Q9450 OC / Temps / Settings

[Zigosity]

Hey,

I can't seem to get my Q9450 stable at 3.6 on my P5Q3 dlx.. Am I just -that- unlucky or am I doing something really wrong here? Best settings I can come up with (40 minutes or so OCCT stable? Heh.) are as follows:

vcore: 1.37500 bios (1.365 idle\1.368 load) LLC enabled
PLL: 1.56
VTT: 1.38
vNB: 1.44
vSB: Auto

GTLs are all auto, I was only going to play with them after I got a stable clock to try and bring my vcore down a bit. VID on my chip is 1.25, I know that's bad, but others aren't having nearly as much trouble with the same VID. I have decent air cooling and a 40mm fan tied to my NB heatsink.

Thanks for any help,

Zigo

[tdenton1138]

Hey,

I can't seem to get my Q9450 stable at 3.6 on my P5Q3 dlx.. Am I just -that- unlucky or am I doing something really wrong here? Best settings I can come up with (40 minutes or so OCCT stable? Heh.) are as follows:

vcore: 1.37500 bios (1.365 idle\1.368 load) LLC enabled
PLL: 1.56
VTT: 1.38
vNB: 1.48
vSB: Auto

GTLs are all auto, I was only going to play with them after I got a stable clock to try and bring my vcore down a bit. VID on my chip is 1.25, I know that's bad, but others aren't having nearly as much trouble with the same VID. I have decent air cooling and a 40mm fan tied to my NB heatsink.

Thanks for any help,

Zigo

Look through my settings and see what might apply to help you out... also see Luci5r's post just below.

[Utnorris]

So, I just got my Q9450 up and running and so far it is looking really promising. 3.2Ghz (400x8) @ 1.18v. What really surprised me was the other voltages. Take a look at the screenshot. These are a lot lower than my Q6600 was at. Of course I haven't tried 3.6Ghz yet, so I am not sure how much higher I will need to go, but as I said, it looks promising.

Batch - L813B225
VID - 1.2500

[Zigosity]

I also got mine working at 3.6GHz now.. (I gave it a bit more NB, which it seemed to appreciate. 1.48v now.) It's 1hr OCCT stable, but I haven't run anything else on it really besides this quick 3dmark06 run, which made me quite happy. Almost 3k more than stock, plus I broke the 20k barrier!



Needless to say, I'm going to make sure this is stable (running small FFTs overnight in p95) and then try and reduce voltages, it's a pretty sweet setup. Hopefully I can get it lower than 1.368 vcore, it's a little high for my liking.

[Loser777]

LLC and 45nm chips are a bad mix... LLC usually interferes with stability with 45nm chips.

[Zigosity]

Really? I've read through this whole thread a few times and most people report having LLC on without issues. The other day however when I was testing it at 3.4GHz I noticed a similar thing: it would run at a certain bios voltage with LLC off but not with it on - very strange as the actual real voltage was way lower with it off! Hmm.

I'll continue with my original plan to run small FFTs overnight, and during the work day tomorrow I'll run the same tests with LLC disabled. We'll see what happens. Is there a particular reason why it would cause instability, or is it just a rumor running around right now? I'm curious.

[Utnorris]

1st question - What is LLC?

A bunch more.

1- It seems that I was not quite stable at 1.18v, had to raise it to 1.22v. Seems strange since it made it through OCCT fine at 1.18v. Not really a question just an observation.

2 - I tried getting 3.6Ghz stable and stopped after it failed with these settings:
Vcore - 1.325v
FSB - 1.4v
NB - 1.39v
PLL - 1.54v
SB - 1.14v
Dram - 2.1v (normal setting for it)
Any ideas?

3 - I have noticed people putting SB really high, how does this effect overclocking? I thought NB and FSB were the ones to pay attention to.

Temps are fine, never go over 54c on any of the cores. I had a Q6600 before and you could set FSB, PLL and NB to 1.6v. On the 45's if I understand it correctly, 1.4v is the max. If that is the case then why am I seeing 1.48v on some? Just risk taking or is the 1.4v a guideline?

