Nice chip nkresho.
Good luck in 4.3 land.
Printable View
No ac, it's only getting up to about 24c outdoors here this time of year. Windows are open.
Temp measured at the front of the case is currently 26.7c, prime blend is getting the hottest core up to 59c in realtemp (other cores are at 56c). Water temp is at 31.7c in my reservoir.
I just ordered a hk 3.0 waterblock about an hour ago, should be here by the end of the week... it will replace my current dtek fuzion v2.
I plan on putting my other res in tonight whenever UPS comes. I'll get more temp results at this clockspeed to compare to then too.
Ok, finally done with 4203mhz. I got blend stable with 1.24375v bios and 1.216 in cpu-Z
http://img195.imageshack.us/img195/2...eblend8htg.jpg
I want to see what a second rad does for your temps, I am as high as I can go at 4500mhz because of heat in blend. I will add another if it does well.
Thanks andressergio.
This chip is to good for my ASUS Rampage Formula board.:D
Here's what I've got so far with these nice G.SKILL F2-9600CL5D-4GBPI memory and my Q9650. The ASUS Rampage Formula isn't the best board for OverClocking these Q9650 CPU's but it's a nice board to clock these G.SKILL sticks.
http://i431.photobucket.com/albums/q...RD6-tREF16.jpg
http://i431.photobucket.com/albums/q...tRD6-tRE-1.jpg
I thought about that myself though I am running a PA 120.3 just for my 9650 which is overkill already. I just grabbed another 9550 yesterday from Microcenter for 159.00 and its a 1.150 vid which is nice, 3850mhz at 1.22v so far is not a problem even on this P5QC board.
DAMN!!! That is insane.
ty guys! :D , the only restraining factor seems my
X38-DQ6 (or... my inability to tweak it.. :P) as my first attempts to find a stable high fsb setting had poor results , partly due to my 4x1gb sticks. above ~415(5:6 - ram 498) starts asking for much vMCH in non linear "response" (1:1 setting is obviously much more demanding in terms of vMCH :S)
guys it is safe to use LLC on the UD3R for 24/7 4GHz @ 1.23v?
Here's the first full run with the second radiator installed. Temps went down just a little bit. Water temps are lower by 2-3c, but cpu temps seem down by only about 1c. I guess I wasn't fully using all the radiator cooling potential I already had. No problem though, I'm sure i'll use it eventually.
http://img34.imageshack.us/img34/928...end8haddec.jpg
Also, here's the log where I added the rad.
pic-log
nkresho: Great blend run, keep pushing. :up:
early report for 3.6 is ~1.1v at load to pass linx , i havent look around for similar setups, but seems to me just an average 9650, nothing extraordinary...
http://img17.imageshack.us/img17/3805/linx3600.gif
O/C curve till now:
3.2 -> 1.008v load
3.4 -> 1.056v load
3.6 -> 1.104v load
PS> temps start to worry me.. if i get 60 C at 3.6/1.1 at (hypothetical scenario) 4.0/1.2+ how high ?:S
after setting 3600 the wolf wakes up slowly slowly :P, eager for more... but have obvious difficulties to raise fsb with 4 dimms on x38-dq6. After trial & error , i found the most stable GTLREf settings for 425 fsb and minimum voltages (VTT +0.1/1.21v, MCH +0.075/1.33v) , to be GTLREF 1 : +3% , GTLREF 2 :-9% , kinda weird ?? , such Voltage Delta between these two , is reported to shorten chip's life? , assuming most ppl use a plain 0.63% (or equivalent, both at -3% in a GB board)
ty in advance again :)
EDIT : more VTT :s (+0.1)
Here are some quick results@4.5. This obviously is not Prime stable yet(will probably need 1.45V), but so far stable enough for my purposes. Yesterday I was priming at 4.4 and kept getting errors no matter what Vcore. Temps were topping out at 73C. Some of you guys have talked about this issue, Prime failing if temps get above a certain point. I realized my 38mm fan was installed backwards (pull config), switched it to push config and temps went down to 63C top, and no more errors. Im sorta glad I didnt get one of those crazy low vids, as I really am liking my temps at my Vcore.
http://i26.tinypic.com/33kzp0y.jpg
Wow, nice clocks viper. And on air nonetheless.:clap:
I can't wait until I get to 4500. Soon, very soon.:D
Great job Viper, congrats.
^I hope. Not sure though.
Question for you guys...
When prime runs fine for a number of minutes, then everything freezes up, what is it usually? I have found that with vcore, you'll typically have a core or two fail, but it will keep priming the others. I keep having freezes, not bsod, not cores failing. Wierd. Any advice?
One sec, i'll add my overclocking notes...
I recently was able to accomplish 12hrs of Prime95 blend + IBT with 1.32vcore on a much higher vid chip (1.2125). Those with Turbo Profile and rams at 1200Mhz 2:40B (PL8).
4500Mhz with less then 1.30vcore it's perfectly doable on a low vid chip using slack ram timings like on standard profile and ram's on 1:1 strap.
Nkresto: the freezes you're having are most likely to be ram related. You will probably need more Vmch or Vdimm to correct the problem.
http://i42.tinypic.com/3580too.jpg
Thanks for the insight.
I'm guessing mch then, as I'm running the same "big boy" gskill pi as ocnub and andressergio.
Here's a copy of the last two pages of my overclocking logs...
http://img6.imageshack.us/img6/9348/83728009.jpg
http://img196.imageshack.us/img196/728/91415873.jpg
trying the ibt stable settings with higher mch and mchref right now... we'll see.
Jor that's a so damn good chip that i dont get tired of lookin' such an impressive low 4500 OC's... i tried on mine and not even at 1:1 wich loosen timming i could go less than 1,392vcore
Did enabling all CE1 states and all that you have enabled help on your OC or you just activate all after stable ?
