Interesting info thx.Quote:
Originally Posted by Metroid
Seems the part that gets unstable very fast is the uncore part. I think the large and "slow" 3rd level cache has its price.
To bad you need a high clocked L3 cache to rock SPI. :D
Printable View
Interesting info thx.Quote:
Originally Posted by Metroid
Seems the part that gets unstable very fast is the uncore part. I think the large and "slow" 3rd level cache has its price.
To bad you need a high clocked L3 cache to rock SPI. :D
You were right, WCG is not like prime 95 100% all the time :D
So 4.0Ghz is good enough to stay, temps are good.:)
I have changed some settings to work only up to 70% of the processor force, each processor will run at 70% = 8 = 5 and half processors :)
Any new OC news Sonofander?
Thanks for opening my eyes to the QPI voltage... I was having a brutal time getting P95 to run > 1m until I pumped that up a bit.
Also, since turning off Loadline and HT, CPUZ only showing 4 threads.... Normal?
HT on = 8 threads
HT off = 4 threads
That's it. :D
thought so, but, ya know... Never a bad thing to say I don't know. It's when you assume you are right and really aren't that you start getting into trouble...
New Test!:up:
Configuration:
Processor: Intel Core i7 920 (Nehalem);
Motherboard: Gigabyte GA-EX58-UD5 - Bios F4K;
Ram: Corsair 6GB DDR3 1600 cas 8-8-8-24 - 1.65v "TR3X6G1600C8D";
Cooling: Air - Asus Triton 81;
Sistem: Windows Vista 64 bit.
Frequency Cpu: 4515Mhz "215x21" - Voltage: 1.37v.:eek:
Super PI 1M:
http://img56.imageshack.us/img56/472...9438nj0.th.jpg
Super PI 2M:
http://img56.imageshack.us/img56/666...6726tp5.th.jpg
Super PI 4M:
http://img223.imageshack.us/img223/8...7443hd7.th.jpg
Super PI 8M:
http://img254.imageshack.us/img254/8...7504sz6.th.jpg
Super PI 16M:
http://img223.imageshack.us/img223/1...6013el5.th.jpg
Super PI 32M:
http://img254.imageshack.us/img254/3...7599qq1.th.jpg
Validazione:
CPU-Z Validator 3.0
Very Good Cpu!:)
Very Nice:)Quote:
Originally Posted by dj883u2
How much do you want for it? :)Quote:
Originally Posted by dj883u2
How does this look? Bolted down v2 :up:
Is priceless.:hammer::DQuote:
Originally Posted by Lukee
Wow. Very nice CPU. Care to share your BIOS settings?
I'll take some pics and post later :)
Lol, that's not even a gold cpu, that's a platinum cpu! BIOS settings do not matter. He's probably one of the two people that will ever get these cpus. Insane clocks on air.
Hi again
max i tested (not prime)
Core i7 920 @ 4400Mhz
Gigabyte UD5 (f4K)
3GB Corsair 1600Mhz
http://img149.imageshack.us/img149/5862/4400mhzbs9.jpg
:up:
Yes, it is. :)Quote:
Originally Posted by dj883u2
What the freaking hell?! 4.5GHz stable on god damn air?!! Can't say I'm not extremely jealous, sorry. :shrug:Quote:
Originally Posted by dj883u2
Indeed...that is one sick OC. I managed to find a setting I am happy with for my Core i7 940 on air. 3.95GHz (19x208MHz) 4-core 8-thread 64-bit Prime95 stable, memory at 1667MHz 7-8-7-22 timings, with 1.336V.Quote:
Originally Posted by 003
Not bad I guess, but not even close to the dude couple of posts above. 4.5GHz on air! WTF! Jealous. :shrug:
Edit: After reviewing his screenshots again...I am suspicious that this is all legitimate on air. Look at his temps and his voltage. I find this quite irregular. Maybe it is true, but I'm skeptic that it is on air. I'd say very good water or SS. On air I was able to SuperPI at 4.2-4.3GHz with about 1.4V at best and temps get up there for sure, especially with longer runs of SuperPI.
