Q9550 OC'in experiences are welcome

[Phunk0ne]

Just check my signature for set up.

But just in case:

CPU: Intel Core 2 Quad Q9950 C1 stepping
Mobo: Gigabyte P35 DS3r V1.0 with F10 Bios
Ram: A-data DDR2 800+ Extreme
GFX: eVGA 8800GTS G92
PSU: Zalman ZM-850HP (850Watt)

Volts: 1.2250 Bios / 1.184 Idle (CPUID) / 1.152 100% (CPUID)
Ram: @ 800Mhz @ 1.8V @ 4-4-4-12
Ratio: FSB:RAM @ 1:1
Multiplier: X 8.5

EIST & C1E are turned on


So I went ahead and carefully tried 3.4Ghz. the only two things I had to change was upping the bios Vcore and turn on the .5 multiplier.
It did crash on 1.2000 vCore in bios. But after turning up the voltage 2 notches, it stabilized.

I did check the Intel site what the VID rage is for the yorkfield, and it seems I have enough room left to hit 3.6.

At this moment, my cpu is running idle C0/C1/C2/C3 @ 49C/45C/45C/39C. I am pretty sure I have to re-seat and reapply some AS5 in order to achieve better temps, other than that I will lap the IHS next weekend (will post some pics, etc).
Under 100% (Prime95), my cpu is topping out at 59C.

(oddly enough, the idle temp has stayed the same, going from 1.15 to 1.2250)

So anyone got some wonderful stories and such which I can incorporate into my overclocking adventure with my first Quad =)

Post your info here, such as mine (rig set up, settings and overclocked achievement).

[Edit 1: Added extra info]
[Edit 2: Added CPU Stepping info]
[Edit 3: Overclockled my cpu further]

 

[Phunk0ne]

Anyone with a Q9550? ?

 

[Phunk0ne]

Got my Quad @ 3.4Ghz now with minor adjustments

reading my temps, is that a safe operating temp for quads during idle/stress?

 

[GundamF91]

I don't have Q9550, but I do have Q9450 and have similar results. I also hit 3.4Ghz @ 1.225vCore but I have to do 425mhz x 8multi to get there. Your next challenge should be 3.6Ghz. I can get it around 1.275vCore. I think you should make it easily with 424x 8.5. The only problem is the temp. I was hitting 65c with that.

You should also test your system with Intel Burn In Test. It'll heat up your system real good and report errors far faster than Prime95 or any other tests. Just be careful not to run too long, 10-20min is enough to tell how your system performs under extreme stress. You'll probably hit 70c+ temp with this program.

 

[Sentry2]

I just picked up a an X3360(Q9550) C1 from newegg. Would like an E0 but I didn't want to wait. So far it's been a great chip.

4.037GHz
475x8.5
1900FSB
1.4v

Temps: ~37c idle/ ~51c load(quad prime small FFTs)

~11hr quad prime stable

http://img513.imageshack.us/my...60ocprimestableey3.jpg]

 

[GundamF91]

I ended up hitting 4Ghz at 1.4vcore, but the temps were approaching 80c. But my room is also warmer, I idle around 46c.

What cooling did you use on yours? 51c is a great Prime95 load temp.

 

[Sentry2]

Swiftech Apogee GTX, 2x GTS 360 rad's(might change soon), Panworld 50PX pump This pump moves A LOT of water.

Room temps stay in the low 70's

 

[Big Lar]

Originally posted by: Sentry2
Swiftech Apogee GTX, 2x GTS 360 rad's(might change soon), Panworld 50PX pump This pump moves A LOT of water.

Room temps stay in the low 70's

Nice pump, pretty similar to mine

Larry

 

[Keysplayr]

n/m. Thought you meant Phenom 9550, LOL. Sorry.

 

[Sentry2]

Originally posted by: Big Lar

Originally posted by: Sentry2
Swiftech Apogee GTX, 2x GTS 360 rad's(might change soon), Panworld 50PX pump This pump moves A LOT of water.