Cheers :up:
Sergio
Very impressive Jor. That is one hell of a chip, I think the guy clocking it has a lot to do with it also. Always nice to see your posts.
lol, its gotta be true...
heh sergio can u link me to some of ur Bios settings past 4.5
i give up trying any more with x38-dq6 , two afternoons spent for nothing, cant get stable 450fsb , no matter what... (with 1.475 MCH -too high already- get random lock ups..), ill stay at low 9x400/1.1 and be "happy" :S .there is a hope in future if i change my ram to some fast 1066+ so with anything than 1:1 the chipset will be much more relaxed.. , but for now "nisafi"- enough
Jor
I tried your same bios settings with all states enabled i was surprised i could boot to W7x64 at such low volts (1,328 cpuz) but playing i BSOD ...so i went back to bios and changed some to get 1,36 on cpuz and i get to play crysis and all games, didnt test blend, Tom did and failed...buts nice to run low volts and cool, also nice to see multi changin from 500x9 to 500x6 when idle lol...
I loosen Gskill timmings and im at 2,08 everest read at PL9 turbo CL5
Will keep trying tough i doubt it could be stable prime but let me play on bit oc'd clocks on the GTX285 for higher OC on VGA i think the 1,46 on the MCH are a must, so a little more vdimm as it dont like to be OC'd with such low volts.
Cheers mate and thanks for those tips :up:
Sergio
Last night I added an 80mm fan to my northbridge. It's sitting on my gpu for now. I was suspiscious of my instability 3-4 hours into a run. And this morning, to my suprise, I found her still chugging away.
I'm deciding whether or not to go ahead and get the EK waterblock for the nb just to keep it under control. If it adds stability I'm all for it.
I read that the nb/sb 5 fits no problem http://www.sidewindercomputers.com/ekwaek5ac.html
If I do that, though, I would probably get the xspc dual ddc res and make it one loop for the nb/gpu on the mcr220 rad, and a separate loop for my cpu with the mcr320... decisions decisions...
Ok, well, the pics...
http://img7.imageshack.us/img7/2316/...mestable8h.jpg
http://img5.imageshack.us/img5/7249/addednbfan.jpg
The high exhalted 4500 is next...
^ Incredibly nice. :up:
Thanks guys. Currently priming at 4500
voltages are:
pll 1.50
cpu term 1.340
cpu ref 0.873 65.15%
vcore 1.35000
ddr 2.06
nb 1.400
nbref 0.928 66.29%
ichio 1.5
ich 1.1
2hours 14mins in ... going strong. cross your fingers for me.
Thanks for the recommendation, but if I'm going to cool the nb, i'm going to do it right. I know thermalright makes good stuff, but I put my faith in water keeping it cooler.
Thanks guys, you two are not doing bad at all. I really appreciate the input from you guys.
There are so many variables on the mix that is hard to tell the formula.
I used exactly those settings Sergio, I did the run and after that got into BIOS, so the C1E state etc were on. I don't think that had anything to do with anything.
Temps are an issue so I kept it under 63oC on the CPU.
As I said I could go around 1.30vcore for 4500mhz but that was on loose profile and auto on mems.
Now summer is over and I'm back at my stable 4800Mhz.
Doing some benches again, my goal is to unify the HWbot.org ranks for the Q9650, I did the test that was missing (PCMark05) and grabbed the 3rd. place, I need Acard power to grab 1st.
Thanks for all the compliments.
Very good nkresho, you are shinning already.
Yeah IBT will need less vcore but will need more vmch and vdimm. Funny thing.
Your cpu is great and it's scaling very well with voltage. You will probably brake the barrier of 4500Mhz stable soon enough.
Keep up the hard work and keep us posted.:up:
Thanks. I'm at 5hours 36mins, still going. Tomorrow my HK 3.0 comes. I'm pretty sure I have a bad mount with my fuzion v2, as core 0 always seems 3-5c higher than al lthe others, idle or loaded.
I have decided to order the nb block. I'm going dual loop. More results to follow. And of course pics.
nkresho: The difference in core temps may be the sensors, I have several 45nm CPU's and all have at least 1 bugged sensor.
I have taken the DQ6 out and put the UD3P back in. I will learn to do the volt mod on the DQ6 and put it back in this rig. If it had LLC I would probably never use this ud3p again, I think the DQ6 is a better all around board despite a lot of people saying it is garbage. I was able to clock my q9550 100mhz higher, blend stable with reasonable voltages. The only downfall was the higher idle vcore because of not having LLC.
Well here's the screen shot. I'm a little bothered by the realtemp reading of 7.5 multi in correlation with the cores 0/1 being on different tests than 1/3. Does this mean I have to do another 8 hours to make sure I'm really stable? I honestly don't remember whether or not it said 7.5 at the beginning...
http://img3.imageshack.us/img3/4789/...mestable8h.jpg
I wouldn't run it again, try realtemp 3.30. Congrats on your 4500 blend, great vcore. :up:
very nice nkresho
What advanced clocks do you use for clock drive and skews?
Thanks ocnub, I'm going to try a few runs with lower voltages here and there. I used jor's template, and I might be able to get a tad lower on some voltages.:hehe:
Thanks everyone for the recommendations, praise, and compliments.
phsinc1: Stock settings on those.
I'm a big fan of the KISS method. If i'm not forced to change it, I don't.
All that I changed that isn't shown/typed out by my pics is pl8 in the memory settings as well as my cr is set to 2t. All else is auto or hard set to stock settings.
I'll try some runs for 4600 after I put the new cpu block in tomorrow afternoon some time.:D