• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Q9650 tuning

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

BobbyBubblehead

Member
Joined
Jan 29, 2009
just dropped in a Q9650 to replace the Q66

set it for 4GHz first boot and started my stress test on OCCT

anyone point me at good source material to tune it to perfection?

havnt set the termination voltage or dabbled with the finer settings but I need to because the test just failed #Chuckles#

any tips?

looking at the result two things apparent

#1 its getting woops style hot on cores 1&2 (best re seat the block me thinks)
#2 the vcores dancing everywhere.
 
Last edited:
Enable the LLC ( load line calibration ). this help for Vcore.

for this kind of FSB you need a lil more NB, PLL, Fsb Term.

VID of your chip ?
 
looks to be a guffy 1.30vid
seems high for what ive read of other folks Q96

and check, check `n check for the others.

have to go find the LLC though... any idea what im looking for in the ep-45 bios?

im not going to screw around with it too much until ive gatherd them there facts.

currently gave it some 1.38125 on the core

temps core 1&2 32`C
core 3&4 39`C (re-seating the block did bog all)
unloaded temps.

NB upped a little & SB

north bridge temp 35`c

I had a go with occt and the temps launched to 95`C ooops!

so story so far is its benchable but not stress test worthy

and stabilities better than the Q6 was @ 3.851

obviously theres some frame increase to measure as well.

just need to find out what im doing with this particular chip and then I plan to run it into the ground :D

(if hitting 95`c on three cores didnt already dig the first shovel full out its tiny grave #Laughs#)
 
Enable the LLC ( load line calibration ). this help for Vcore.

for this kind of FSB you need a lil more NB, PLL, Fsb Term.

VID of your chip ?
This but For 4Ghz I didnt need to touch my PLL, it was still 1.50v, YMMV of course.

NB 1.36
FSBt 1.3

95C under water????????


LLC is towards the bottom of the MIT section, Load Line Calibration.
 
oh that reminds me EarthDog, im going to get a real case for my rats nest that you seemed quite "moved" by #Chuckles#

im thinking TJ07... no reason I carnt manage cables in one of thoes #Laughs#

(just for you.... & airflow/self respect #Chuckles#)

also canning the alu cooling stuff for a rather nasty experiment... stay tuned for that one ;)

no I7 though ive drawn the line, just plan to run this one till the big gal squarks.
and then see if I even care about daft spend PC`s

anyhow seems my vcore dancing between 1.31 - 1.33 as things stand.

guess I need another thread that has relevance to this one... planning to get a good block for this CPU (current ones alu :rolleyes:)
 
I'm making a similar jump too, i expect my Q9550 will be here next week, going from a Q6700. Hopefully my P5Q-E can make it prime stable at 4

didnt seem any effort getting it to post and get to windows but stable in real terms?
think thats where the luck & skill combo come in #Laughs#

really want to try 4700 with the ram at 1066 but im probably dreaming right on thinking thats going to hold up for any length of time.

I was running the Q66 at up to 3.850 so a wussy 30% odd clock isnt that exciting yet still good news on a few more fps and lower temps to go.
 
Dtek Fusion v2 or Apogee GTX would be solid blocks.

what rads would you go with in the event of no swiftech triple rads to buy in this tiny grim country :shrug:

looking for two triples.

the pickings are

thermochill
xspc
tfc
black ice

(planning to cool the gtx260 its two 80mm fans are doing my head right in)

probably do two loops CPU/NB & GPU/GPU (north bridge because its pre-blocked and I can I think #Chuckles#
 
what rads would you go with in the event of no swiftech triple rads to buy in this tiny grim country :shrug:

looking for two triples.

the pickings are

thermochill
xspc
tfc
black ice

(planning to cool the gtx260 its two 80mm fans are doing my head right in)

probably do two loops CPU/NB & GPU/GPU (north bridge because its pre-blocked and I can I think #Chuckles#

There's a 775 XSPC block. i think its called the delta V3. Works wonderfully, and stacks up against the D-tek and EK for MUCH cheaper. Should also be readily available in the UK!
 
didnt seem any effort getting it to post and get to windows but stable in real terms?
think thats where the luck & skill combo come in #Laughs#

really want to try 4700 with the ram at 1066 but im probably dreaming right on thinking thats going to hold up for any length of time.

I was running the Q66 at up to 3.850 so a wussy 30% odd clock isnt that exciting yet still good news on a few more fps and lower temps to go.

yeah by stable, i mean no BSOD ever... i had my Q67 @ 3.8 for a while too but sometimes it crapped out on me. Brolloks said the chip I'm buying ran 4ghz at only 1.28v under load, so it should be a winner :)
 
I hit 95`C earlyer (prize dingbat!)

its like scratching a new car :-/

maybe turning your thermal protection off and havin a stress test with a crap block is a z list plan after a second thought :rolleyes:

well I just poked in bios and realized I hadnt set the termination voltage on my last visit there.

just upped it by .020mv and now the vcores jumping the other side of 1.33 to .34mv

maybe drop the voltage a little now and see if it helps its standing temp any.

just trimmed 1`c off the board and north bridge temps.
or maybe its getting cooler anyhow we shall see #Chuckles#

woot core 1s lost a degree too.
think I was a tad over zelous with the mv`s earlyer.

note. I switched from zalman grease to arctic silver 5 on my gtx260 earlyer and its lost 10`c at idle #Laughs#

neither maybe many peoples favorites but I think theres a clear winner with it clocked further too #Chuckles#
pleasing result because ive still got at least 11 and half grams in the tube :D

got to say I think a Q66 @ 3.8 is a fair taste of a Q9650 @ 4G about half to three quarters of the way there depending on what your doing.

much less heat to get a result though.
 
95C under water????????

well I tweaked and benched again and no errors and the hottest core was 46`C

so im sure the 95`C was a miss report when it failed the test.... simply will not get that hot with my cooling :shrug:

settled for 4005GHz curios to get an 8.5multi & 533Fsb though (tease myself about the future #Laughs#)

thanks all for the clues :thup:

new case next week and no more alu :D
 
Back