4GHz on Q6600 would need liquid cooling for 24/7 usage, that Kent is too much of a heat ball...Quote:
Originally Posted by zodiac
For fan inside Tuniq, try Noctua NF-P12 or Scythe S-Flex F 1600rpm.
4GHz on Q6600 would need liquid cooling for 24/7 usage, that Kent is too much of a heat ball...Quote:
Originally Posted by zodiac
For fan inside Tuniq, try Noctua NF-P12 or Scythe S-Flex F 1600rpm.
I just installed the copper plate, and the temps seem higher than before.Quote:
Originally Posted by zodiac
that's weird... for me, the copper plate & transpiper has lowered my temps... :shrug:
do you really think that it is a good idea to install the copper plate between CPU and heatsink?
Because it means double the amount of Thermal paste and double the chance of not having an even adhesion between heatsink and CPU.
I don't think that it is the right choice.
Better use a fan to decrease the temps of the north bridge.
I think I will try again tomorrow. I was getting load temps of 75C within about 5 seconds and then a BSOD....Before I was maxing out at 72C or so after several minutes.Quote:
Originally Posted by emoners
@emon
how about transpiper without the copper plate?
after a lot of searching and doing the math 1 question still remains here... what about GTL? Anandtech has a huge article on this but, stupid me, i still dont get it. i see guys with CPU VTT at 1.2V and the GTL around 60-65, which, according to Anandtech should be around 80?
can someone explain how to find the settings that the GTL should be around?
http://edgeofstability.com/Quote:
Originally Posted by Mr Alfabet
There are some good articles there, read the GTL part, especially toward the end.
Read it twice from front to back, but it still doesn't explain to me why some users get some incredible results with GTL settings way below default. the site says VTT at 1.2V should do with GTL around 80, but i've seen nice results with VTT at 1.2 or 1.17 with the GTL at ~65. It seems the sweet spot is well below default...?Quote:
Originally Posted by ether.real
Well, I think the point is, you have to play with it,but going too high or too low will kill you.Quote:
Originally Posted by Mr Alfabet
When using a 45nm CPU the board uses a different VTT/GTL table and the settings are of the lower value like you are seeing.
you have an example of this table so that i know what i'm looking for?Quote:
Originally Posted by Praz
Over at the Tech Repository
http://www.thetechrepository.com/showthread.php?t=228
Thats the same as the 65nm tables, still no sign of ~65 GTL'sQuote:
Originally Posted by soundx98
I haven't had time to do a table for 45nm dual core processors. It literately takes days. These are prime stable VTT and GTL I used for 500x8.
http://www.edgeofstability.com/image...ios/volt_l.jpg
Sorry, I was not aware you were familiar with the 65nm table.
But it is a good example :)
I doubt there is a simple "dial in".
So many settings in the bios affect stability.
Not sure what MHz, FSB or voltage/temps you are looking for but I feel they all affect GTL settings.
Don't get me wrong, "auto/disabled" may work fine for some but even changing a value +/-1 has affected stability for me.
I've "copied" various setting that Praz has posted and they pass the obligatory stability and 3D benchmarks.
I've been using some Mushkin d9s using identical mem settings but I've needed slightly more Special Vid % and I've set my GTLs at 59/61/62 on the E8400 and 60/60/62 withe the X3110.
No idea for GTL settings on the new quads but these vales have been pretty good starting points on two Wolfies.
Wish ya luck, I can only share what worked for me.
I'm using your settings right now ;) allready found them(read most of these 177 pages) i'm priming at 8x510 right now, 1.344Vcore under load. haven't been able to find any difference between the enabled and disabled mode for Vdroop.Quote:
Originally Posted by Praz
by the way... how do you save the bios settings to a neat pic like that? print screen doesn't work i'm guessing ;)
edit:
Soundx, i get what you're saying, and thanks for sharing those other settings. Since the 65nm VTT and GTL have a 67% 'link' i thought i'd figure what the link is for 45nm and play with that. right now i'd say it's around 52%, so VTT*52=GTL setting. Sounds right?
U'll need a digital camera for that.
Praz's pics aren't made with a camera are they? they seem way to neat and pixelized for that.Quote:
Originally Posted by fireice2
Btw, trying to get 500x8 stable right now, but my NB needs way more V then yours, i'm at 1.61V right now. Does that mean i should have a higher NB GTL? or is de NBGTL dependent on the cpu VTT like the others?
edit:
i'm tweaking GTL right now, and i think this is a stupid question because i'm allmost sure you guys are gonna say yes, but do the GTL settings influence eachother? 58/60/65 primed for 1h16m, and 59/60/65 stopped afer only 8m. now stressing at 57/60/65, but if 58 seems to be the sweet spot for the first setting, does that mean i should leave it there if i'm altering the other settings? will 58 allways be the sweetspot regardless of the other settings? VTT will stay the same of cause.
63% seems like a good starting point for all three GTLREF settings when using 45nm dual core processors. The settings are independent of each other. They don't need to be set to the same value and most likely when optimized will have different values.
The above BIOS shot was captured with an external frame grabber. The one below I did with a camera. The frame grabber is a lot faster but also expensive.
http://www.edgeofstability.com/image...800/main_l.jpg
you have a steady hand ;) and how about the NBGTL? is it only dependent on the cpu VTT? or does my high NB voltage need a higher NBGTL? i'll be finding out soon anyway, but i thought i'd ask first.Quote:
Originally Posted by Praz
here are my findings right now:
http://members.home.nl/theplague/gtl.JPG
All GTL reference voltages are derived from VTT. What would be an optimal NB GTLREF voltage with the NB and VTT voltages at default would still be optimal with the NB voltage at 1.5 if VTT was still at default.
Tests Systems
E8400 Retail 1.64 Vcore
DFI P35 UT T2R
1000 WATT Silverstone Op Psu
2X 512 MB DDR2 800 MHZ RAM
SINGLE STAGE -57C IDLE BLOCK
E8400 1.64 Vcore 5244 MHZ FSB: 582
Validate: http://valid.x86-secret.com/show_oc.php?id=346970
http://img73.imageshack.us/img73/3153/5244nk5.jpg
SP1M No Tweak
http://img73.imageshack.us/img73/5660/978sp1mfh1.jpg
At last someone put it clear for me :DQuote:
Originally Posted by Praz
Actually till now I set NB Ref GTL according to NB VTT - not just CPU VTT... time for "new approach" tests :)
That means once we've found the optimal GTL settings, we only have to change GTL setting if we change VTT. Did I get it right? :D