Room temps stay in the low 70's

Nice pump, pretty similar to mine

Larry

Nice. Are you running an Iwaki or a Panworld? Great stuff.

 

[Phunk0ne]

Edit to follow soon:

In short, CPU is lapped, but have to do another lapping session to get the surface as flat as possible. TRUE120 in PushPull config (2xS-Flex SFF21F 1600RPM).
Swapped out my A-Data Extreme 800+ with OCZ Reaper HPC 4x2GB (8GB) DDR2-1066.
at this moment, the cpu tops out C0/C1/C2/C3 - 61/57/52/52.

bbl, time to take a nap

nice setup Sentry.. how about some pics =)

 

[Sentry2]

Yeah, if I get a chance this weekend I might do that. Nothing to special...got a Codegen S-201(B) server case and I'm still running out of room. Looking real hard at a Mountain Mods Ascension CYO. All said and done though it comes out to about $540 + shipping for the anodized black setup and an etched top window. A bit more then I wanna spend right now but they sure do make some bad ass cases! Anyway...

 

[Patrol]

hey guys im new here ,this for realy good, Very nice guide i will reply later on maybe tomorow and shair my trouble with u guys , again very nice forum see u soon keep it uo guys ,but im still waiting for Mr Phunk0ne ,new reply and se what hapend with his overclocking eperience ,cool

 

[Tempered81]

some people are getting high clocks out of increasing the voltage of the 12v rail on the power supply in conjunction with increasing all the various bios voltages to the cpu. I'm sure quad-core overclocking is difficult when trying to raise the FSB to compensate for the low multipliers, but I don't have any first hand experience. These guys are getting 500 to 600mhz FSB's on 45nm quad cores. So along with FSB, Vnb, GTL's, VTT, PLL, Vcore, SB, etc.. - you can now manage your PSU's delivered voltage, i presume.

Here is a post:
"I've been following this thread for awhile now, and from what i can see NapalmV5 is right on the money, he is making perfect sence in my head, we are raising the voltages on every bit of out boards/cpus except the 12, 5 and 3.3 volt supplys to it, from what i understand by doing this it is keeping volts not only stable but above what they need to be...just like what we have to do to overclock our cpu's...but there comes a point in overclocking where it is necessary to overclock/over volt other things besides vtt and nb..hence over the years how we have seen so many more options for voltage control, doing the psu thing is just the next step....i say well done NapalmV5, your thinking outside of the box, and thats why you have, and will do better...
Regards Andrew"

http://www.xtremesystems.org/f...d.php?t=197802&page=14

Screenshot of 560mhz FSB and 5.0Ghz validated on a quadcore:
http://valid.x86-secret.com/show_oc.php?id=423924

 

[Patrol]

hello guys i have Intel Core 2 Q9550 E0 stepping A, i got it OC 3.400 , stable , but i cant get it any higher then 3.400 GHz , ASUS P5QE , PSU corsair 1000 W , GAINWARD 4870x2 ,sound card is AUZEN X-FI prelude 7.1 , CORSAIR (800MHz) DDR2 ,4-4-4-12 at 2.1V. well i did set the CPU voltage up tp 130000 ,SB 1.10 , NB 1.20 , DRAM 1.84 , CPU SPREAD SPECTRUM disable , plz any help that can OC higher like up to 3.6 or even 4GHz on the ASUS P5QE , plz

 

[Patrol]

sorry i Forgot to( Mention above ) that im using 8GB RAM , 1 TB SAMSUNG

 

[KingstonU]

Just got my Q9550, have been dying to find the time to finish putting it together and try my very first OC on it. Will post when I get it going.

 

[mrSHEiK124]

Just got an X3360 (Xeon version of Q9550), just finished building it, I'll try some overclocking stuff tomorrow.

 

[Phunk0ne]

Patrol:
one of the reasons you are having problems getting your quad stable beyond 3.4 could be the 8GB's of ram you are using. It's stressing your northbridge way more when using 8GB's compared to using only 2 ram slots.

