^^^ so basically lower multi, higher BCLK will provide lower temps as you can lower your QPI/VTT voltage and more stability? I use a excel sheet sometimes to do gather some data together, helps out a million :)
Printable View
^^^ so basically lower multi, higher BCLK will provide lower temps as you can lower your QPI/VTT voltage and more stability? I use a excel sheet sometimes to do gather some data together, helps out a million :)
Some results from tonight, far from final settings, lots of tuning to be done!
http://img.photobucket.com/albums/v4...estar/bnch.jpg
With those temps, hell to the no :p:
Check sig :)
edit: sorry if it's not clear, everest temps are in system tray:
Yeah I agree... my PC totally hates the 211 x 19 setup .. prime gone in 10 seconds .. in fact 190x21 it doesnt like too much either ..
I'm on Air .. while other P6T/i7 920 guys here are loving those setups on water .. so I can only think extra CPU cooling or NB cooling is doing the job ..
Classic Satch etc .. you guys on water are you putting blocks on the Northbridge .. anywhere else ??
I have the NBMAX on the Northbridge. Was also thinking about getting the EK MOSFET blocks and the EK NB/SB full waterblock. Going to hold off on that until after the holidays. Hoping someone else gets them and I can see if there are any performance gains.
I am just cooling the CPU. NB stays relatively cool on my board, it has a little fan on it from the factory.
Classic, what was the longest prime run you ever got at 200x20 ??
For what its worth I got about 1.5 hrs Prime with these settings:
200x20
HT On
DRAM 1603Mhz
1.4V CPU
1.68V DRAM
1.35625V QPI
SpeedStep/Turbo Off
LLC Enabled
CPU/PCI Spread Spectrum Disabled
C-State Enabled/Auto
ACPI 2.0 Enabled
Rest ASUS Default
Gone retry tonight to see if it wasnt a fluke .. but I might be happy enough with this for everyday gaming use (which is all I care about in long run) - played Crysis & Assassins both over an hour with these settings no problems at all ..
Also primed for over an hour (stopped manually) @ 1.45V CPU .. but with crazy temps ..
I think I got to about 40 minutes before it failed at 200x20. But I've run over 14 hours at 211x19 with 1.36vcore.
HT is on
QPI-1.35v
Speedstep/Turbo off(systems defaults to OFF when using anything other than AUTO for a CPU multiplier)
DRAM voltage is 1.66v
The rest is the same as yours. There's something obvious that I might be missing but I'm happy with 4.0Ghz @1.36v for now.
I figured out why I couldn't post at a multi of 196, I forgot to disable turbo mode, so 210x21 equals 4.4 GHz which at 1.35v I dont think will boot. Imma try it again and disable turbo mode to get 4.0 GHz then try to work my way up to 4.2GHZ at x19
Interesting, post back in when you get a chance to monkey around a bit more. I'm not sure, but I heard once that if its not at least 3 hours Prime stable it might not be truely gaming stable (not sure from my own experience, just heard that from somebody once).
Yeh .. looks like I might have the 200x20 curse like most ppl ..
I'm currently running these settings from freshy ..
http://www.xtremesystems.org/forums/...&postcount=657
And I've just hit the 100 min mark .. farthest 4Ghz I've ever gone with lowest temps too ...
I think the key is low QPI/Vtt Volts for us folks on Air .. I had my best effort on 200x20 with a low Vtt above there as well ..
When it falls over gonna try 200x20 again with even lower Vtt
Hmm, I have uset 1.35 - 1.4V QPI voltages, should I try to lower it :)
My system won't post at 211 Bclk with a qpi/vtt lower than like 1.75. I've tried lower and it sits there.
I'll give it another shot when I get home, but I have no clue how you guys are doing this with 1.39
Well my head is officially done in ..
No sooner had I achived my own Prime Personal Best (over 2 hours and still running) with flashys settings from above there.. now when I try go back to those settings.. I can barely even get more than 1 Cinebench run .. sometimes cant even get into Windows !
I think I've given up .. I've gone back to basics .. set 200x20 .. the rest Auto .. it seems to run multiple loops of Cinebench and all my games .. thats good enuf for me ..
