Do you have a QPI frequency of 7400Mhz? Maybe I'm wrong but I think you mean 7400MT/s instead of Mhz.
Printable View
Batch # 3838A937
HT off
CPU Vcore 1.35000v in bios, 1.312 idle, 1.312 load OCCT - Prime, 1.296 load linpack
QPI/VTT Voltage 1.315v
CPU PLL 1.860v
DRAM 1.600v
PCIE 1.500v
QPI PLL 1.140v
IOH Core 1.160v
ICH I/O 1.580v
ICH Core 1.140v
http://www.saunders.fr/images/oc5.jpg
Dont know what you guys are referring to when you say MP, because my gigabyte BIOS does not have anything called that. Also, QPI and Uncore are both seperate multpliers of the bclock. Thats why you can have different QPI and Uncore clocks given the same bclock.
Say, you have bclock of 190...
You can have uncore multi of 12-16 not sure which ones my BIOS allow, but either way yeah. And you can have QPI multi of 36 44 46 I think are the options in my BIOS.
I'm sorry it was not clear but I refer to MultiPlier with the abbrevation MP, like you use multi.
Only on my Asus you can't litterly select a multi. You can only choose between a predefined selction of speeds which are ofcourse indirect multipliers. But they are not mentioned that way.
Could someone post the updated spreadsheet (as a spreadsheet, not image)? Thanks!
^^ He has HT disabled and that will remove a good amount of heat 5-10C idle.
:p: here is not even near 0º, right now my room temp is at 22ºc heres my idle temp now:
http://www.saunders.fr/images/oc8.jpg
the magic could be coollaboratory liquid pro, lapped i7, EK-SUPREME LT, HT off running on quite low voltage for 4.2Ghz.
id like to see how cold it would be on arctic condition :slobber:
http://www.saunders.fr/images/02.jpg
Final test:
=======
Prime95 v25.8 Build4 X64 - 17H10M
LinX v0.5.6 X64 - 1H34M (20 Loops)
Batch: 3838B079
CPU Freq.: 4005MHZ
QPI: 7400MT/s (even selecting 10000MT/s@BIOS it goes to 7400MT/s)
PPL: 1.80v
Multi: 19X
LOAD Vcore: 1.28v
CPU VTT: 1.27v
Hyperthreading: ON
LOAD core temps: 74/69/68/63 (Coretemp)
Cooling: NOCTUA U12P
http://img413.imageshack.us/img413/9...17h4ghzxo5.png
http://img27.imageshack.us/img27/6497/73131122ww2.png
Nice rig but where are the HDDs?
Just got a B to play with from a buddy of mine that lives near microcenter $229 and guaranteed B stepping cant loose :)
3837B317
3.91ghz
186x21 HT on
Vcore: 1.28
QPI/PLL: 1.275
Vdimm: 1.64
10x linX
GTZ + BI 3x120 3xslipstreams @ 5v
Load temps where 71/70/70/66 but didnt get a shot of them sorry. You can skip this post as I just popped it in though, I expect alot more out of it More to come!
Well, having the same batch number I don't think is any guarantee that you will have a good clocker.
I bought the 920 3836A772 which is the exact same chip as Cardnut99668 (3836a772) and he gets 211x19 @1.376 with temps at 85,83...
For the life of me, I can't go over 205 and the best I can do is Linpack stable but not prime95 stable (go figure..) at 20x200 and 1.49V. I can only get this stable at around 3.8 GHz and lots of volts (I'm on water so temps are much lower than Cardnut's)
So, I have given up with this 920. I ended up buying a 940. The extra multipliers (22) give you an additional 400 MHz to play with around 200 making it much easier to overclock.
I now have full stability at 4.1 GHz in one try. It will boot at 4.4 (22x200) GHz but not into windows. This is stock auto voltage. Maybe I'll see if I can get this to work!
It's also 3836A like the other two. But again, it won't go higher than 205. Perhaps there is some motherboard chipset influence in some cases. The eVGA Classified should let us know about this. First benches bring the Bclk up to 233.
So in the end it's back to square one? If I know that people got 3838B079/080 or 3837B079/080 to run decent clocks with around 1.35-1.4 it's still not safe to say that I'll reach that with the same batch. Or is it just a trend that B batches are better?
I will update and add the other features tonight.
seems like the Gigabytes and ASUS need less voltage than DFI one. Still havn't seen 4200 ghz with HT and Turbo on with an UT DFI...
With my 3837B395 I'm not able to get 4000 Mhz stable with 1,46V + and low DDR3 speed. System crashes with 191x21 or 200x200 =(
Perhaps i shout get a Gigabyte additionally to check out the differences.
Definitely. IMC seems to make the chip poo all over itself at ~4.0GHz without raising various voltages quite a bit, even on the good chips. Oh well, these are Intel's baby steps with IMC, and the chips still clock better on air/water than AMD's second generation Phenom, so poo. Pun. What?
Maybe D0 will have some IMC tweaks? If Intel can bolster the IMC slightly and lower temps ~10°C, I'll be a happy camper. I'm not asking for much, am I? Please, Intel, pllleaaasseee. I love you. I'll send you Canadian beer and bacon! :D