Thanks,
Utnorris

[Zigosity]

The general consensus seems to be 1.4 VTT, 1.6 PLL, 1.6 NB (depends on your chipset though, mine doesn't like to be above 1.5), and 1.370 vcore.

LLC is Load Line Calibration, which is supposed to almost eliminate vdroop. With LLC enabled my idle voltage is 1.36, and load is 1.368, to give you an idea.

As far as SB volts go, I have mine at 1.20.. I don't think it makes much of a difference.

I tried a large FFT run just for the hell of it, and it failed after 21 minutes. Going to up the GTL of the failing core and disable LLC, see if it helps.

EDIT: Two quick failures on cores 3 and 4. Great. Guess LLC isn't the culprit, after all.

EDIT2: 1.400 vcore in bios, 1.320v under load, core 3 fails in less than 5 seconds during large FFTs and idle volts is over 1.37. I'm going back to using LLC. I'm really hoping that vcore isn't the problem here, because I can't boost it any higher. I'm starting to think my chip just -really- sucks.

EDIT3: Clocked back to 3.2 again... taking longer to boot now and I'm getting weird horizontal lines across the screen while it's doing it. And my USB ports apparently detected a new device I didn't attach. Going to leave it off for the night and see if this doesn't go away, I hope I didn't just break something...

[Utnorris]

Thanks for the info, that explains a lot and gives me some idea of where I can go. At first I thought I had a "golden" chip, but after reading a bit, 3.2Ghz on stock seems to be the par. Now that I know what my limits are I can start tweaking to get at least 3.6Ghz. I was able to run my Q6600 at 3.725Mhz 24/7 with no problems, hoping I can get there with this chip. 500Mhz is a lofty goal for quads, but I should be able to do 450Mhz with no real issues. As far as LCC, it was one reason I went with the Blitz. My Gigabyte didn't have it and I couldn't get 3.7Ghz stable on it since the bios stopped at 1.6v and the vdroop was terrible. Anyway, again, thanks for the info, that helps a lot.

Utnorris

[Zigosity]

Just spent an hour reseating everything, clearing cmos, making sure my HD4850s weren't fried and re-installing catalyst. White horizontal lines still present during boot sequence, but at least the long boot time and weird USB thing is gone. Meh. I think I'm going to call it quits at 3.2GHz and keep a close eye on my mobo and cpu to make sure I didn't accidentally fry something. Good luck getting to 3.6 Utnorris, hope you have better luck than I did!

EDIT: Went away overnight, thank god. Something weird must have overheated. Scared me, though.

[Utnorris]

Thx, sorry to hear about the problems. Good luck on getting them fixed.

Utnorris

[mit78]

my q9450 stay al the time at 3,35ghz,in 1,15v,temps its 45-47 idle,and full lload 59,many people says "temps on q9450 its not real,its a problem with that,i test with a digital termometer and says 30-31 grades,its a problem or not?

[SuporterPoli]

Hello mit78 use Realtemp for the correct temperature.
My temps with 1.34vcore and 3.52GHz and aircooling are maxed out at 60C.
This with push-pull on TRUE and everything is on the table.

[Amurtigress]

Hi folks,

I posted earlier here with my rampaged maximus board being stuck at FSB 436 with a Q9300.

I am planning to switch over to DDR3 RAM soon, and would like to know what the currently best chipset is for DDR3 and Q9300 overclocking. Any advice?

Does the X48 really make a difference with the Q9 CPUs compared to the X38..what about the P45 and NVidia 790i based boards?

Looking forward to your opinions!

[Utnorris]

Ok, so I have 3.6Ghz stable. Here are my settings:

Auto Settings
Vcore - 1.42v
FSB - 1.23v
NB - 1.58v

Manual Settings
SBridge - 1.14v
PLL - 1.58v
CPU GTL - 63
NB GTL - 67

Temps
Idle Cores - 44-38c
load Cores - 60-55c
NB - 31c
SB - 40c

Any issues with these that I need to be aware of? I.E. Any set too high?