I have an Asus Rampage II Extreme. With the stock fan I can get my i920 to a Bclock of 160MHz. I raised the CPU voltage to 1.4V but this did not really help.
I have a Mach II GT cooler which cools the CPU down to -50C. Under a 200W load it goes up to about -10C. The thing is that colling the CPU did not change the Bclock performance. I was still only able to get a Bclock of 160MHz at -50C. Once I had the CPU at -50C I also raised all of the CPU voltages to their maximum safe settings (core voltage at 1.5V) but this did not help.
I am assuming that it is not the CPU that is holding me back. I reduced the memory speed so as not to exceed the rated speed of my memory so that is not the issue. Can anyone give me any pointers?
OK, my system is up and running (P6T with 920).
Must read a little bit about Nehalem Overclocking (coming from S775), but don't have the time before christmas holidays begin.
So would you guys be so kind to give me a little starting help ?
Some good starting settings and more important values not to exceed (like QPI voltage or the well known dram voltage of 1,65v) would be awesome ;)
got me a L383B107 Batch, maybe someone has tested one already and knows how it clocks ?
lol 1.75V qpi... talk about stressing the chip. :p:
Ok, I'll try 1.4 Qpi/Vtt and if that doesn't work, I'll come back and post my system specs as well as bios settings. Hopefully, someone out there will be able to tell me where I've gone wrong.
I've had to go back down to 195 - 200 bclk since I really didn't like those voltages.
Yeah i wouldn't go over 1.5v while testing max qpi / bclk etc. One reason is because that's a sort of limit i set myself based on what i've seen and another is that i already know i can get up to 216bclk with 1.44v stable, and that i won't be wanting to go that high on my final OC anyway.
Really have to play your own game with voltages, for example i'm stable @ 4.3ghz with 1.456v, but i now know i'm also stable @ 4.2ghz with 1.37v (perhaps less, not finished testing). So there's no way i'm going to set a whole .08v higher for only 100mhz extra when the clock is already so high that it won't make a difference to me personally.
Please do post your bios settings though, can't figure out why you need such ridiculous VTT. When i started to find max bclk all i did was:
Bclock 200 (my starting point as i knew it was stable already)
CPU Multiplier 12x
Uncore Multiplier 12x
QPI Multiplier 18x (36x in bios)
Memory Multiplier 6x
Vcore 1.275v
VTT 1.46v
Vdimm 1.64v
Then i started raising bclk until i hit walls. Under this setting your roof is only going to be how high your QPI clock tops out - meant to be around or just under 4ghz.
I'll give this a shot as well. Iirc I did try it when I first got the chip and I couldn't go over 200 Bclk with 1.46vtt. I then raised it to 1.55 and was able to hit 205.
Anyways, I'll try this again, maybe I missed something
I am running 6 x 2GB memory, so would that require a higher vtt?
I'm wondering if anybody here has written a guide/tutorial about overclocking the core i7 :D:yepp:
Plenty tips kicking around:
http://www.xtremesystems.org/forums/...=205688&page=4
http://www.linustechtips.com/how-to/...clocking-guide
Not sure, maybe someone else running the same can comment. I wouldn't think it would need a whole lot more juice. Maybe at higher (1600mhz?) clocks, but for now try reducing memory multi to remove that bottleneck and see what bclk / QPI / Uncore you can get :up:
No, there is no definitive guide yet. I think it is still to early for a stick guide. We are all still testing, trying to find the holy grail. I think more so, we are all waiting to see if any chips die/degrade, and if they do, what was the cause.
He he, I hear ya. I've been begging for a step by step for a while. I guess if nobody else makes one I'll go ahead and document and photograph my entire experience when I OC here in a few days. Only problem is that its my 1st OC, so I'm not so sure how professional of a source I'll be. ;) I've been trying to do my homework though. From what I can tell it involves lossening everything up (to maybe 12) and then seeing how high you can safely get youe bclock... then you keep that and start tightening the next component. Does that sound right guys?
his is only $500
Most of my components are in my signature. I have been meaning to start a thread of my build, and will do so tomorrow - my package for performance pc's is scheduled for delivery tomorrow. Once I have it, I am going to do my final build.
