Yup, the RAM would just go in the yellow slots but easier in the green and its surprising what those vanes can do to your fingers if they get a chance.:eek:Quote:
Originally Posted by RobPuba
http://img.villagephotos.com/p/2004-...7/DSCF0852.JPG
http://img.villagephotos.com/p/2004-...7/DSCF0851.JPG
please share with us the "secret"Quote:
Originally Posted by Oese
first of all a 3h prime screenie @ 267x12 :D
ok dont know perhaps the secret is low temps.. this is water..
whatever just set htt the same as boot up htt then be careful that ht is not over 2000mhz (i set it to 1,4ghz) disable c'n'q and lower nb-multi a bit.
set voltage manually to whatever required nb and core-voltage seems like must be totally the same..
and thats what it was..
this is vista 64bit and 4 dimms at 533mhz :shocked:
now 24/7 and "idle stability" testing can begin.. :D
i did what you say and still got no juice out of thecpu...
what stepping?
mine is 0810GPAW
perhaps the cpu is just a bad clocker.. what i can say now its not the board that was what i wanted to confirm and try out especially what htt oc does..
how high can you tune the htt with your cpu? how high is the nb then? did you lower the ht-speed to under 2000mhz?
perhaps try setting trfc for all dimm slots (unoccupied as well) to some 195ns.. i do that always since my first 4600+ did not wanted high htt without that step..
Cool refHT speed you use the GSkill's? Lower temps definately helped here too. Not that i need lower voltages but the system is more stable at higher voltages (>1.4V).Quote:
Originally Posted by Oese
BTW: That mosfet cooler does not help. With lower cpu voltages (<1.375V) temps are ok (~50°C) but with >1.45V temps exceed 70°C.
http://www.abload.de/thumb/mosfet1v7.jpg
Well I know it sucksQuote:
Originally Posted by ea6gka
That's why I just bought a (HP OEM) LSI MegaRAID SAS 8344ELP
Now I just have to wait for my day-off to setup the new RAID 0 :D
The onboard SI3132 RAID sucks , I have been experience data corruption with my RAID 0 setup almost once a month. I took the board to my local dealer and they said there isn't anything to do about it. One of the tech support guy told me probably related with crappy BIOS :shrug:
That's a "Antec Spot Cool" fanQuote:
Originally Posted by Oese
I bought the Corsair RAM cooler but it doesn't fit with the Ultra 120 :down:
Quote:
Originally Posted by justapost
pitty.. but remember these digital pwm are build to run at high voltage.. but whatever lowering temps would have been nice..
no i'm not using the gskill right now because they are needed in 2nd rig.. that doesnt work well with 4dimms.. weak imc..
I use corsair dhx cl4 4x1gb
aha be was right guessing :DQuote:
Originally Posted by imamage
K, thanks phenom seems to love 1GB dimm's.Quote:
Originally Posted by Oese
In theory it should not be too difficult to make an waterblock for the pwm section whom is still connected to the heatpipe.
lol i didnt mean high voltage, i meant high temp ^^
you think phenom likes 1gb dimms? would be nice i'll get 4x1gb crucial 667 cl3 in some time...
I did some testing today under linux 64bit.
I used 1.4V CPU VID varied frequencies in 200MHz steps from 1Ghz to 3GHz and noted max/min power consumption (at wall) and cpu temps after ~3mins running monitored for a minute. I used stresscpu2 to generate load.
Temps started at 43°C and went up ~1°C with every 200MHz step. At 2.6GHz after a few minutes the temps jumped up 5°-6°C. At 2.8GHz and 3Ghz the increase was ~1^C again.
I repeated the whole testing serie and this time it did not jump at 2.6 but at 2.8GHz. Repeated again and it jumped at 2.6GHz. I also noted that the PWM temps started to exceed 50°C.
I added an extra fan and voila, as soon as the pwm temp was below 50°C the cpu temp dropped ~2.6GHz. I increased the frequency to 2.7GHz and still had no temperature jump. I removed the pwm fan and as soon as the pwm temps get in the 55°C range the cpu temp also jumps fom 49° to 56°C.
I added a fanmate and can repeate that behaviour forever now.
At 2.8GHz however a lower pwm temp does not help the temp increases ~8°C (58°C) compared to 2.7GHz(50°C). Going to 2.9GHz the temp is 59°C and at 3GHz 60°C at 3.1GHz 61°C at 3.2GHz 62°C at 3.3GHz 64°C. (Those above 3GHz are not stable at 1.4V but run stresscpu2).