Thanks,
Utnorris

[RunawayPrisoner]

vCore seems a bit too high for 3.6GHz? Is that after vdroop or before vdroop?

[RaZz!]

skimmed over the thread and couldn't find what i'm looking for: how far will the q9450 be able to clock on stock volts (and with stock heatsink)?

btw, what's the stock vcore actually?

tia

*edit* board will most likely be a p5q-e - just fyi if it matters to answer my question

[Zigosity]

Ok, so I have 3.6Ghz stable. Here are my settings:

Auto Settings
Vcore - 1.42v
FSB - 1.23v
NB - 1.58v

Manual Settings
SBridge - 1.14v
PLL - 1.58v
CPU GTL - 63
NB GTL - 67

Temps
Idle Cores - 44-38c
load Cores - 60-55c
NB - 31c
SB - 40c

Any issues with these that I need to be aware of? I.E. Any set too high?

Thanks,
Utnorris

If that's Vcore after droop, it's a little too high; I wouldn't go over 1.38. NB is a little high, but your temps are good, so it should be fine. PLL might be high too, if those are bios settings and not real voltages, but it might also be fine.

Really all you need to worry about now is trying to reduce that vcore a bit, I'd say. Most people aren't even going that high for 3.8! Then again, if you have a chip like mine... You probably won't get much lower. So you could run it like that, or reduce your OC a bit, depending on when you feel like replacing parts = P. Good luck!

-Zigo

[Utnorris]

Thanks. Those are the actual voltages reported inside windows. I have a Blitz with LLC turned on, so very little vdroop. I may try to lower vcore to 1.39v, but this was after trying several settings and I finally threw those on auto to see what the board would put them at. 3.8Ghz probably won't be attainable without going to 1.45v on vcore. I left it on auto and the board put it to 1.43v and it wasn't stable at 475FSB or 463FSB, so it may be that this chip just can't do 3.8Ghz or my board isn't that good with 45nm chips. I will keep playing with it.
Thanks again,
Utnorris

[Zigosity]

Hmm, I'm thinking about trying for 3.6 again, because when I ran memtest86 earlier today, it gave me a few errors (but only on the second time through, first time ran fine..).

I'd like to know.. Do large FFT tests stress the ram enough that if there's a problem there, it will fail? I'm running one now at my stable 3.2GHz clock (I never bothered to before), which runs everything I throw at it just fine, from gaming to rendering stuff on all four cores in Cinema4D. We'll see how that goes.

-Zigosity

EDIT: vNB 1.5v and slightly less clock on my RAM gets me 2 hours large FFT stable so far, 1.368-1.374 vcore. Might try reducing voltages or narrowing down if it was RAM or NB, or both giving me problems. Gonna stress it more overnight, see what that gets me. : ).

[Loser777]

I was looking at the chips my local store had... Anyone heard of an L816?

[Loser777]

It's ok, I found out already. 1.25000v VID ...Gay.

[RaZz!]

skimmed over the thread and couldn't find what i'm looking for: how far will the q9450 be able to clock on stock volts (and with stock heatsink)?

btw, what's the stock vcore actually?

tia

*edit* board will most likely be a p5q-e - just fyi if it matters to answer my question

hate to quote myself, but i'm still confused about the q9450's default vcore.

are there differences in the default vcore or what? while browsing the inet i found a lot of cpu-z shots of stock q9450s where none was equal to the others... the voltages varied from 1,1xx to 1,2xx. what's this crapload of different stock vcores about?

...the last time i've fiddled around with overclocking was in the athlon64 era, hence i'm quite confused with all this core2 stuff right now

[SuporterPoli]

The VID is the "stock" vCore for each chip, and it varies from chip to chip based on Intel's testing procedures. The VID will tell the motherboard: "This is the voltage I require at stock speeds when the VCore setting is on 'Auto' for proper operation." Nothing more. It is said that a lower VID value will yield better overclocks, but nothing's been conclusively proven.

Q9450 like other cpus have a range of vids. That is why you saw different stock vcores.

[Shadowthor]

but lately it seems intel has been setting vid higher on recent cpu production.