# to CAS# Delay: 5
DRAM RAS# Precharge: 5
DRAM RAS# Activate to Precharge: 18
RAS# to RAS# Delay : AUTO
Row Refresh Cycle Time: 55
Write Recovery Time: AUTO
Read to Precharge Time: AUTO

DRAM Static Read Control: DISABLED
DRAM Read Training: AUTO
MEM. OC Charger: AUTO
AI Clock Twister: AUTO
AI Transaction Booster: AUTO

CPU Voltage: 1.3875v
CPU GTL Voltage Reference (0/2): AUTO
CPU GTL Voltage Reference (1/3): AUTO
CPU PLL Voltage: 1.58v
FSB Termination Voltage: 1.34v
DRAM Voltage: 1.96v
NB Voltage: 1.30v
NB GTL Reference: AUTO
SBridge Voltage: AUTO
PCIE SATA Voltage: AUTO

Load Line Calibration: DISABLED
CPU Clock Skew : AUTO
NB Clock Skew : AUTO

Right now all I'm looking for is 24hr Prime stable @ 458 - 460 FSB .. nothing more then that! But I'm just wondering if there's any more room up there, any more settings up there I can fiddle with that can help me.

Any advice?

Thanks guys!!

[Cob]

I would seriously try to work out your GTL settings. Taking them off Auto and setting them manually can make all the difference.

Start at 0.62x and work your way up to about 0.72x. You'll also likely find that both won't be the exact same setting.

[rdv]

Here is all the BIOS settings I had for the stable 463FSB:

Test Sessions BIOS Settings: #17 3.7G 463FSB STABLE
Extreme Tweaker menu:	
CPU Level Up [Auto]	Auto
Ai Overclock Tuner [Auto]	Manual
Super Memory Profile [Profile #1]	
OC From CPU Level Up [Auto]	Auto
CPU Ratio Setting [Auto]	Auto
FSB Strap to North Bridge [Auto]	333
FSB Frequency [XXX]	463
PCIE Frequency [XXX]	100
DRAM Frequency [Auto]	Auto
DRAM Command Rate [Auto]	Auto
DRAM CLK Skew on Channel A/B [Auto]	Auto
DRAM Timing Control [Auto]	Auto
	
1st Information : 4-5-5-15-3-42-6-3 (These values are auto-detected)	CPU-Z says    5:6 1112MHz
CAS# Latency [5 DRAM Clocks]	5
RAS# to CAS# Delay [5 DRAM Clocks]	5
RAS# PRE Time [5 DRAM Clocks]	5
RAS# ACT Time [15 DRAM Clocks]	18
RAS# to RAS# Delay [Auto]	Auto
REF Cycle Time [Auto]	Auto
WRITE Recovery Time [Auto]	Auto
READ to PRE Time [Auto]	Auto
	
2nd Information : 8-3-5-4-6-4-6(These values are auto-detected)	
READ to WRITE Delay (S/D) [Auto]	Auto
WRITE to READ Delay (S) [Auto]	Auto
WRITE to READ Delay (D) [Auto]	Auto
READ to READ Delay (S) [Auto]	Auto
READ to READ Delay (D) [Auto]	Auto
WRITE to WRITE Delay (S) [Auto]	Auto
WRITE to WRITE Delay (D) [Auto]	Auto
	
3rd Information : 13-5-1-5-5 (These values are auto-detected)	
WRITE to PRE Delay [Auto]	Auto
READ to PRE Delay [Auto]	Auto
PRE to PRE Delay [Auto]	Auto
ALL PRE to ACT Delay [Auto]	Auto
ALL PRE to REF Delay [Auto]	Auto
	
DRAM Static Read Control [Auto]	Auto
Ai Clock Twister [Auto]	Auto
Ai Transaction Booster [Auto]	Auto
Pull-In of CHA PH1	
Pull-In of CHA PH2	
Pull-In of CHB PH1	
Pull-In of CHB PH2	
	
CPU Voltage [Auto] (Vcc)	1.35000
CPU PLL Voltage [Auto]	Auto
North Bridge Voltage [Auto] (NBCore)	1.51
DRAM Voltage [Auto]	2.1
FSB Termination Voltage [Auto] (VTT)	1.34
South Bridge Voltage [Auto] (SBCore)	Auto
SB 1.5V Voltage [Auto]	Auto
Loadline Calibration [Auto]	Auto
CPU GTL Voltage Reference [Auto]	Auto
NB GTL Voltage Refernece [Auto]	Auto
DRAM Controller Voltage REF [Auto]	Auto
DRAM Channel A Voltage REF [Auto]	Auto
DDR2 Channel B Voltage REF [Auto]	Auto
Voltiminder LED [Enabled]	Enabled
CPU Spread Spectrum [Auto]	Auto
PCIE Spread Spectrum [Auto]	Auto
	
	
Advance Menu/CPU Configuration:	
CPU Ratio Setting [Auto]	Auto
C1E Support [Disabled]	Disabled
CPU TM Function [Enabled]	Enabled
Vanderpool Technology [Enabled]	Disabled
Max CPU ID Value Limit [Disabled]	Disabled
Intel Speedstep Tech. [Enabled]	Disabled
	
	
Advanced Menu/North Bridge Chip:	
Memory remap (for 64bit OS's) [Enabled]	Enabled

Sorry about the formatting but the data comes from a spreadsheet. I did not want to submit one of those long posts.

