• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Some Q9550 OC advice?

Associate
Joined
10 Oct 2008
Posts
353
Location
Dundee
edit: oops theres a specific overclocking board :( move please? Sorry I don't know how i missed it

I havn't done a lot of searching because I want some good personalised advice and I know people here dish it out in rivers. I recently build a system for a friend and my intention is to get a nice overclock on his q9550, here is all the system information:

------
case: NZXT m59
mobo: DFI Lanparty DK-T2RS Plus
cpu: core 2 quad 9550 E0 SLBV8 batch: L930B834 MM: 899133
hsf: ac freezer 7 pro v2
ram: 2x2GB Corsair XMS2 1066MHz
psu: OCZ ModXtreme 600W
gpu: 7900gs (getting ati 5xxx or nvidia 3xx later)
hdd: samsung spinpoint f3 500GB
------

I have the cpu at a nice 3.2GHz but trying to get it to 3.4 pretty much results in prime 95 not responding - as windows puts it (nope, not worker thread stopped).
The VID of this chip is 1.3 which seems a tad high to me (core temp told me, can I trust it?)

Firstly, I have the RAM dialed back to 800MHz with the fsb at 353. and the multiplier at 8.5, cpu voltage can sit around 1.25 and seemingly be stable and run at load max temp 51C. Other voltages at deafault/auto.

Now, the bios seems to be version 6.00 - but I have no idea how recent that is, the ones on the site are only given by date and the update process is annoying: I need a floppy drive or windows xp to do so - the system is running windows 7. I tried boot from usb but I can't ever get that working, when I download the dos files and extract them the path becomes 'dos/dos/[content]' now, I presume I format it with the HP tool choosing the 2nd folder (dos/dos) - but incase that is wrong I shall slap myself later, and possibly whoever is responsible for making the bad zip archive too :mad:

Even if I get a floppy drive I then need to get a floppy disk :O

But anyway: I know the board is excellent for the fsb - but what voltages should I expect the cpu to require? And as I raise the fsb and cpu voltage - and NB voltage when appropriate. Do I tweak other voltages too? With my p5n-t deluxe I had to change the "1.2v voltage" to 1.25 - after investigation it turned out that was the voltage of a bus somewhere.

Now I hear the 9550 is generally safe under 1.4V, so I don't want it to go above 1.35V for long term use - even if the temps would permit a higher overclock, with proper adjustment of bios settings should I be able to attain 3.4 or maby higher? I got my q6600 to 3.8GHz on my nForce board - just didn't leave it there because I want the chip to last.
 
Last edited:
Core temp doesnt tell you the actual voltage. For that, you need CPU-Z, also use it to check what speed your RAM is really running at. And from what i've heard, the q9550 should be able to do ~ 3.6GHz on stock voltage which is 1.275v
 
Core temp doesnt tell you the actual voltage. For that, you need CPU-Z, also use it to check what speed your RAM is really running at. And from what i've heard, the q9550 should be able to do ~ 3.6GHz on stock voltage which is 1.275v

I know... I am not looking for the voltage, I am looking for the VID of the batch.....

does coretemp do it by looking up the batch with a table? And mine is too new so it's not listed and so it just returns 1.3?
 
I have my Q9550 at 3.6Ghz with stock volts (1.275v vid) however, i can't go much past that due to X48 boards being fsb limited with 45nm quads.

Some good info on clocking the cpu:-

Here and here.

Info on clocking the board here.
 
cant get anything done with it because of the retardedly broken bios and the impossibility of updating it. I have tried all methods and its completely impossible.

tried floppy boot, tried installing xp and using the windows tool, tried usb boot.

all methods are impossible to get working, usb boot it just restarts.... floppy boot I get an error "the file's bios-lock string does not match with your system" and googled that leading to Info that tells me what launch parameters I need so I use them perfectly and it says "Error Parameter.".
DFI havnt provided any information on what the parameters mean, here is the command I am using:
WinFlash.exe PD45D914.BIN /nbl/cks/sn/py/cc/cd/cp/ld/gi/WB/ch
how the hell do I get it to update the bios if no matter what its broken. And I wish they would get a proper translator to write the documents, damn.

