Q9450 OC / Temps / Settings

[JohnnyV]

New, small issue... With my current strap vs FSB settings I can either run my memory at 1140mhz which I can do at 6-5-5-20 1T settings or I can choose for 1540mhz. I have switched back to 2T and loosened the timings to 7-7-7-24. I did 10 runs of memtest86 test 5 and it was ok but the problem is this. As soon as I start up a game it hangs somewhere during loading.
What would this indicate? Do I need to increase NB voltage since the memory is pretty much stable during memtest or do I need to increase DRAM voltage to get it to work in games as well?

[Singh400]

First all, thanks for this thead, a ton of info here - anything and everything you need to know about the Q9450 is in here. So cheers!

I was hoping that someone could take a look at my settings and suggest any possible improvements.

I’m currently running at 3.4 Ghz stable (linpack 10x, and Prime95 12hrs+)

Here are my settings:

Ai Overclock Tuner 		[Manual]
CPU Ration Setting		[8.00]
FSB Frequency			[425]
PCIE Frequency			[100]
FSB Strap to North Bridge	[266 MHz]
DRAM Frequency			[DDR2-1066MHz]
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		[Manual]
 1st Information:		5-5-5-15-3-52-63
 CAS# Latency			[5 DRAM Clocks]
 DRAM RAS# to CAS# Delay 	[5 DRAM Clocks]
 DRAM RAS# Precharge		[5 DRAM Clocks]
 DRAM RAS# Activate to Prechar	[15 DRAM Clocks]
 RAS# to RAS# Delay		[Auto]
 Row Refresh Cycle Time		[Auto]
 Write Recovery Time		[Auto]
 Read to Precharge Time		[Auto]
  2nd Information:		7-3-5-4-5-4-6
  READ to WRITE Delay(S/D)	[Auto]
  Write to Read Delay(S)	[Auto]
  WRITE to READ Delay(D)	[Auto]
  READ to READ Delay(S)		[Auto]
  READ to READ Delay(D)		[Auto]
  WRITE to WRITE Delay(S)	[Auto]
  WRITE to WRITE Delay(D)	[Auto]
   3rd Information:		14-5-1-6-6
   WRITE to PRE Delay		[Auto]
   READ to PRE Delay		[Auto]
   PRE to READ Delay		[Auto]
   ALL PRE to ACT Delay		[Auto]
   All PRE to REF Delay		[Auto]
DRAM Static Read Control	[Disabled]
DRAM Read Training		[Auto]
MEM. OC Charger			[Auto]
Ai Clock Twister		[Auto]
Ai Transaction Booster		[Auto]
 C/P: A1 A2 A3 A4 | B1 B2 B3 B4
 LVL: 10 10 10 10 | 10 10 10 10

******** Please key in numbers directly! ********
CPU Voltage			[1.21875]
CPU GTL Voltage Reference(0/2)	[0.370]
CPU GTL Voltage Reference(1/3)	[0.750]
CPU PLL Voltage			[1.50]
FSB Termination Voltage		[1.26]
DRAM Voltage			[2.10]
NB Voltage			[1.40]
NB GTL Reference		[0.370]
SB Voltage			[1.10]
PCIE SATA Voltage		[1.50]
*************************************************
Load-Line Calibration		[Enabled]
CPU Spread Spectrum		[Auto]
PCIE Spread Spectrum		[Auto]
CPU Clock Skew			[Auto]
NB Clock Skew			[Auto]

I started at a vcore of 1.33750 and have slowly dropped it down to 1.21875 doing linpack 10x every time I drop the vcore.

[Slovnaft]

SO I decided to post here about my recent bad/good news.
The bad news is that my Q9450 is dead, probably from trying to push over 4ghz with more than 1.5v. The good news is intel is great about replacing chips pretty much no questions asked, and I should be getting a brandspanking new one on wednesday.
So I would warn anyone, especially those with P45 boards, against pushing even to 4ghz, as it could kill your chip more easily than most. I only spent a handful of hours altogether making abuseful attempts on my chip before noticing that it would no longer post at stock. moving it to a new board confirmed, it's dead.

[Big Lar]

First all, thanks for this thead, a ton of info here - anything and everything you need to know about the Q9450 is in here. So cheers!

I was hoping that someone could take a look at my settings and suggest any possible improvements.

