I got an i7 920 as a gift and am wondering how high I would be able to overclock these things without going to water cooling. Would it be possible to get a nice stable 4.0GHz overclock on air with an ASUS P6T (non deluxe) and 3GB corsair DDR3-1333 ram? What heatsink would you recommend?If 4.0GHz isnt possible on air what is the highest I would be able to get?
Dang, I wish people would give me gifts that were that expensive. Some people will tell you that it will overclock easily to 4GHz on air but that generally isn't the case. It can certainly be done but isn't as common as the fanboys will have you believe.
It depends on your particular sample. The RAM will also hold you back if you're not able to clock it past 667 MHz (1333Mhz effective), though most DDR3-1333 has some headroom once you up the voltage to 1.65Vdimm.
The most common heatsink I'm seeing for people reaching 4.1-4.2 GHz air is the Thermalright Ultra-120 Extreme. It's also the most capable 1366-compatible heatsink I know of, once installed properly.
Typical stable peaks for the 920 range from 3.8 to 4.2 GHz. There are many settings involved, and it's very easy to mess up and think your CPU won't go past 3.3 or 3.5, so people have written overclocking guides for the 920. I'm familiar with one such guide on EVGA forums, but the terminology used for BIOS settings varies with different board manufacturers.
#1 do not run othos at 4.2ghz you can game it and use it but do not abusize it! remember the i7 is core2 plus the nb mem controller and heat is higher.
orthos at 3.8ghz is 72-73c as pictured above - not seated with yet with artic silver - i use artic for shipping since it is forgiving mx-2 is brittle
mx-2 is better
orthos at 4.3ghz i@1.43v s 80-85c unseated (new system) drops to 78-82c but do not run it here. no need to obuse your cpu.
with a single 120 fan non-true -the true-e = regular TR 120
if you can not understand what wrote it is cryptic research it!
dual fan true will run 70-75c at 4.3ghz
i would keep it at 3.8ghz unless you monitor the cpu!
If a machine can game but not pass Orthos/p95 or Linpack, I don't consider it stable. You can easily play a game that uses 1-2 cores or is poorly CPU optimized, but what happens later when someone writes a well optimized game that heats up the CPU to p95 levels? There are many 4.3-4.5GHz reports, but some are Pi-1M/8M/32M shots - probably not even game-stable.
lol! i do understand that you do not understand - why? the cpu is also a mem controller and getting too hot is not unstable
running a cpu unnessararly hot is not good for it - it may be stable but it reduces it's life. if the cpu (the i70) only use 2-3 of the 8 cores and 2-3 a little bit 10-20% during normal use then you can push it up higher then if you run all 8 cores at max
really not 8 cores there is 4 cores with 4 synthetic cores but you packing more data into 1 and calling it 2 and therefore it runs hotter
lol! ok a game will use 1 or 2 cores at max, background will use 3or 4 - there is 8 with the i7.
VERY GOOD POINT: First: "what happens when someone make a program for all 4 cores" you load profile 2 from the asus tools. the system ships with 2 settings!
setting 1 profile 1 is normal use - 4.2-4.3ghz air cooled
setting 2 = folding at home or orthos 3.8-4ghz
when WE ship systems AT warpedsystems WE HAVE FIRST RUN THE SYSTEM AROUND 10 DAYS:
1) we run it in bios for 1-3 days 72 hours is best DESK TOP BUILD NO CASE -cpu,mobo,psu.gpu
2) we build it - add drives and all other parts no os
3) we run it in bios built 1-3 days - slow build - we skip this for a rush ship and load os and test in os asap
4) we program the os and load all programs
5) run stress tests full ram. half and half, full cpu for 1 day for hour plus (in steps 5-7 we run the system5-10days)
6) we run multiple tests - prime 95 blend, orthos and stablity test for 24-72 hours
7) we run 3dmark08 while running the stress test above
8) we game it
the purpose is the ultimate out of box experince and to reduce the probablity of any failure anther 50-90%
i.e. psu will die fast or show spikes after 10+ days of testing your changes are 1/10 that of dell
when we are done we have a 1-2% return rate and 5-10% call back rate - 3% of the systems need anadjstment
we run the tests at 3.8 and 4.3 but we do not run the 4.3 for extend periods due to high temps
the system is 100% stable though all tests
running 3dmark with orthos and other test at the same time is the game
synthetic test
we only game a few systems, first few we test all with games, then as we go on none.
i have just told many pc builders how to do what we do - we do it better they anyone!
system is 100% stable, you want to fold at home or run prime 95 running profile 1 you want to game/im/listem music/surf the net all the same time load setting 2 at 4.2ghz. you run 40-50c gaming 45-55c mutltitasking
orthos runs 72c
Message edited by dragonsprayer on 12-30-2008 at 08:54:49 AM
wow! dragon i am in total awe of your setup bro... wanna switch?
------------------------------E8500 oc'd 4.5 @ 1.44 vcore with 92mm Zalman
ATI 4850 oc'd 680/1158 with aftermarket Zalman
Asus P5Q Pro mobo
2 gigs 800 Corsair ram @ 4-4-4-12
Reply to werxen
Dragonsprayer, how you run your business is not for me to comment. My concern is only with calling it stable when it has not run the same stress tests that another chip is stable with. If we take your 4.2-4.3GHz i7 figure and compare with P2's typical 3.9GHz, but you test with Crysis and they test with p95, then the notion that i7 oc's to higher frequencies on air is fallacious. If all the P2's ran Crysis, they probably could break 4GHz, too, because of the lower thermal output.
You are about to answer a thread that has been inactive for more than 6 months. If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.