edit:
ok right, I got it to run from inside xp with the /nbl parameter in place. I heard a faint click from the mobo and that was it, hopped into the bios and its the same old buggy one. It would help if they actually gave version names rather than dates, surely I didn't already have the newest one? My previous experiences of updating bioses from inside an os took a lot longer, had to wait around whilst the system hung until it completed and the update actually happened. Is my experience here normal for this board/nb chip?

but this cannot be the newest bios, it doesnt have the options that the overclocking guides for the board do - also I am experiencing absolutely massive vdroop and vdrop (over 0.25V). far more than my 780i system gets.

edit again:
i can confirm the update did nothing - my settings such as full screen logo persisted.
 
Last edited:
ram is always as low as I can get it while I focus on cpu overclock, i have 3.4GHz at the moment with these settings:
vtt: 1.32V (didnt need to put this above default for 3.2GHz but 1.32 is 80mV above default!)
NB: 1.295V - put this up with the vtt, i will drop it after p95 has run with these settings for a while to see if it was a factor in the success
vid add: 137.5 mV - making it 1.36V under load, what! will try and lower this a tad to experiment
ram is at 800MHz with ~1.9V 5-5-5-18 T2

this OC seems to be good, but every1 is shouting about how this board and cpu is meant to get up to 3.4GHz with no changes of voltages? what! These voltages are higher than my q6600 in a 780i board :S

edit: my core0 is always about 7C warmer at load than the others, and about 3 above at idle... duff sensor? or hot core?
also core 3 seems to always be the first to fail at p95 on any of the tests :S and core 1 is always a little slower (maby 10 seconds behind for each iteration when doing blend, so 10 seconds behind for test 1, 20 seconds behind for test 2... eventually the other cores are 4 or 5 tests ahead of it!)

i think my cpu is an unlucky candidate.

I have 3.4 stable now I am going to lower the NB first and see what happens, then vcore a tad, and see to VTT too. after that I can get the ram a little faster - though messing with timings annoys me.
Oh also, on auto settings my ram is put at 800MHz 5-5-5-18 even though it is specced for 1066MHz... cant remember the timings, but they are likely 7-7-7-18 (not on the box they came in or on the side of the sticks!)

edit:
ok ive stopped with the cpu at 3.4.
I can probably get it higher but I begin to get worried about voltages. My final voltages were:
VID add: 137.5mV (should be 0 for q9550!!!!!!!!!!!!)
... thats it, all other voltages were default - most of those as low as the options allow, I can probably get vtt a bit lower than default 1.24, but I need to start getting the ram nice and fast - usually when I up the ram to anything above 800MHz I cant even post and have to reset the cmos, current settings:
NB: 1.32V
RAM: 2.1V
ram timings: 5-5-5-18 (default, next adjustment though)

though I am confident of changing those ram settings, won't all the other like 30 timing options on auto be too strict for this? I might be destined to have my ram sat at 800. Though with no cpu overclock at all I could get the ram to 1066 with these ram and NB settings... so maby upping the nb should do it? but is there not a ram bus voltage somewhere, I cannot find it but should there not be one?
 
Last edited:
Is this a E0 stepping Q9550? Mine is running at 3.6Ghz with stock volts but i need a lot of NB volts for the board. NB voltage is 1.36. Any faster and i need 1.4v on the NB which i am not comfortable with. The board is my problem. Think i will offload it and get a P45 to replace it. Not a DFI though. Some good deals around if you look hard enough. Is this your ram? if so timings are 5-6-6-18 @2.1v. Mine is supposed to run at 1066mhz 5-5-5-15 @2.1v but i am currently only at 1008mhz with timings of 5-4-5-12 @2.1v. All of the secondary timings are on auto.
 