I’m currently running at 3.4 Ghz stable (linpack 10x, and Prime95 12hrs+)

Here are my settings:

Ai Overclock Tuner 		[Manual]
CPU Ration Setting		[8.00]
FSB Frequency			[425]
PCIE Frequency			[100]
FSB Strap to North Bridge	[266 MHz]
DRAM Frequency			[DDR2-1066MHz]
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		[Manual]
 1st Information:		5-5-5-15-3-52-63
 CAS# Latency			[5 DRAM Clocks]
 DRAM RAS# to CAS# Delay 	[5 DRAM Clocks]
 DRAM RAS# Precharge		[5 DRAM Clocks]
 DRAM RAS# Activate to Prechar	[15 DRAM Clocks]
 RAS# to RAS# Delay		[Auto]
 Row Refresh Cycle Time		[Auto]
 Write Recovery Time		[Auto]
 Read to Precharge Time		[Auto]
  2nd Information:		7-3-5-4-5-4-6
  READ to WRITE Delay(S/D)	[Auto]
  Write to Read Delay(S)	[Auto]
  WRITE to READ Delay(D)	[Auto]
  READ to READ Delay(S)		[Auto]
  READ to READ Delay(D)		[Auto]
  WRITE to WRITE Delay(S)	[Auto]
  WRITE to WRITE Delay(D)	[Auto]
   3rd Information:		14-5-1-6-6
   WRITE to PRE Delay		[Auto]
   READ to PRE Delay		[Auto]
   PRE to READ Delay		[Auto]
   ALL PRE to ACT Delay		[Auto]
   All PRE to REF Delay		[Auto]
DRAM Static Read Control	[Disabled]
DRAM Read Training		[Auto]
MEM. OC Charger			[Auto]
Ai Clock Twister		[Auto]
Ai Transaction Booster		[Auto]
 C/P: A1 A2 A3 A4 | B1 B2 B3 B4
 LVL: 10 10 10 10 | 10 10 10 10

******** Please key in numbers directly! ********
CPU Voltage			[1.21875]
CPU GTL Voltage Reference(0/2)	[0.370]
CPU GTL Voltage Reference(1/3)	[0.750]
CPU PLL Voltage			[1.50]
FSB Termination Voltage		[1.26]
DRAM Voltage			[2.10]
NB Voltage			[1.40]
NB GTL Reference		[0.370]
SB Voltage			[1.10]
PCIE SATA Voltage		[1.50]
*************************************************
Load-Line Calibration		[Enabled]
CPU Spread Spectrum		[Auto]
PCIE Spread Spectrum		[Auto]
CPU Clock Skew			[Auto]
NB Clock Skew			[Auto]

I started at a vcore of 1.33750 and have slowly dropped it down to 1.21875 doing linpack 10x every time I drop the vcore.

Where did you come up with those gtl ref #'s? Kinda surprised at that is all. After tweaking my gtl's I ended up with:
0/2= 0.670
1/3= 0.675
NB gtl= 0.640

I understand that every setup is different.

Larry

[Singh400]

Where did you come up with those gtl ref #'s? Kinda surprised at that is all. After tweaking my gtl's I ended up with:
0/2= 0.670
1/3= 0.675
NB gtl= 0.640

I understand that every setup is different.

Larry

Well alot of reading here, taken alot of information from people here who have Q9450 and P5Q Deluxe's.

I'm no OCing expert, I'm a complete newb at this

[Earzz]

I'm no OCing expert, I'm a complete newb at this

Well same here, all i know i learned from reading over here and im still reading a lot
XS is making us all better people

[Justintoxicated]

you expecting magic when there is very little there.

I can do 400x9 with 1.1v vtt all day long, but if you EVER want to be stable at 440 plus your going to need at least 1.3v if not higher to achieve this.

I have been working with this board, overclocking quads, since its release (almost a year now), I know exactly how it behaves at different FSB levels and what the required settings are. Basically what I'm saying is that you are wasting your time trying to reach mid 400's with such a low vtt, and you need to tweak gtls when you have that higher voltage, otherwise none of the gtl tweaking you did carries over...

I never trie dover 400 FSB at these voltages.

Problem is even with Vtt @ 1.35 and CPu at 1.4v I can't boot up over 400 FSB... My only stable overclock is at stock voltage with custom GTL's...
3200MHZ is my Max overclock with CPU and NB under water cooling. Actualy if I take out 4GB of ram my max stable overclock is 3400MHZ with high CPU and Vtt voltages. I guess my CPU or Motherboard just sucks? I must have wasted over 48 hour straight playing with this, and I keep going back, to 3200 MHZ completely stable. Its very strange to me I guess my intel just overclocks like asz.

When I say stable I mean at least 50 iterations of Intel Burn Test, often it will fail onyl after 20 tests or so.

[SNiiPE_DoGG]

I never trie dover 400 FSB at these voltages.

