Braking OC Theory?

[Nekroze]

ok, this is unfortunately my first OC ever. i have a i7 920 D0 on a asus P6T SE mobo with water cooling... admitting the water cooling could be better but i lacked the cash for more fans on the rad wich should be here any day... back to the topic:

 

after gradually increasing the base clock, i finaly reached my target of 4ghz(although i have reached 4.3ghz aswell but more on that soonish) although everything i have read on "overclocking Theory" tells me that voltage is the prime contributer to heat and of cause i agree and can produce settings to show that if i felt like it. however upon reaching 4Ghz i had the voltage a 1.3v and to reduce heat wanted to lower the voltage untill i came up with atleast something to indicate it may be unstable. so i started to lower it however today it is a 1.25256v or so i forget the last number but the point is that i can even run it a 1.25v flat and no stabilty test has ever dissagreed with that, however and this being a very minor thing compared to the coming at 1.25v occasionally when i boot it stops POST and tells me that the OC is unstable but if i simply exit bios and the instantly close it it wont show up again until the next reboot sometimes it jumps 1 reboot. i have tried many different tests on many different settings there has never been a problem(nothing else was overclocked at all no ram bump i even lowered that as there was not the max stock speed present i used a fair bit lower) so is this a mobo problem or bios even though i have updated that and all. however the main thing that i contest with especially to bracking OC theory, is wait for it.

 

lowering the Vcore results in more heat!... not always but the first time i had the CPU at 4ghz on stock volts 15 mins of p95 and it hits 95c...startled the hell out of me so i put it up one placeand the OC seems perfect to me still the the very max i have seen it is 80c but thats after awhile and at a rate that was not so alarming. however i have made a folder for a bunch of collected results and notes for each stage of the OC going from stock 133 to 200 in 10 point jumps. I have the results saying the max temp at 4Ghz at 1.3125v presented a max temp of 71C now to me this is showing a situation where as more voltage is lowering the temp. is this normal as i have never seen this anyware and is quite perplexing. i have not recently tested the CPU again at 1.3125v but every result i have after testing each OC a few times after multiple reboots to make sure of any errors.

 

To me this is braking the OC theory or can some one explain it, also the fact that adding HT increase the max temps by 10 sometimes 20 i have only done this a couple of times as i dont like seing those kinda numbers. i have also reached 4.2ghz by uping the multi by one permantly(i have never had the powermulti thing on sense stock tests) but that required a Vcore of 1.26875v which is a rather large jump to me. the temps where as expected a tad hotter i also attemped higher but i could not reach ubove 205 bclk without 1.33+ Vcore which i am not willing to attempt at this point!

 

i think thats all i would really like some comments on this one, i am at the moment running the 4ghz at 1 step up from 1.25v sorry that i dont have this actull number atm may well add it latter and my ram is at 1603mhz 7-7-7-20 at 1.4 V the latancy and volts are stock and the stock mhz is 1333 so i liked that oc. the last question is more a summary question should i have my Vcore at 1.3odd or leave it around 1.25 and dose the heat make sense to anyone.

[Kingfisher]

After reading all of this I am not sure what "braking the OC" means.

As for your temp questions, the i7 chips will run hotter. It is up to you how comfortable you are with your temps.

I like to run prime95 blend test for 24 hrs and at least a run of ten intel burn tests.

 

If temps stay within my comfort zone(especially when the IBT) then I would be done with it.

To answer your last question, I would cloclk my cpu at the lowest stable voltage possible along with the lowest temps.

[Nekroze]

sorry if i was not very clear with that last night.

 

my saying "braking OC theory" was that it seems reducing the voltage is making the temps higher. even though it is still just as stable. and may i ask what is IBT?

[Nephilumos]

sorry if i was not very clear with that last night.

 

my saying "braking OC theory" was that it seems reducing the voltage is making the temps higher. even though it is still just as stable. and may i ask what is IBT?

 

 

IBT = Intel Burn Test

[thebridgerofdoom]

You are not breaking any overclocking theory. There is no fix "theory' as all chips overclock differently. Or could it be that the ambient temperature is lower when you tested your temps after you lowered your voltage? Or some weird factor affected your proc temp? Or the temps might be wrong?

 

PS: can u rephrase the 2nd last paragraph? not trying to be the grammer police or anything but i can't seem to understand what you are trying to say.

[Nekroze]

sorry i have a really bad habbit of explaining myself and not actually making any sense the 2nd last paragraph of the OP was more the end questions i am seeking answers to let me try to bake them down in a more understandable way and sorry once more.

 

1. is there any reason one could find that having lower voltages lead to more heat rather then less.

2. is it normal that HT increases the heat a great deal

3. is it completely normal that having stock Vcore i7 with 200bclk is stable but making the bclkc 205 requires almost 1.3Vcore to become stable.

 

also the main articles i read for learning how to OC was title OC Theory granted there was a separate OC Theory for i7 proc's but to me the entire process and main constants in overclocking is theory. as main constants go i mean a base number of any kind where the proc's speed is a multiplier of, increasing numbers usually requires increased voltage which increases heat.

 

these things and more are in my words OC Theory and when i see my proc getting higher temps on lower volts it leads me to believe the theory is somewhat broken or there is something wrong with my rig that i cannot figure out and i need some help from the best so here i am once more posting on OCC.