I do not remember you mentioning is whether or not you applied the Pencil Vdroop mod. Here is a post where I think your mobo (P5Q Deluxe) was mentioned being moded: http://www.ocforums.com/showthread.php?t=557776
If you can I definitely recommend the mod with a 3B pencil.

I hope this helps.

[jas420221]

Instability! Can't get past it.
I used some of the advice I got from you guys last evening and I was able to push the FSB up a bit. Right now I'm sitting @ 3664Mhz (458 x 8). The thing is, it's still not 24hr Prime95 stable - what's happening is that at least 1 core gives me "worker thread stopped" while the others continue without failing. One of the 4 cores always gives me (and they rotate) "worker thread stopped".

This screams GTL settings at this point. Up the GTL on the core that gave the error first. Its all trial and error now!

[LUCI5R]

I would seriously try to work out your GTL settings. Taking them off Auto and setting them manually can make all the difference.

Start at 0.62x and work your way up to about 0.72x. You'll also likely find that both won't be the exact same setting.

Cob ... Agreed 100%. In fact, I'm going to be taking quite a few things Off of AUTO and start playing with actual values - GTL's being the first. Quite a few successful Q9450 OC'ers are using these settings - that's one difference I realized.

I do not remember you mentioning is whether or not you applied the Pencil Vdroop mod. Here is a post where I think your mobo (P5Q Deluxe) was mentioned being moded: http://www.ocforums.com/showthread.php?t=557776
If you can I definitely recommend the mod with a 3B pencil.

Actually I haven't applied the Pencil Vdroop Mod yet - I don't know anyone using a P5Q Deluxe who has, and I really didn't know if you need to or not. Actually I'll make a post in the P5Q Deluxe and see if anyone knows anything about this.
I went through your link but I couldn't spot anything pertaining to a P5Q deluxe. Did I miss it? If it's helpful & needed in the P5Q ... I'm definately doing it. Will get a 3B as you recommend.

Couple of Questions regarding your BIOS settings:
Doesn't AUTO on CPU PLL take it too high .. towards unsafe limits? I haven't tried it on AUTO yet because I wasn't sure you could.
Same with NB ... I'm not sure what's the max supported for your board but I was told for the P5Q not to take it over 1.4
Only other thing is, you have CPU TM Function ENABLED .. does that help towards stability at all?
Other then that your settings look worth a shot - I'll give it one!

Thanks guys!!

Edit:
This is what I'm planning on trying tonight ...

JumperFree Configuration Settings
AI Overclock tuner MANUAL
CPU Ratio Setting 8
FSB Strap to North Bridge 400
FSB Frequency 458
PCI-E Frequency 101
DRAM Frequency DDR2-917
DRAM Timing Control MANUAL

1st Information
CAS# Latency 5
DRAM RAS# to CAS# Delay 5
DRAM RAS# Precharge 5
DRAM RAS# Activate to Precharge 18
RAS# to RAS# Delay AUTO
Row Refresh Cycle Time 55
Write Recovery Time AUTO
Read to Precharge Time AUTO

2nd Information
All AUTO

3rd Information
DRAM Static Read Control DISABLED
DRAM Read Training DISABLED
MEM. OC Charger ENABLED
AI Clock Twister MODERATE
AI Transaction Booster MANUAL
Common Performance Level 10

CPU Voltage 1.400v
CPU GTL Voltage Reference (0/2) 0.67
CPU GTL Voltage Reference (1/3) 0.67
CPU PLL Voltage 1.54v
FSB Termination Voltage 1.34v
DRAM Voltage 1.98v
NB Voltage 1.28v
NB GTL Reference AUTO
SBridge Voltage 1.20v
PCIE SATA Voltage 1.60v

Load Line Calibration ENABLED
CPU Spread Spectrum DISABLED
PCIE Spread Spectrum DISABLED
CPU Clock Skew 400ps
NB Clock Skew 100ps

Advance CPU Settings
CPU Ratio Setting MANUAL
CPU VID DEFAULT
All Others DISABLED

Let me know if you guys have any thoughts/recommendation on these settings ..

This screams GTL settings at this point. Up the GTL on the core that gave the error first. Its all trial and error now!

First thing I'm doing today!!