Maybe its just cold there? :hm:
I think you're right. Those temps are BS. At 1.3v on stock cooling and 3.8GHz, I hit load temps of 100C, and idle at 55C. Absolutely no way those temps are real.Quote:
Originally Posted by dejanh
He must be on chilled water or really good normal water. I'm just not buying that he is on air. Unless he took the PC outside in -15C weather or something like that, which HAS been done before. But that would seem to be a rather significant detail to omit.
superpi stable and real stable are two separate thingsQuote:
Originally Posted by dejanh
what does it do on quad prime?
This is very interesting to me. Back in965 days, i remember the asus p5b beating the :banana::banana::banana::banana: out of my gigabyte ds3, with numerous different revs.
In this thread, every brand of mobo seems to give the same general results. I'm going to be running chilled water w/ fans on the board, with stuff that'll require the 8 hr prime stability. I do it right now with a 3.92 GHz q6600 g0 an p5k deluxe.
Here are my 24/7 settings so far but I just can't prime more than 3 hours with one more tick on the Bclk CAN'T DO IT! So I am running 210X19 at like 3.985Ghz (gay).. I mean man I just don't get it, doesn't matter how much voltage or what I change I can't do 4.0Ghz stable... Anyway I am really happy with the performance of this machine it is awesome! Fastest computer ever that's all there is to it.... My read, write and copy are incredible so I'm cool with it....
Attachment 90428
And WCG stresses the CPU more then most games. :up:Quote:
Originally Posted by Metroid
So expect even lower temps when playing games (but maybe that will change when game get multithreaded, but till then we already have westmere :D )
Does anyone know if there is an issue running a 30 multiplier on 133bcl @1.35V with a 965?
It's the heat, you are running too hot guaranteed, just like me. With better cooling you would be able to lower the volts and up the clocks. With the stock Intel cooler I could not get the Core i7 940 past 3.4GHz to be rock solid stable. After upgrading to the CoolerMaster V8 I can do 3.99GHz rock solid stable, but not higher. I noticed too that a lot of people with better cooling are managing to lower volts and get higher overall clocks. Currently, I am settled on the settings I mentioned earlier as the balance between volts and clocks is the best. Need to put this under some serious water to go higher. Currently under Small FFTs the temps peak at 85C, under Large FFTs around 80C. Also, I reviewed the Intel technical specifications for Core i7 and they say that min and max voltages are 0.8V and 1.375V respectively.Quote:
Originally Posted by Guacamole
what about raising your memory timings?...have you tried that?Quote:
Originally Posted by Guacamole
That won't help him. I already tried it on my rig. I loosened all the timings and dropped the multiplier to the max to no effect. It is not failing on memory. It is unstable because of heat. I'm 100% sure of it.Quote:
Originally Posted by Ace-a-Rue
8 hours of prime still going @4ghz, 191 x 19 with turbo on,
Hey guys, I have a few questions
1) What's the word on VTT and DRAM voltages? Is it to just keep them within 0.5v of each other like the guys from bit-tech say?
2) Can someone give me some general settings for a 3.2 (20 x 160) Ghz oc on the 920? (I know it's weak, but it's plenty enough for my needs)
I was thinking:
Vcore - 1.2v
QPI/VTT - 1.2
DRAM - 1.5 (since I'll only be doing 1280 Mhz)
Appreciate your guys' help!
Why not ask him and let him post pictures?Quote:
Originally Posted by 003
I find the lack of trust disturbing.
I see ppl not getting 4ghz, and I get 4.2ghz easy as pie.
I saw Fugger get 5ghz on air
He is so full of :banana::banana::banana::banana: that if he gets that on air, I wont belive it, not even if he post images due to people postprocess images.
He is obviously faking that.
If I cant get 5ghz on water, he will never get 5ghz on air, no way in hell.
People even buying he get a special made cpu from Intel?
Come on, that is so fake.
If you question people, I would prefer you ask them directly.
And besides, if you question one guy, you have to question them all, even Fugger.
;)
You can get 3.2ghz mostly by just raise vcore a little.Quote:
Originally Posted by nemesisrobot
leave the rest at auto.
and yes, 0.5 with the vtt/ram thingy.