 

i hope that makes it a tad easier to understand my thoughts as sometimes i end up typing what i am thinking before structuring them into actual words or sentences. thanks all for having patients to atempt to understand what i say.

[thebridgerofdoom]

1. There might be other factors that is leading to this phenomenon.

 

2. Yes. Enabling HT also might make your overclock unstable.

 

3. Yes. Generally, most i7 920 require very high voltages for over 200 blck.

 

Why not do a check again? and post some ss.

[Nekroze]

at this moment i cannot do the tests due to time restraints in RL. probably tonight i will try to recreate the huge increase in temps by lowering the voltage. it could be that the water cooling loop was doing poorly at that time or whatever but i have never seen temps that high.

 

out of curiosity what temps do u get idle and load on that 920 of yours under water, and what OC you have. this is also a problem i have been working with recently temperature monitors and the tjmax crap... god i hate it. i have at the moment realtemp running tjmax's of 90/90/95/95 (=4cores) but i also keep speedfan up always as it is the program i tested every overclock temp with so its numbers accurate or not will show the increase or decrease with the phenomenon that i have mentioned.

 

when i retest it i will do one hour of occt and one hour of p95 to gauge the temps possibly if i feel its need will do a small ffts tests with p95 on top of the usual blend i do. if there is any contestation with these methods please tell me. but nothing that is to lengthy as i have little time at this point and i am lacking money to do multiple 24hr tests.

[thebridgerofdoom]

I haven't set up my main rig yet, but on my sec rig, the i7 950, im getting 68 to 72 degres on full load (prime 95) Overclocked to 4ghz

Edited June 25, 2009 by thebridgerofdoom

[ccokeman]

On an I7 heat is yor biggest concern. Running with HT on means you have the capability to load 8 threads instead of 4. By loading 8 threads equally with Prime 95 25.8 you can find out where your stability point is as well as the thermal threshhold that running 8 threads fully loaded will overload your cooling solution. I have a CO 920 that will do 4.1 GHz with 1.41 volts but runs at 72C fully loaded after 48 hours with a smallish WC setup. GTX 240, D5 pump and Swiftech GTZ block.

[Nekroze]

OK then i must have a real problem with my watercooling as currently although im getting new fans soon, my temps at 4ghz with 1.251Vcore is (with speed fan) idle 38 and load 70+ although with realtemp my tempretures come up 10C above the speedfan ones but i am not sure which ones are accurate as i do not have a clue what Tjmax is ok for a i7 920 D0. i think there must be something wrong i have checked the waterblock seating almost as many times as i have checked or even re-applied the TIM i am starting to think perhaps lapping either the IHS or Waterblock is in order... i got a dual 120mm rad but at the moment that is mounted to the top inard of my case as that is the only mounting point the NZXT Tempest has stock, which i no is not the best with heat rising upwards and all but at this point its all i got.

 

i have a 90cfm 120mm fan on the rad on the inside and the stock(very soon to be replaced with a much more powerful) 140mm fan ontop that i fliped to push the air at the radiator. the second 140mm fan on the top, right beside the one i flipped is on the exact flipside of the rad to the 120mm thats under. it setting to exhaust the air outwards however... it died not long after i got the case, still dont know why but it matters not.

 

the point is my rad is not up to par yet with fans on it but will be soon... is it so bad though as to result in 70+ degrees under load for 4ghz no HT, or could it be something else at fault other then the current cooling on the radiator. an a day or two i should get my new fans and i know that will be a HUGE increase in airflow as i have not seen to many fans as lousy as these stock ones

[Copie]

OK then i must have a real problem with my watercooling as currently although im getting new fans soon, my temps at 4ghz with 1.251Vcore is (with speed fan) idle 38 and load 70+ although with realtemp my tempretures come up 10C above the speedfan ones but i am not sure which ones are accurate as i do not have a clue what Tjmax is ok for a i7 920 D0. i think there must be something wrong i have checked the waterblock seating almost as many times as i have checked or even re-applied the TIM i am starting to think perhaps lapping either the IHS or Waterblock is in order... i got a dual 120mm rad but at the moment that is mounted to the top inard of my case as that is the only mounting point the NZXT Tempest has stock, which i no is not the best with heat rising upwards and all but at this point its all i got.

 

i have a 90cfm 120mm fan on the rad on the inside and the stock(very soon to be replaced with a much more powerful) 140mm fan ontop that i fliped to push the air at the radiator. the second 140mm fan on the top, right beside the one i flipped is on the exact flipside of the rad to the 120mm thats under. it setting to exhaust the air outwards however... it died not long after i got the case, still dont know why but it matters not.

 

the point is my rad is not up to par yet with fans on it but will be soon... is it so bad though as to result in 70+ degrees under load for 4ghz no HT, or could it be something else at fault other then the current cooling on the radiator. an a day or two i should get my new fans and i know that will be a HUGE increase in airflow as i have not seen to many fans as lousy as these stock ones

 

Might be worth listing your Loop so the cooling experts can have a look and find what the main issue is/could be.

 

As for mine, im requiring 1.38v to keep mine stable at 4.0ghz, so 1.3v is not bad at all, also im having trouble understanding what exactly you are getting temps wise, you are hitting around the 70 mark with P95 at 4.0ghz? and thats measured with speedfan?