I'd say both statements are true.Quote:
Originally Posted by Oese
From what we have seen 2GB modules don't go that far ref HT wise.Quote:
Originally Posted by Oese
yeah i think the cpu temp depends on pwm temp because the whole area heats up quite a lot.. but i like the phenom temp sensors all 4 cores are the same and not too unrealistic..
and with the 2gb modules..: could be as well, perhaps i'll test with the gskill one time..
can someone tell me.. is there any possibility yet to lower cpu fid/vid in windows other then c'n'q? i know crystal does not seem to work..
if not.. would it be risky to let the nb run at 1,45v or whatever it needs to do its 2,4xx ghz all the time and let control the cpu fid/vid via c'n'q?
because obviously i want to lower my idle power consumption without loosing my overclock :D i can do 3,12ghz @ c'n'q, but 3,2ghz does not work because of the nb speeds which i have to set to a lower value if using 267mht htt..
on the other hand: is using c'n'q a good idea at all? because it was stated that it is kinda outdated algorithm not optimized for faster up to date processors and does switch too often especially under vista. amd is working on another driver but did not seem to succeed by now..
guys look at this..
295mhz htt 3,53ghz 3dmark stable^^
http://www.forumdeluxx.de/forum/show...postcount=1186
hehe this was by far world record for 9750 :yepp:
http://www.ripping.org/database.php?cpuid=715
Nobody knows?:confused:Quote:
Originally Posted by SocketMan
somewhere it was stated to have something to do with the p-states.. p0 is high power p1 is power-saving.. but i didnt find it does something..
Ahh ok Thanks man:up:Quote:
Originally Posted by Oese
I find that P0 is a lot more stable with AOD then p1
that's about it. Nice work on that 9750;)
thx :D
i left p0 there also ;)
i think i will run into thermal and financial problems running these settings:
http://www.abload.de/image.php?img=dsc02492e9b.jpg
watertemp:
http://www.abload.de/image.php?img=dsc02493u8a.jpg
Do you think its risky to have ram @ 2,13v whilst nb is @ 1,25v??
If P1 would mean to start in p-state-1 it would mean that your system runs at half the stock frequency with 1.05V. Bios uses this p-state sometimes but the filed in the bios has nothing todo with it.
To be save with CnQ I use the same multi for the lower p-state with just an 4 instead of an 2 divider. I run 3GHZ/1.375V 750MHz/0.9V since a week or so without problems under linux 64bit.
Saves ~10-20W in idle.
Oese that ref HT speed is cracy what bios version do you use? Congrats to that chip, nice clocker.
Wattage eh, I saw 400W already here. :) As for the NB voltage, 1.25V with mem at 2.1V should not cause issues. My Tracers died short after i started to apply above 2.3V. :shrug:
I had one phenom whom liked an equal NB / CPU VID all others could not care less.
Gotta join over at forumdeluxe. :)
good to hear that 1,25v nb / 2,1v vdimm won't be a problem..
bios version was 5/5/08..
i'm right now struggling to get 3,2ghz stable again, it seems my 4°C more coretmp today is holding the chip back.. but i think i'll switch over to a c'n'q stable setting soon whatever, have to read about how to change p-states... but right now i want another prime screen @ 1,4v/1,25v nb at higher nb speeds :D
yeah forumdeluxx is no.1 german forum but most people are more sophisticated over here (at least in the amd section ;) )... whatever amd section is big at forumdeluxx so it makes pretty sense to have a look over there too ;)
edit: I made a interesting observation: with the 05/05 official my 3,2ghz is not stable, with 05/05 beta it is!! wtf?? there are only two possibilities: something went wrong with flashing the official, or there were changes by dfi between beta and official..
edit2: it wasnt the biosses.. its the temperature... yesterday and today its 30°C and coretemp went up to from 42°C to 46°C and that seems really to make 3,2ghz impossible at the moment..
phew :up: solved:
http://www.abload.de/image.php?img=phenomsgi.jpg
was the ram temps, now with cooler it flies again :D
http://www.abload.de/image.php?img=dsc02495r0i.jpg
So why would they put:
"AMD CPU Stepping....." if it's related to the P-states?
Oh yea it's dfi I forgot:rofl:
I might have a source of AOD problems here:
Every time you start the AOD it sets it's own
voltages(vlotage in dfi language :rofl:) (lower than BIOS)
For example it sets 1.25 for the CPU instead of 1.3,so naturally if your cpu speed is set to a nice 2.8-3.0 (at 1.3 or more) when the aod starts -the
voltage drops and you get a freeze.
yep thats right i experienced that also.. only solution is doing all the overclocking via aod and let it load the profiles at startup.. of course thats not a good solution and therefore i do not really use aod at all..
vlotage :rofl:
Oese could you give the settings you used for booting into Vista x64 above 3GHz? I am able to boot into Vista x86 or Win XP at 3.2GHz without any problems but Vista x64 doesn't boot above 2.85GHz.