Problem is even with Vtt @ 1.35 and CPu at 1.4v I can't boot up over 400 FSB... My only stable overclock is at stock voltage with custom GTL's...
3200MHZ is my Max overclock with CPU and NB under water cooling. Actualy if I take out 4GB of ram my max stable overclock is 3400MHZ with high CPU and Vtt voltages. I guess my CPU or Motherboard just sucks? I must have wasted over 48 hour straight playing with this, and I keep going back, to 3200 MHZ completely stable. Its very strange to me I guess my intel just overclocks like asz.

When I say stable I mean at least 50 iterations of Intel Burn Test, often it will fail onyl after 20 tests or so.

r u kidding? burn test is bs and nothing loads a cpu like that so why bother?

that's like taking a coat hanger and hanging a 100lb weight from it to test its reliability, when all you really need it to do is hang a coat

I've never had a system that was 2hr OCCT stable crash on my before so that is my stability benchmark.

[jas420221]

Well alot of reading here, taken alot of information from people here who have Q9450 and P5Q Deluxe's.

I'm no OCing expert, I'm a complete newb at this

Those GTL ref are JACKED up... .37 and .75? The difference between those 2 numbers is something I never saw before and thats across 3 forums. Something just doesnt seem right. I cant believe that is stable at all on those cores.

SO I decided to post here about my recent bad/good news.
The bad news is that my Q9450 is dead, probably from trying to push over 4ghz with more than 1.5v. The good news is intel is great about replacing chips pretty much no questions asked, and I should be getting a brandspanking new one on wednesday.
So I would warn anyone, especially those with P45 boards, against pushing even to 4ghz, as it could kill your chip more easily than most. I only spent a handful of hours altogether making abuseful attempts on my chip before noticing that it would no longer post at stock. moving it to a new board confirmed, it's dead.

I highly doubt it was vcore that killed it, or your P45 board (unless you got a POS one). You abused it to death via PLL or VTT.

Ive been trying to get 4Ghz stable for a month+ now, and havent quite done it. I think I have settled on the settings in my sig though. PLenty safe.

[Justintoxicated]

r u kidding? burn test is bs and nothing loads a cpu like that so why bother?

that's like taking a coat hanger and hanging a 100lb weight from it to test its reliability, when all you really need it to do is hang a coat

I've never had a system that was 2hr OCCT stable crash on my before so that is my stability benchmark.

I have had systems that pass 48 hours of prime crash on me before.

Stable is stable I don't want "stable enough" thats not acceptable for me. I'd rather be truely stable than pretend that I am. If the CPU can pass it at default speeds it should be able to with a so called "Stable" overclock. 3200MHZ is all I can hit and adding voltage to the CPU does not help. I have spent over a month total trying to go higher, I have only managed to do so with 4GB memory and alot of voltage. I feel much better knowing my CPU can take anything I throw at it and will be stable. Sure nothing loads it up like burntest but thats not to say it will never make a mistake unless it is under a load. 50 iterations of Burntest takes about 4-6 hours to complete with 8 GB of memory. I think tis very reasonable.

[SNiiPE_DoGG]

I have had systems that pass 48 hours of prime crash on me before.

Stable is stable I don't want "stable enough" thats not acceptable for me. I'd rather be truely stable than pretend that I am. If the CPU can pass it at default speeds it should be able to with a so called "Stable" overclock. 3200MHZ is all I can hit and adding voltage to the CPU does not help. I have spent over a month total trying to go higher, I have only managed to do so with 4GB memory and alot of voltage. I feel much better knowing my CPU can take anything I throw at it and will be stable. Sure nothing loads it up like burntest but thats not to say it will never make a mistake unless it is under a load. 50 iterations of Burntest takes about 4-6 hours to complete with 8 GB of memory. I think tis very reasonable.

no, your being unreasonable, my computer never crashes and I use it all the time for all different kinds of things. Dont complain to all of us that you cant get your cpu to OC high when you put it under ridiculous stress and then say "oh its unstable"

[jas420221]

no, your being unreasonable, my computer never crashes and I use it all the time for all different kinds of things. Dont complain to all of us that you cant get your cpu to OC high when you put it under ridiculous stress and then say "oh its unstable"

Good for you, and here is a cookie. *cookie*. Congrats on 4Ghz and being able to ssurf the net with it.

Most feel a 8 hour test (P95, PCCT or that new intel pgm) is 'good enough' unless you F@H. Then you undoubtedly should test it for 24 hours. I dont think he is being unreasonable. Especially considering its a generic standard that he is following.

[Singh400]

Those GTL ref are JACKED up... .37 and .75? The difference between those 2 numbers is something I never saw before and thats across 3 forums. Something just doesnt seem right. I cant believe that is stable at all on those cores.

So I should lower those figures? What's an "ideal" figure?

It is stable Prime95 (12hrs+) and IBT.

I've now further dropped the vcore to 1.16250 - IBT stable, currently Priming.