Last edited:
Is this a E0 stepping Q9550? Mine is running at 3.6Ghz with stock volts but i need a lot of NB volts for the board. NB voltage is 1.36. Any faster and i need 1.4v on the NB which i am not comfortable with. The board is my problem. Think i will offload it and get a P45 to replace it. Not a DFI though. Some good deals around if you look hard enough. Is this your ram? if so timings are 5-6-6-18 @2.1v. Mine is supposed to run at 1066mhz 5-5-5-15 @2.1v but i am currently only at 1008mhz with timings of 5-4-5-12 @2.1v. All of the secondary timings are on auto.

yes its E0 said in my first post :D the batch VID is 1.3 I am almost certain because it needs a fair bit of vcore to get OCs going. for 3.4 I have it at 1.36 under load and it backs down to 1.12 when whatever that energy saving thing is called drops the multiplier for me. the system is away with its owner now though. but I can still get to it if I want to have a play with settings, ive got all the setttings written down so I cant really mess it up.

As for the ram, that probably is the ram or at least its very similar. didn't get it from ocuk (was more expensive at the time) though so it could be different. Have trouble getting the ram to run fast when the cpu is overclocked, oddly I get a message after post something along the lines of "cpu overclock failed" and I dont trust the automatic timings after I manually set the ones I know about to something like 8-8-8-18 and thereabouts without it helping. So I just opted for 960MHz ram with default (5-5-5-18 in cpuz) timings and upped the NB voltage to get it stable.

Here are my final changes from "optimized defaults" in the gene bios menu:
fsb 400
(default 8.5 multiplier)
cpu VID add: 137.5 mV (lowest stable! any lower - quite a big jump for the next step down - wouldnt get me to desktop)
RAM voltage: 2.05V
cpu vtt (fsb): 1.2V (1 step above the seemingly stable option but its still below default)
NB: 1.3573V (very nicely cooled by the angle of the cpu heatsink and where the fan ended up!)
ram ratio at whichever one gave me 960MHz :D

do the ram ratios change anything except the ram? ie: can they have strange effects (affects?) on anything? why do they originate... just some thoughs I had

Couldn't push the ram to 1003MHz even with 1.4V on the NB (heard that 1.4V is about as high as it's safe to keep the p45 for 24/7). Because I had issues with the ram timings and needed to keep them on auto.

Basically I was dissapointed with the VID of the cpu, but hey hes not paying for electricity ontop of his rent at uni halls... though hes moving into a room in our house next year and energy usage is going to be something we need to pay attention to. But I have no experience of how much a lower VID would cut off a quarterly bill... something like £10?
 
Use Linpack like Intel Burn Test for fast rough stability testing.
It's lot heavier on CPU than Prime and hence shows most errors lot faster.

The VID of this chip is 1.3 which seems a tad high to me
Intel probably put better tweaked production lines to making Nehalems when it was released so later Yorkfields might well be worser overclockers.
But neither is VID absolute truth, my 1.25VID E0 made (or packed) in 09/02/08 is running at 3.7GHz using 1.2V.

Now, the bios seems to be version 6.00 - but I have no idea how recent that is, the ones on the site are only given by date and the update process is annoying: I need a floppy drive
Motherboards of every serious maker should by now support updating from USB memory stick and have flashing software built into BIOS.
Don't even remember when I last needed floppy for updating BIOS manually...
 
Use Linpack like Intel Burn Test for fast rough stability testing.
It's lot heavier on CPU than Prime and hence shows most errors lot faster.

Intel probably put better tweaked production lines to making Nehalems when it was released so later Yorkfields might well be worser overclockers.
But neither is VID absolute truth, my 1.25VID E0 made (or packed) in 09/02/08 is running at 3.7GHz using 1.2V.

Motherboards of every serious maker should by now support updating from USB memory stick and have flashing software built into BIOS.
Don't even remember when I last needed floppy for updating BIOS manually...

Intel probably put better tweaked production lines to making Nehalems when it was released so later Yorkfields might well be worser overclockers.
But neither is VID absolute truth, my 1.25VID E0 made (or packed) in 09/02/08 is running at 3.7GHz using 1.2V.

I thought the same thing, the best raw materials and such would be going to i7/i5/i3. It's a shame but its just the way it is they still run to spec so what can we do :D


The system is set up and good now, but i think ill head over later and try the intel burn test to be sure - hey it's free electricity there too.
Also, I got the usb update to work. I had used the wrong dos folder, because the zip extracted to a folder that had another one and another one inside lol. Turns out it had been shipped updated to the latest bios anyway.
 
Back
Top Bottom