[SuporterPoli]

Can i please have an opinion from you guys:

Q9450 @3.2GHz 1.264V with a lapped TRUE+Scythe SFF21F(1600rpm)+AS5 on it.
Why is there a so big difference between the cores? I checked the cpu and its absolutely straight. Also i did the washer mod on the TRUE and it doesnt move one mm now. Is the mobo temp sensing wrong? Those temps are at idle but with a rather high ambient temp, around 30C or more.

[Cob]

A difference is normal.

Currently running Prime95 Blend atm and my temps are 51 56 57 49 (that's under water ofcourse).

[rdv]

LUCI5R your questions were:
1) "I couldn't spot anything pertaining to a P5Q"
> Here is a better reference: http://www.xtremesystems.org/forums/...age=4&t=111156
> See the user "jVIDIA" signature (post#80) it quotes a P5Q Deluxe and he said "I did the pencil vDroop mod and the results @full load are : before : 0.07 vdroop now : 0,005 ~0.01 vdroop".

2) Doesn't AUTO on CPU PLL take it too high .. towards unsafe limits?
> You are correct I have changed the setting to 1.51v but that ends up as a real 1.648v the same as if the setting was left on auto. I did a number of tests at lower vPLL but all failed. As mentioned before this is the voltage I am most concerned about.

3) Same with NB ... I'm not sure what's the max supported for your board but I was told for the P5Q not to take it over 1.4
> In my research the OC guidelines I have for the NB are "Not past 2.0v (<1.75 and lower)". It was very difficult to find widely accepted 24/7 guidelines for the vNB. My Rampage Formula has on board LED indicators with three colors (Green, Yellow and Red). As you can imagine Green is for safe and Yellow if for High while red is "Crazy". I am on the lower end of Yellow (1.51~1.73) with my vNB being a real 1.54v so I am assuming this is high but safe. My temps under prime95 load are at or under 50C. I have also seen a number of quotes of people running in the 1.50-1.55v range when OCing. Right now I am going with what the mobo and forums are telling me.

An additional comment on the high vPLL, I found an article (http://www.xtremesystems.org/forums/...d.php?t=179580) and it has an information table which shows a number of voltage readings (see: Bios to Actual vDroop listing). They show BIOS settings, BIOS Thermal Monitor values. their own multimeter readings and multimeter readings under load. These are all for the Maximus Formula mobo which is the sister mobo to my Rampage Formula. With the only difference being Maximus having an X38 rather than the Rampage X48 chip. I used the table information to extrapolate what my vPLL "multimeter" readings under load would be (three different ways) and they came out to be in the range of 1.580v to 1.599v.

This puts me on the knifes edge of being over the guidelines. I am not sure if I am deluding myself because I want to run at 3.7Ghz 24/7 or this is just good research. Of course the counter argument if Intel's own recommendations of max 1.575v then what the heck am I thinking?

I will make my mind up on how much risk I will take for a 0.1Ghz improvement after I finished tuning and stability testing.

Hope you have some luck obtaining your targets.

[LUCI5R]

rdv, thanks for your detailed responses! Always helps.

LUCI5R your questions were:
1) "I couldn't spot anything pertaining to a P5Q"
> Here is a better reference: http://www.xtremesystems.org/forums/...age=4&t=111156
> See the user "jVIDIA" signature (post#80) it quotes a P5Q Deluxe and he said "I did the pencil vDroop mod and the results @full load are : before : 0.07 vdroop now : 0,005 ~0.01 vdroop".

That post is actually from 2006, and jVidia had the P5B back then. He has a P5Q Deluxe now which is why his signature shows P5Q Deluxe. (Signatures on this forum are backward-post compatible). I don't believe he was referring to doing to the pencil mod on the P5Q Deluxe. I've searched extensively in the P5Q Deluxe thread as well and found one other person (besides me) asking the same question - but no answers. I'll keep looking.

> You are correct I have changed the setting to 1.51v but that ends up as a real 1.648v the same as if the setting was left on auto. I did a number of tests at lower vPLL but all failed. As mentioned before this is the voltage I am most concerned about.

Makes sense. With the other massive changes I'll be making to the BIOS today, for now I'll restrain the vPLL between 1.54v ~ 1.58v

> In my research the OC guidelines I have for the NB are "Not past 2.0v (<1.75 and lower)". It was very difficult to find widely accepted 24/7 guidelines for the vNB.

I quite agree with you - I don't think there is a widely accepted or standardized guideline of the vNB. Being on air - I think I'm being a bit more conservative. I think Cob and some others on Water are up much higher then I'm at.

I am not sure if I am deluding myself because I want to run at 3.7Ghz 24/7 or this is just good research. Of course the counter argument if Intel's own recommendations of max 1.575v then what the heck am I thinking?

That article was definately an interesting read.
Quite honestly, there's always a thin line between 'really good research' and 'self delusion in order to achieve goals'.

For now I have enough settings to fiddle with, so I'll probably be going through some trial & error myself on my setup - but I'll keep reporting here cause I get real good feedback & advice from you guys.

Thanks!