3.2ghz is nothing and you do that on stockcooler.
For one the temps are BS on air with the i7. No way that's possible unless he took the PC outside in like -15C weather (or something similar), which I did say was a possibility. And you know Fugger is a different case here, Intel is personally selecting the best chip out of more than 100,000+ for him, and then disabling various sensors. Also note how he has not posted since. I know if people questioned my legit overclock I would come to the defense.Quote:
Originally Posted by flopper
I dont buy this, isnt there a pic with some guy priming 4ghz on the stock cooler hitting like 98c and its still stable? :shrug:Quote:
Originally Posted by dejanh
Dudes, how to enable SLI on X58?
I just installed 2x 280 GTX and Forceware 180.84´s and thwere is no SLI option anywhere and GPU-Z says SLI is disabled.
Wow.. I can get 4 threads semi stable, but when I enable HT, it just locks up.Quote:
Originally Posted by freshy
What did you raise when you enabled HT?
Is it OK to run i7 920 on 4 Pin CPU ? When I use 8 Pin CPU, Keyboard doesnt work at all. Any suggestions ?
Has there been any testing running the RAM above 1.65v? Can anyone confirm it does in fact damage the chip? If so, can you link me to it? It would be greatly appreciated.
hi there.......
http://valid.canardpc.com/show_oc.php?id=459765
http://valid.canardpc.com/cache/banner/459765.png
just want to remind don't use auto setting on cpu/qpi/dram coz in my case it wupps up to cpu 1.41..v qpi 1.375v dram 1.8v.
core i7 965 on true black 120
asus p6t DLX bios 0905
2x1gb apoge GT 1600 + 1gb Adata 1600x
ati 4850 512mb
Yes, you are right, there is, and maybe there are a few total. There is even a guy here that runs low volts 4.2GHz OC perfectly stable on air. Golden chips and therefore not representative of the majority of parts out there.Quote:
Originally Posted by Bobbylite
Any ideas on why scores increased in time when oc increased :down:
Maybe cores are crashing? :shrug:Quote:
Originally Posted by Bobbylite
It's a single threaded app, and BL has his affinity set to core0 I believe, if the core throttled, or failed his run would have aborted.Quote:
Originally Posted by dejanh
Yeah, I thought so...hence the shrug at the end :p Something must be causing it however...Quote:
Originally Posted by ReverendMaynard
Finally, as it is a single threaded app, why not then disable all cores but one and go for max OC to one core to get the best Super PI score?
its on only 2 cores as disabling all but 1 was worse results
I have a qusetion about OC with HT on.
Let's say that you have a stable config w/ HT off. If you turn it on, do you just increase Vcore, or are there other voltages that you would have to increase as well?
My question is : HT or higher clocks?
Which is more 'useful' in 24/7 computing?
Depends on your personal definition of computing. I browse the web and play games and thats it so I just assume leave it off.Quote:
Originally Posted by bmilos
Hi
Core i7 920 @ 4Ghz
191x21
HT ON
QPI/VTT : 1.235v
Vcore : 1.31v Bios (1.264v Full)
Uncore : x16
Ram : x8
Gigabyte UD5 (F4K)
http://img255.imageshack.us/img255/3271/4ghz126vij4.jpg
:up:
Can you take a MM to your board and measure the actual voltage? That seems awful low for 4GHz, though there has been people that have some nice chips...Quote:
Originally Posted by mascaras
On a different note, I needed to get some proof that temps play a huge role in really limiting the OCs and that lower temps == lower volts and higher OCs, so I devised a little experiment.
My normal room temperature is about 23C. I cooled it off to about 10C-12C, and did an open case test at 4GHz (212x19) and a closed case test at 4.1GHz (216x19). I have tightened memory timings too to 7-8-7-20-2T and have the memory running at 1700+MHz. As I have anticipated, both Prime just fine with 1.35V and 1.43V respectively.
So as you can see, good cooling goes a long way here. This is just an air test with low ambient. Water should be better, and good water even better. The volts on the 4.1GHz OC should be able to come down to as low as 1.38V - 1.40V when the temps are kept below ~70C.