[jas420221]

Naa... if it works... I gues it works. Default values for those are .67. Ive never seen one as low as .37. Same with the NB GTL.

[Singh400]

Naa... if it works... I gues it works. Default values for those are .67. Ive never seen one as low as .37. Same with the NB GTL.

I'm gonna try those default values tomorrow, just out of interest.

Also, should I carry on dropping my vcore? Surely the lower the better right?

Thanks for your help dude

[Justintoxicated]

no, your being unreasonable, my computer never crashes and I use it all the time for all different kinds of things. Dont complain to all of us that you cant get your cpu to OC high when you put it under ridiculous stress and then say "oh its unstable"

To each his own, I want more MHZ, just not badly enough to be less stable. Theres not a huge difference in 3200 to 3400 to 3600 anyways. I mean sure software isnever going to be 100% stable on windows anyways but as much as possible is #2 importance for me, #1 being that I don't have any leaks in the water cooling.

Or maybe your just scared to run 50 iterations of burn test? LOL :P

[TheScavenger]

http://www.newegg.com/Product/Produc...82E16820145194

That is the Ram I have, here is some more info on the chip they use TWIN2X4096/CM2X2048-6400C4DHX XMS640X v1.X PSC 128MX8

You think these have any room to overclock at all?

I'm running that RAM at 900MHz on 2.12V on a G35 chipset.

[Singh400]

Naa... if it works... I gues it works. Default values for those are .67. Ive never seen one as low as .37. Same with the NB GTL.

Well I tried them at the values you suggested and the result is a non-booting system. Completely dead. Working now of course.

I even tried lowering them by one increment (-0.005) and the result was the same, PC went dead on next bootup.

[truehighroller]

I have both of mine set to .63 and my fsb term is 1.22 "that one won't do less then 1.22 for sure" and my Core voltage is 1.245v~. I haven't tried to drop it in a while..

[Slovnaft]

wow, 3.85 on 1.248vCore, nice gtl tweaks!

[TheScavenger]

wow, 3.85 on 1.248vCore, nice gtl tweaks!

That is pretty awesome voltage! I'm jealous!

[Slovnaft]

i think the gtl tweaks might have been the magic there?
i never fooled around with them because i didn't quite know what to aim for.

[carepolice]

I know that this is about 9450 but well 9450 & 9550 is nearly 1:1.

I cant get grip of my cpu!? Its so full of suprises! I have always tought that it likes applevoltjuice, but i first tested whit 3,8ghz clock boot whit vFSB 1.32V ! Well then i rebooted and changed vFSB 1.10 wich is well known as 45nm default value. Doh system booted and is working fine

I have to keep high vcores always on high OC's, But when i try play whit NB and this vFSB voltages i CANT lower vcore even for one snap! I can just run whit default vFSB and vNB on 1.26 whit high vcore

Feels like that my CPU is scumbag who just likes only kind of volt!

in few words = Why my vFSB & vNB doesnt make any difference, any value that i give?

[Slovnaft]

I know that this is about 9450 but well 9450 & 9550 is nearly 1:1.

I cant get grip of my cpu!? Its so full of suprises! I have always tought that it likes applevoltjuice, but i first tested whit 3,8ghz clock boot whit vFSB 1.32V ! Well then i rebooted and changed vFSB 1.10 wich is well known as 45nm default value. Doh system booted and is working fine

I have to keep high vcores always on high OC's, But when i try play whit NB and this vFSB voltages i CANT lower vcore even for one snap! I can just run whit default vFSB and vNB on 1.26 whit high vcore

Feels like that my CPU is scumbag who just likes only kind of volt!

in few words = Why my vFSB & vNB doesnt make any difference, any value that i give?

First off, 1.26 is a pretty low vNB especially for FSB speeds 450+. When pushing high OCs on this chip I get stable around 1.4-1.6vNB depending on the board and chipset.

Second, what's the highest you've ever had your vcore? Instability at varying settings could be a sign that you've fried your chip. Try booting at stock everything. If you can't, try seating it on a different board if you have that option. If it won't post on another board then you've probably killed your chip, but don't worry, intel will RMA it as long as you throw a bunch of technical terms at their RMA support.

[DrPain]

I run a 9450 on an Asus Striker 2 NSE and just by putting up the fsb can run it stable at 3.6. I run SLi with a BFG 9800GTX and a Asus 9800GTX+ and on Futuremark '06 score 19848. So believe me when I say I love this CPU I found that running with the water on both the north/cpu keep temps below 25 even after a couple of hours of gameplay. I shifted from a Q6600 and the difference is extreeme.You do need the 1002 bios and nvidas newest drivers, as previously I had nothing but trouble. So good to see there are some real people out there. But has anyone tried the 2.8 I wonder how sweet it must be.