I am also patiently awaiting the RMA of my 4870x2 I sent to Asus on the 8th. Haven't heard anything thus far, but am trying their live chat right now...
edit: I sent it 2-day air, yet they are claiming they didn't get it until the 16th. Such BS. Asus is trying my last nerve... I very well may sell of all my asus components and boycott.
Ok guys, as promised. I tried booting into windows at 1.35 up until 1.69 until I was finally able to do so..
I went with the following setup to remove memory and CPU votages out of the equation.
CPU Multi 13x
Qpi Multi 36x
Uncore Multi 15x
Memory Multi 6x
Let me see what else
Turbo Off
HT off
Performance - Standard
any ideas?
One source said some bclk's would max out at 190's, some 200's other ~220. Mine has a very sharp cut off at 215.
With multi 15, QPI 36, UC 16x, mem 8x, Vcore 1.4, Qpi/Vtt 1.4, DRAM 1.66, HT on, my max bclk is 215, and can run spi there. 216 will not boot.
Then I can take it outside in 5C air, UC 14, mem 6x, Vtt 1.68, +/- DRAM 1.8+, and +/- red line all other volts, bclk will then go to 217 barely load windows then freeze. no higher regardless of settings. I walked both Vtt and DRAM up and down, (one poster said got high bclk with increasing dram volts +vtt both despite low mem setting, is why I walked DRAM up +/- vtt nothing worked.
your bclk seems to go from 205 to 211 like mine goes from 215 to 217. I think both are just near their limits.
Yep. I can run 200 Bclk on <1.4 Qpi/vtt, but for me to run 205, I need about 1.55-1.65. I can't remember off the top of my head, but somewhere within that range. You saw what it took to get 211 so for me, I'll have to settle with 200 x 21 for now.
Btw, when it "hang" and doesn't boot, the bios error code that I get is 6F which on the Gigabyte EX58 Extreme says:
1. Initialize Floppy Controller
2. Setup Floppy related fields in 40:hardware.
I don't have a floppy connected. lol.
Sounds like you're limited by your chip after all then lukee. What's with all those other settings though, cpu pll etc? I've hit pretty high clocks without having to touch that stuff (not even sure what it does), i wonder if they're high enough to cause instability.
I agree if you want to 'challenge' someone to a benchmark create a thread for it. Alternatively, there are probably a hundred reviews by now detailing how much faster i7 is than penryn at any given task..
4.3 prime stable 10 hours - HT off, unlikely I will get 4.4 on air but must try!
The spread sheet was rge's design - good approach!
I think 4.3 on air is the max for me - 4.32 @ 1.55v is way to hot!
Interesting that it actually got through linpack x5.
On real temp - when it reaches 100C - thermal status goes to "HOT" for that proc.
I've managed 4.2ghz so far, but i keep getting random freezes and bsods, i think its a memory issue though. Running 1.475v in bios, somehow it ends up at 1.55v in windows though. Should be able to push farther as temps are still under control, ive just been too lazy to care as this thing is so much faster than my q6600 ever was. Running at 3.2ghz (dummy overclock in bios) daily as its rock solid stable and still plenty fast.
Ya, i think i also need to update the bios too. Its running 8x on memory and 16x on uncore.
veedubfreak - I have standard kingston 1066 memory, no heat spreaders - I am using qpi x36, uncore x16 and memory x6 - it was suggested earlier in this thread that we might need to keep uncore/memory freq. ratio to 9/8.
rge - did I get that right?
That was dejanh with the uncore to qpi ratio of 8:9 on i920's. uncore to memory is 2:1 or higher.
Yeah, i should have said before all other voltages on auto. There's a post on this thread from way back with someone who was using all sorts of voltage increases, cpu pll / qpi pll, IOH core increases, etc and some others from that page of the bios. I tried out his setting with all those tweaks and while it was stable, i then tried it with all of those on auto (just keeping the vcore, vtt, and dram), and it was also stable!
No point in raising values for the hell of it eh :D
If they allowed you to set your own multis, from 12 to 50 or whatever, everyone would be happier.
Is there a reason they only have for example 3 to choose from on QPI, does anyone know?
965 = everything unlocked.
And expensive... ;)
Aha, even inbetween the x36 / x44 / x48 for QPI? I figured it was just the cpu multi that was unlocked. That really increases the value of the 965 then, how many people are complaining about not being able to choose a certain multi for qpi etc.