I will post screens in a few minutes.
Edit: Here are the screens...
Finally, a reasonable OC running P95. Temps seem a bit high, given my cooling, or aren't they?
http://www.mediafire.com/imgbnc.php/...3c4e2ed22g.jpg
i7 needs watercooling
Every CPUz screenshot I see shows 1T timing on the RAM?
My i7 is watercooled. Still 90c load.
Seems that all the revision C0/C1 920s are in Europe and England. Anybody know of one of these uber chips being purchased in the United States?
is the cpu in a loop by itself?Quote:
Originally Posted by bmilos
with NB, but thats it
That has got to be a bad mount, or high voltage.Quote:
Originally Posted by bmilos
I have my NB and a 4870, and I get to 72C only on LinX.
with how warm the NB runs on my board.. you'd get better temps if the cpu had it's own loop.
a 360 extreme radiator for just the CPU?????
Is there any other possibility? Would too much thermal paste cause it? Air pockets in the loop? There has too be some other explanation!
I have a BIE 360 and their not so great. I use a single pa120.3 for my cpu.
Air pockets often result in high water temps, did you "shake" your rad after you refilled it? (with shake i mean turn your rad in all different directions with the pump running).Quote:
Originally Posted by bmilos
To much thermal past also can cause bad temps, i usually only use a small blob or strip in the middle of the cpu and let the preasure do the rest. (never failed me till now)
i got mine from mwaveQuote:
Originally Posted by roller11
i stopped prime after 13 hours last night , it was going strong , stable as a rock @4ghz, HT on, turbo on , i might try get 4.2ghz stable today or might have a day off and just enjoy 4ghz for a day and get onto it tommorrow again :D
With how much heat these produce nothing is surprising...Quote:
Originally Posted by bmilos
In between Super PIs, has anyone with a P6T Deluxe notice this? This is not only my board.
http://chuckbam.com/Posts/IOAPIC.PNG
Something isnt right Im on modded fuzion v2 mounting with a MCR-320Quote:
Originally Posted by bmilos
I hit high 50c's @ 1.3vcore 4.1ghz and I have a 4870x2 in the loop as well.
what kind of v's are you throwing at it?Quote:
Originally Posted by bmilos
83C max small fft on air here mate, something is awry with your loop.
Is it a fresh build and if not, has it been flushed?
Well, i finally have a decent OC, prime is 26 minutes into it, but so far so good.
211 x 19
Turbo off, HT off
Vcore is 1.3 in Bios.
If I turn on turbo, I've tried raising Vcore to 1.45 and i still get BSOD or prime fails, same thing if I raise multiplier to 20 with turbo off.
Crazy if you ask me. I'm still working on lowering Voltage though. I haven't hit a floor yet.
What CPU PLL would be safe to go up to? Going to 1.84 helped get me to 190x20/21 with Turbo and HT on, CPU 1.35v. But at 191 I fail, even if I go up to 1.45 vCPU.
It looks like I've got the same problem. Any idea what this device is conflicting with?Quote:
Originally Posted by chuckbam
Attachment 90504
When I've raised BCLK and it wasn't stable, raising QPI/vtt helped.Quote:
Originally Posted by Aberration
Try traising it a couple of notches.
My CPU/PLL is 1.88 atm.
Well if you look at my sig, it is not from overclocking. A member on abxzone.com pointed it out to me today. Only on ASUS boards. Other boards with that same memory range have no conflict. PS: My SAS controller is disabled - can't be that one.Quote:
Originally Posted by msgclb
Heres my settings :up:
i920/ex58-Extreme/g.skill 1600 9-9-9-25/fuzion v2/mcr-320
Voltage:
1.375 vcore set in bios LLC enabled
1.41 QPI/VTT
1.86 PLL
1.66 Vdimm (probably could be lowered)
All other voltages Auto
21x200
Turbo enabled even under load (thanks f4j :clap: )
HT disabled
x36/x18/x8
PCI-E 101
CPU clock drive -800mV
PCI-E clock drive -900mV
Hope it helps
Well, I am really thinking to sell my Core i7 940 and get a 920 instead...but I guess I ought to set up a water loop first to see what I can get out of it then...Quote:
Originally Posted by Bobbylite
Same conflict on R2E as well...I wonder if it is the two types of advanced power management features that are present in the BIOS.Quote:
Originally Posted by msgclb
Strange...anyway here is a screenshot...
