Starting to put my componants together see below for setup, anyone using an Core I7 965 be able to give me some tips on settings would be greatly appreciated.
Printable View
Starting to put my componants together see below for setup, anyone using an Core I7 965 be able to give me some tips on settings would be greatly appreciated.
Now Available 0903 Official :)
http://i287.photobucket.com/albums/l.../Untitled1.jpg
Now with TRUE
damn it when will all the coolit sales staff get back from CES
need my back plate...
quick question can i fit a HR-03GTX thermalright cooler backwards on this board with a Coolermaster v8 cooler and a hr-05/hr-05sli NB cooler? i also want to add the mosfet coolers(hr-09) but i dont think they will fit i need to take some measurements, but i dont know how to measure the space a HR-03 plus a hr-05 take together... cause im not sure how far the heatpipes for the Hr-03 extend how
noticed 0903 i official now...
I have a new stable low-voltage OC now on the R2E. See it here http://www.xtremesystems.org/forums/...postcount=1784
3.8GHz at 1.2V :shocked: Me thinks not bad :up:
What is the clock skews for? should we leave them on auto or disable them?
anyone can help me out here, I just put my machine together, asus extreme ii with latest 903 bios, intel 920, memory kingston 2000 hz memory, evga 8800gtx and corsair 1000hx.
Everything set at default works fine. When I go to the bios and enable the XMP settings for the memory, it shows up and I choose profile 1. But when I save the settings, the machine boots and tells me overclocking failed.
Shouldnt the XMP settings work without anything other adjustments?
Yes, in theory, but as all chips past 1333MHz are outside of Intel specifications their specific support is really dependent on how well the board manufacturer has done in terms of making the given DRAM compatible with their board. Quite likely, you DRAM is not "compatible" in that sense. But, don't despair. Just set all of the settings for your DRAM manually and try it that way. You will have to set the following parameters.
QPI/DRAM voltage
DRAM voltage
DRAM timings
Refer to the specs of your DRAM for values to use here. Quite likely you will need QPI/DRAM at about 1.65V, DRAM at 1.65V, and DRAM timings at something like 9-9-9-24-2T to start.
Keep in mind that QPI/DRAM voltage past 1.35V is outside of Intel's maximum specification (see http://download.intel.com/design/pro...hts/320834.pdf, page 22, Vtta and Vttd) and that there is at least a chance that going beyond this limit may kill your chip in the long run.
Thanks for the reply, this is the same ram. How is it he is able to just set it at XMP?
Kingston KHX16000D3K3/3GX
http://www.legitreviews.com/article/834/3/
Actually, he is not using XMP it seems. He just showed that XMP can be set, but he goes on to do a manual configuration of his memory. Try it. I found XMP to be less stable too than manually configuring my DRAM, Dominator 1600MHz.
Edit: I forgot to mention something in the OP. To reach 2GHz memory the system must be over-clocked, so you could also be failing there, but try manual config first.
ic
also noticed with some ppl that with different c states enabled
their everest scores for write is much higher.
personally i think this is a just a software flaw during the c state transition to give false readings
even booting takes longer with cstate..
cstate disable - vista 2 1/3 bars
cstate enable - vista 5 bars
jcool was having a debate with me about orthos
when he tested with prime only
orthos priority 10 like above vs linx latest version which stresses all cores
different in temp is only 3 C.
but voltage requirement is the same.
my best guess on this is because the IMC is being stressed also.
but he has a point that if i was running qpi voltage of 1.65 etc .. the difference is going to be bigger
but that defeats the purpose.
small fft is just to show the vcore
and memtest hci to show qpi dram voltage
with final confirmation of linx is a better way
the above is 72C
for linx is 75C
obviously he didnt test orthos and made a conclusion
prime for i7 is pointless as smallfft and memtest hci would show any errors
ok for skewing for noobies out there
when do u skew
first skewing doesnt give u lower voltages
skewing corrects ram speeds to cpu speeds and also IMC waves
so when to skew.. when ure running orthos like the above screenshots
suddenly it will bsod like in 20-30 minutes.
theres two possibilities
1.the LLC is at the borderline on the low vcore or the high vcore stress is fluctuating
2.skewing to correct the waves to be insync.
for 3.6ghz at set 1.2 i need cpu/ioh delay both at 200
for 3.8ghz at set 1.216 i need cpu at delay 300 ioh at 100
again this takes experience
those who clocked ddr3 with rampage extreme on skewing will find this board easy.
0903 vs 0805
mem tables for 1n vs 2n are different
1600mhz Cl7 bclk 200 1n
0805
DRAm A trip Delay = 58
DRAM B trip Delay = 60
DRAM C trip Delay = 63
0903
DRAm A trip Delay = 50
DRAm B trip Delay = 52
DRAM C trip Delay = 54
0903 linx failed in 4 passes
0805 linx passed 50 passes
so back to 0805 for 1600 Cl7 bclk 200
This is indeed not bad :clap: but i find a strange thing, in CPU-Z you're multi is 19 x 200 = 3.8GHz but if you look @ RealTemp it says 17 x 200 = 3.4GHz.
http://i42.tinypic.com/15zkvav.jpg
Now to be sure this is not a flaw or something i did the same test and with me it says CPU-Z 18 x 200 = 3.6GHz and RealTemp also 18 x 200 = 3.6GHz, weird ha :shrug:
http://i39.tinypic.com/2moouhs.jpg
well ocaddict i know the clock is because of c state enable for cpuz
but for real temp..hmm interesting
also for linx i usually test in windows diagnostic mode
cause u can use max mem and for 3gb kit 2.5gigs free
hello from germany. know somebidy something about a vdropmod? pencil or something else? my i920 drop and with the llc it is to much vcore for my opinion :( something exist to stable it?
eh llc increase the voltage for everybody afaik
set 1.2 example is actually 1.22v on my dmm on load
RealTemp accurately shows C-state transitions, CPU-Z does not. The screenshot was taken after the run was complete at which point the multiplier drops as the load is removed. There is nothing strange about it. It's just transitions at work :)
Edit: I will attach a screenshot here (see bottom, but later when XS stops failing to upload) of how everything looks while the stress test is running...
Yup, indeed, LLC will overvolt your set voltage somewhere in the range of 0.02V - 0.03V. Not really a bad thing IMHO.
hm, cpuz show at 3,8ghz 1,208v. in bios i set 1,225 for this and everest 1,25-1,263 :( this is al showing while prime is running