They are pretty expensive though, £850 is the absolute cheapest i could find available here when i was buying.
Still, cheap as chips compared to the QX9770!
Here you go...
http://www.xtremesystems.org/forums/...39#post3522239
:welcome:
reverandmaynrad: I have the 965 and so far several folks with 920s and 940s are doing better than I am.
QPI options for me on ggbt extreme mobo are: auto, slow, 36, 44 and 48.
Is this the same for the folks with 920 and 940s?
QPI needs to be much more adjustable - like uncore - x12 to x48.
My I7(RMA) is going to be shipped out tomorrow, I hope that I get I7 B' batch, as it seems to run on low Voltage as per the members in this thread.
@ bmilos
got the same case as me, looks good, here is a few shots of mine:
@ GNU
Check out the pics if you'd like to, they are cell phone pics, but oh well, you can still see the system.
Mine consists of thermaltake pump and reservoir, Koolance 120.1 radiator, and Swiftech Apogee GTZ water block.
http://img135.imageshack.us/img135/5...3081923an9.jpg
http://img135.imageshack.us/img135/2...3081924bi5.jpg
http://img135.imageshack.us/img135/4...3081926mp9.jpg
I've noticed the same thing, but I have also noticed many turning HT off, which I am unwilling to do. As a photographer, I built this rig for editing and adobe CS4 seems to use the additional threads.
I think, after a month of obsessing over trying to find stability at 4.3 or above, I am officially giving up. I have put forth way too much time and effort. If, by some bios update or some other miracle, it becomes a possibility, great. If not, 4.2 @ 1.425 with load temps around 75 is fine. knowing that I can run at 4450mhz is somewhat of a prize.
I can no longer justify spending all my free time rebooting my freaking computer!
One thing that I am considering is just buying a SS. I hate throwing *more* money at a problem, but, at this point, anything to avoid the BSoD.
Xello: For qpi I only get slow, x36, x44 and x48 and I'm not sure it is a processor locking issue.
I would hope it is a mobo bios issue that a bios upgrade can fix!
From what I can tell so far - at least with the 965 (on air) - heat does seem to be the limiter -
What a tease last night. I tried the following:
200 x 21
QPI 36x
Uncore 17x
memory 8x
It was prime stable for over an hour, so I was getting a bit hopeful.
I stopped prime95 and cranked up Everquest 2 and some sounds on the speaker and lockup!!
Oh well, back to 15x uncore and 6x memory.
Cant run QPI 36x Uncore 18x memory 8x ? I run 18x but fail at 17x :P
Yes, but just uping the cpu mulitplier also comes with much more heat.
OCing with lower multi and high bclock keeps temps down - my results so far -
I agree. I don't know what kind of binning Intel are doing with these chips, if any, because so far we've seen plenty of 920 and 940s going to insane clocks just like some 965's are, under xtreme cooling, obviously the 965 is just a more efficient option if you are doing that because of the unlocked multi.
On a positive note, their manufacturing seems to be untouchable at the moment - the vast majority of chips are going to or near 4ghz no problem. With penryn C0/C1, this was definitely not the case, you could easily get a QX9650 and struggle to get to 4ghz under air / water. I did, mine needed really high volts for it.
I'm really impressed with i7 and x58 in general in this regard, the bios's have been coming hard and fast and pretty much every board has a feeling of maturity about it even after such a short time. It's easy to complain when things go wrong but it should be noted when things go so right :up:
yep i admit im happy with my 920 , 4ghz - 4.2ghz stable with HT on is good, but i do feel sorry for the ppl that spent more on 940's and 965's and are almost getting the same clocks , ok maybe 4.3 - 4.4 prime stable* if there lucky, if the 920 goes to 4.2ghz prime stable then the 940 should go to at least 4.6ghz prime stable and the 965 at least 5ghz prime stable, but thats just my opinion :D
Do we pay much of a performance penalty by setting qpi to slow (gigabyte boards).
It looks like this is the setting I must use for qpi - if I want to use high bclocks with HT tuned on.
I think qpi on slow is something like - 100xbclock, so for 200 bclock that would give a qpi speed of 2000 vs stock speed (for 965) of 3239 mhz.