Umm. Fuzion V2?? How?Quote:
Originally Posted by Bobbylite
http://clubnboc.com/forum/index.php?topic=4.0Quote:
Originally Posted by 003
I have the SAS controller, the Marvel Storage Controller, Marvel LAN 2 and the Onboard 1394 Controller disabled.Quote:
Originally Posted by chuckbam
Your sig is Australia. Did you get it from mwave in the U.S.A.?Quote:
Originally Posted by freshy
Those are great temps. I am getting 72C @ 1.300 in BIOS on the GTZ.Quote:
Originally Posted by Bobbylite
I am on R2E and do not have this conflict.Quote:
Originally Posted by dejanh
Windows XP maybe? We have all been on Vista so far...Quote:
Originally Posted by Aberration
Edit: Well, I have what looks to be an answer and here it is... http://support.microsoft.com/kb/930062/en-us
Do what I did.Quote:
Originally Posted by bmilos
Put your radiator in the window with six fans on it.
1.7v on a q6600 g0, 3.95 GHz, and it doesn't step over 55c.
Gotta love wintertime. :)
yes :yepp:,Quote:
Originally Posted by Lukee
my 13 hour prime @4ghz is
19 x 191 HT on, turbo on
vcore 1.375v
qpi/vtt 1.315v
vdimm 1.64v
mem 1528mhz 9,9,9,24
qpi link speed x36
uncore frequency x18
pcie frequency 100
ive tried to drop vcore a bit but it crashed, im just priming now with vcore back at 1.375v but trying to see if i can drop qpi/vtt from 1.315v to 1.295v
but yea ya must lift qpi/vtt voltage or its just not stable
il try get 4.2ghz stable after i know what settings im staying with for 4ghz
edit.... nup she needs 1.315v for qpi/vtt
yep mwave in the USA,i got good friends over there , they sent it over for meQuote:
Originally Posted by roller11
over 2 hours of prime and still going @ 4116mhz
19 x 196 HT on , turbo on
1.45v vcore :sick:
1.435v qpi/ptt :sick:
max temps 80c, are these volages and temps ok for 24/7 ?
Well, I striked my watercooling loop last night (what a PAIN!) and re applied the thermal paste, changed the fans to pushing, refilled the coolant (no more air pockets) and it seems to have done the trick. I have a 18 hour prime stable 4.1 mhz @ 153 x 27 w/ 1.39375 and a qpi @ 1.40000 , HT on. Temps were in the 80c range at load.
that sounds more like it :up:
nice to know 920 is kinda keeping up with 965 :D
depends on what you own...
thats true, personally i expected more from the 965, they were $2k here in australia :shocked:
freshy has one of the nicest i7 stable overclocks. sad - this new arch isn't all that interesting to me.
Yeah, I am not going to lie but I am pretty pissed at the moment. I purposely bought a Core i7 940 because I figured they would be higher binned chips than the Core i7 920 and hence get better overclocks. I know this is not guaranteed, but at least in AMD world it held pretty true. However, not only can I not get better overclocks (tried 2 CPUs so far), but the i7 920 actually often beats my highest stable overclocks on air.Quote:
Originally Posted by freshy
A couple of points I noticed...
(1) Lots of chips out there will not run 20x multiplier. Both of my Core i7 940 chips won't. Above or below that is fine.
(2) Voltage requirements and temps for same clocks are a complete crapshoot. My first i7 940 would heat up ~15C more than the one I use now for the same volts and clocks.
(3) Jumps in volts for additional few MHz are stupendously large (e.g., going from 200x19 to 200x20).
(4) Having 22x multiplier (23x turbo) on the i7 940 is actually a major hinderance in comparison to the i7 920 as running this chip with 4-cores/8-threads at 22x200 is basically not possible thereby making 23x200 practically impossible. Turbo sort of gives a multiplier for free, except that means nothing when you cannot run your chip at its default multiplier to allow for turbo in the first place.