Jaredpace:
My thoughts start to shiver what kind of cooling the 5GHz Q9650 needs to achieve such speeds . . .

and the post you made that comments about voltages really makes sense. Telling me it is time for me to replace my P35-DS3R. making my Urashima 2.0 have a little update to 2.1 to an X48 chipset (with the idea of making him into a crossfire rig @ 2x 4870X2). But that's for sometime later.

Hopefully Intel will use the Hydra technique sometime next year:
http://www.anandtech.com/tradeshows/showdoc.aspx?i=3379

 

[mrSHEiK124]

Hmm, current results with my X3360, I've got it at 3.4 GHz and it looks like it might need one step up vcore; CPUZ is reporting 1.2V.

Full Info:
ASUS P5Q Deluxe 1004 BIOS
Intel Xeon X3360 (Q9550, Xeon version) @ 8.5 * 400

FSB Strap to NB: Auto
CPU Voltage: 1.21250, going to try 1.21875
All GTL Voltages on Auto
CPU PLL Voltage: 1.5
FSB Termination Voltage: 1.3
DRAM Voltage: 2.0
NB Voltage: 1.38
SB Voltage: 1.10
Load-Line Calibration: Enabled
CPU/PCIE Spread Spectrums both disabled
CPU/NB Clock Skew both on Auto
CPU Margin Enhancement was on Optimized, going to try Performance Mode

IntelBurnTest on 50 passes goofed a few in the first 10, ran like 30 stable, goofed a few more, and then stable till the end. Prime 95 conks out on what Windows reports as the first core nearly instantly, the other 3 kept going. Max temps on IBT are 60C, 50C in Prime95. Idle is 38C on the hottest core.

 

[mrSHEiK124]

Set BIOS vcore to 1.225V, CPU-Z reports 1.208V idle and 1.216V under load (load-line calibration apparently cancels out the effects of vdroop), Prime95 has not crashed on Core#0 yet, lookin good so far

I wonder if I can hit 3.6 GHz; using a Xigmatek HDT-S1283 in a P182 with the fan pulling air from the front of the case and exhuasting out the back 120mm fan. I might point it up and have it exhaust out the 140mm fan in the top if that doesn't block any RAM slots and my X-Fi in the first slot is not in the way. Is 51C Prime95/60C Linpack/37C Idle regular for a Q9550 running at these settings? I'm used to my Manchester X2 staying nice and chilly at like 50C under load and 34C idle.

EDIT: WTF is up with my temps? Cores 0-2 are usually all hanging around at 50C with Prime95, Core 3 is always a few degrees colder (50/50/50/45 right now )

 

[mrSHEiK124]

X3360_2.png

FSB Strap to NB: Auto
CPU Voltage: 1.21875 (1.208V idle in CPU-Z, 1.218V under load, I attribute it to load-line doing it's job)
All GTL Voltages on Auto
CPU PLL Voltage: 1.5
FSB Termination Voltage: 1.3
DRAM Voltage: 2.0
NB Voltage: 1.38
SB Voltage: 1.10
Load-Line Calibration: Enabled
CPU/PCIE Spread Spectrums both disabled
CPU/NB Clock Skew both on Auto
CPU Margin Enhancement: Performance Mode

I'm going to attribute the second pass of Linpack failing to SuperFetch hogging a bit of RAM and causing Linpack to use VM instead; once I freed the RAM up it ran 48 passes in a row perfectly.

Speed Step was enabled during the Linpack run and it worked exactly as it was supposed to, didn't affect any stability. Idle is 6x400, load is 8.5x400, if I only I could figure out how to work some kind of p-state undervolting for lower multipliers like on laptops. Is it worth messing with C1E too?

Next Try
420 MHz * 8.5
5-pass Linpack stable @
CPU Voltage: 1.26250V (1.249 Idle in CPU-Z, 1.262 Load)
All GTL Auto
CPU PLL Voltage: 1.56V (it was at 1.50V from the 3.4 GHz overclock, 3570 MHz was not stable until I bumped PLL)
FSB Termination Voltage: 1.36V
NB Voltage: 1.40V

I'm still within safe areas, right? Hottest core under Linpack is at 63C