Not sure what i'm running on my QPI, but my evga board only has "auto" then 16x-40x on the uncore so i couldnt drop down to 15x. I did drop the memory multi down to 6x and that seems to have stabilized things. This board is still giving a lot more voltage in windows than i am setting in the bios though. I dropped the voltage down to 1.4v in bios but in windows its still running at 1.475. Going to knock it back to to 1.35v in bios and see if i stay stable.
I game. When I've been able to have a system run stable for 2 hours on Prime, I don't have any issues gaming. My PC isn't on all day, I just use it for 2-4 hours Max.
On another note, because of this chip's inability to hit over 200 Bclk with reasonable qpi/vtt, would you recommend getting another one and selling this one? Obviously I'd mention that anything over 200 bclk would be near impossible without high piq/vtt to the seller, but I'm sure there are people out there who would be ok running 200 x 21 or something below that.
link for linpack ? please
ive been more than 20 hours prime stable blend and small ftt and my pc still crashed when running 2 smp cllients with ht on, priming now with f4r bios 16 hours still going
191 x 19 + turbo on, HT on
uncore x16
mem x 8
qpi link x36
vcore 1.38v
qpi/vtt 1.35v or1.375v not sure i forgot , half asleep when i started it
vdimm 1.66v
still need to try drop voltages a bit more
Looks like you have a really nice OC going there. How far do you intend to drop those voltages, do you think your current ones would risk you eventually burning out your CPU early? I've seen people all over the map with voltages... its come to the point where I'm not even sure what considered OK anymore except Intel's offical quotes (1.65v).
you can use LinX which is a gui for linpack
http://www.xtremesystems.org/forums/...highlight=linx
Quote:
Core i7 owners, to fully stress your CPUs you'll need to run 2 instances of LinX at the same time.
I dont plan to do Extreme overclocking with I7 on my Stock Cooler..... till i find a 1366 kit for my Tuniq tower or little cheaper heatsink.
What do you thinks about below OC on Stock Cooler :
Bclk : 200
CPU Multiplier : x15 (3 Ghz -3.2 Ghz)
QPI : 18x
Uncore : 17x
Mem : 8x (1600 Mhz)
hi bro, just thought I'd share this with you, with the stock cooler I was able to do 170x 20 = 3.4 GHz ram divider 8. uncore x16. The voltages were all auto except for dram which was set to 1.65v. Max temps were about 80-90ºc but that was during prime. Daily use and 3dmark runs all stayed reasonable at around 60-70ºc. I say reasonable because these things run scorching hot out of the box so, you get my point.
My 920 absolutely HATES the 20x multiplier. I can't get it to pass a 3DMark Vantage test at 200x20, Turbo On(4.2Ghz actual) to save my life, even at 1.43v. However, it runs at 211 Bclk x 19 all day long at 1.36vcore.
I think that I'm missing something simple but I've done extensive research and I can't figure out what the hell is holding me back.
Cooling is not an issue because my max temps @4.0 range from 69-73C across all four cores with HT enabled.
Yes, I was lazy and put a Koolance Exos-2 LX with an Apogee GTZ water block on it but my CPU is the only thing in the loop and it seems to be working very well. Plus, I must admit, if you heat this stuff in hot water and then apply it correctly, I think it's paid off royally:
http://www.acousticpc.com/innovation..._compound.html
My temps are consistent with the most "hardcore" H20 rigs and I think it's because I used the best water block and TIM, not to mention that my CPU is the only thing being cooled.
Do you mean 10 hr small fft + 10 hour blend, or 20 hour small fft + 20 hour blend? You shouldn't waste your time with small fft, it's for testing max heat output - good for testing your cooling. It puts less stress on the cpu than large fft, which is why Blend is the best option :up:
I always use 24hr blend as the pass mark for my final OC's - has never failed me.
i have a 965xe. Am i right in saying _ALL_ i need to do to overclock this is just increase the multiplier (along with Vdim) and not worry about the QPI? I am currently running at 4ghz stable with:
BLK: 133Mhz
Multiplier: 30x
Core voltage: 1.275
My QPI is currently 3.2GHz.
I use everest, but for some reason it does not show my real cpu speed or multiplier, for that i have to use cpu-z, any dieas ? is everest not compatible with i7 ?