(5) At theoretical top clocks assuming 200MHz BCLK (even) there is a 400MHz difference between top clock on i7 920 vs i7 940 which likely translates to at least 0.3-0.5V (likely more) difference in extra voltage needed to achieve stability which is again practically impossible as most regular i7 920 will already require too much voltage to run at their theoretical maximum of 200x21.
Overall, pretty pathetic. With the current state of affairs and unless you are a top-notch enthusiast going for the higher possible bench score with some amazing cooling there is no point in buying either the i7 940 or the i7 965 EE. Simply put, they are a waste of money.
I am going to have to agree with you, on all fronts. One would think $700USD would get you a bit further than the 920. Ive seen consistent 1.3 - 1.5 overclocks from the 920 and I am having trouble reaching a 1.0 mhz overclock. What sense does that make?
Given, this chip seems to thrive on SS and ln2, that would be an extra $800 for a phase changer... Not worth it.
I am still keeping the faith, though. Maybe through some miracle bios or untapped voltage, the 940 and 065 will prevail. I've reached an unstable 4.4mzh boot, but prime95 has been beating me up from 4.2 on...
dejanh..yes you are right..therefore i want the people see what real performance they get for the money with my chess test!
If you look to my list then you see clearly how good the core i7 920 is compared with the 940 & 965.
JP.
The one thing that the 965 allows is more control over what you want to OC your memory to. Given that, that's a hell of a lot of money to OC your memory a couple hundred MHz.
My i920 HATES 20 X CPU multiplier.
I've managed to run 211 BCLK, but at obscene Qpi/vtt voltages.
I'm hoping that as time goes on, Gigabyte will put out a bios that will allow me to use that x20 multiplier.
Right now I can run 211 x 19 pretty stable at pretty low voltages. It's just not worth running higher for me. The increase in heat and voltages just aren't worth the possible gains.
That being said, if I saw a 965 for $500 I'd jump on it in a heartbeat!!
Lukee, why not try 20x200 with turbo (running a bios that leaves turbo always on)
have you tryed settings in my sig?
I've tried for like 5 days and nothing. lol.Quote:
Originally Posted by Bobbylite
I've tried 21x200 , 20x200
My bios setting are smiliar to yours :
Voltage:
1.375 vcore set in bios LLC enabled
1.41 QPI/VTT - I think I was a bit lower on this.. like 1.31
1.86 PLL
1.66 Vdimm (probably could be lowered)
All other voltages Auto
21x200
Turbo enabled even under load (thanks f4j )
HT disabled
x36/x18/x8
PCI-E 101 - I have PCI-E 100
CPU clock drive -800mV
PCI-E clock drive -900mV - I have 700mV
I'll give it a shot while changing those 3 settings.
The funny thing is with the 211 x 20, I run it at 1.3V in bios. I turn turbo on and it still gives me problems even at 1.45V in bios That's a lot of juice.
Quote:
Originally Posted by bmilos
Hi
thats strange , im using 3.8Ghz @ 1.20v with watercooling (and BAD Thermal Compound)
playing GTA IV for 1 Hour max temp = 46º
http://img380.imageshack.us/img380/1...sistemaqv3.jpg
and prime95 = 55º/60º
http://img87.imageshack.us/img87/9391/3800120vev4.jpg
:up:
Quote:
Originally Posted by mascaras
Not, strange. You are using 1.20 vcc. I'm using 1.4v vcc
Quote:
Originally Posted by bmilos
haaaaaaaaaaaaa , 1.40v (didnt saw that)
Edit: with 1.35v i get ~70º
:up:
I emailed Asus about this issue. Their support team is a JOKE!Quote:
Originally Posted by chuckbam
How the hell can bugs get fixed then? What the hell is their support good for if it can't be in the loop? Thats the most BS I've ever heard! :rofl:Quote:
Originally Posted by Asus Support
Anyone using a Peltier? I'm thinking it would be a good investment that would yield some results...
no need to laugh at the guyQuote:
Originally Posted by mascaras