ive read on afew sites that this actually works ...
has any one tried it here ?
with a Pentium M ? or a p4...
Printable View
ive read on afew sites that this actually works ...
has any one tried it here ?
with a Pentium M ? or a p4...
yes it works if you do it properly, you can get your cpu to run on much lower voltages at the same frequency.......
is there any patricular way u recommend I do it ?
run your cpu at the lowest voltage possible that allows you to run a benchmark that stresses it out, like running two instances of prime, or occt....dont use sandra....you should be able to get into windows and start the bench, but the pc will crash if you run the bench and start surfing the net and play a game or whatever at the same time - it is only partly stable.....E.G. i ran my 820 at 1.475V(rock solid stable), then gradually got it down to 1.425 so far, by starting to bench at 1.45V and going down incremantally once it became stable :)
prime will stop and say there was an error in the calculation...then you run it again and again, and eventually the number of errors decrease, until it is stable at that voltage....the longer you do it, the lower vcore you can achieve, well to a point anyway.....then you overclock a bit more and raise the vcore until its stable, then benchmark and slowly drop the vcore again and so on.....
the lower your vcore, the lower your temps, the higher you can overclock.....
keep your cpu disciplined, give it the absolute minimum voltage it needs to benchmark and then repeat unlil it becomes stable, could take quite a few hours though.....
you can also do this for memory....use s&m and do a memory test....eventually you will be able to drop your vdimm as well(hopefully!)
i saw the best results when i ran my 820 at 3.025GHz, and managed to go from its stock vcore of 1.325V to 1.020V - can you say ambient room temp of 24'C and load 33'C under dual prime and s&m memory testing?? :slobber:
the same applies to benching the cpu at a set vcore, then slowly raising its clockspeed until it is partly stable, then benching it until it becomes stable, but the first method seems to work better IMO, but play around.....
wow, that's a lot of stuff.....okay bubye :p:
EDIT: if it is too unstable and prime stops all the time, then you can initially use sandra for a few hours and then start to use prime once it is stable....just make sure to disable the hardware monitoring feature when you do the burn in with sandra, otherwise it is also going to stop all the time.....(and maybe do an s&m memory test simultaneously when you do sandra just to put that extra bit of pressure on the cpu[s&m mem test puts 100% usage on the cpu on its own anyway] remember to use the windows task manager to make sure that the cpu is indeed awake and sweating!)
So let me ask you... what do you do when you need 1.6v to just boot the PC, but it's 100% stable in windows?
drop the vcore to the lowest that allows you to boot into windows, as well run a benchmark, but will still give you errors....(the pc doesnt crash, the benchmark just reports calculation errors)
eventually the number of errors decrease
and temp's shouldnt be a problem because you are ginving it less and less vcore....
okay, i see what you mean, sorry, kinda tired - up your fsb until it is just a bit unstable and reports errors like i described, then bench and drop the vcore or raise the fsb even more, soz bout the misunderstanding :)
It's ok, but it doesn't work out for me that way... it's werid, maybe this is why i don't believe in "burning-in", this is on my old cpu, i'm using as an example. On my old 3700+ sandy i did, 305 x11/x3 @ 1.75vcore, i couldn't boot into windows when i did anything over 305 and i couldn't boot into windows when i tryed to do 1.73, so what is your solution now?
never had any problem like that, maybe the mobo is the limiting factor? but try run it at 300x11 or 290 and try lowering the vcore progressively...
in theory if you do the burn in at lower clockspeeds it SHOULD help you at higher speeds anyway....
remember to only burn in one thing at a time, if you are doing the cpu, relax the ram timings and use dividers, keep the fsb low(unless you know the mobo is okay) and make sure everything except the cpu is having a vacation.....
also, high vcores are very psu dependant, maybe the psu cant give the cpu the vcore it needs to go over 305 as well as giving the mem and everything else extra voltage - try taking out any HDD's you arent using, as well as cdroms, run everything(like memory) at default voltage, and only overvolt the cpu - this makes sure that your psu should handle it....(use dividers for mem)
i ran fine at certain speeds and voltages, but by just adding 0.1V vdimm the pc became unstable, and that's when i noticed the little 420watt furnace i had going in my psu haha
(just so you know, i know diddly squat about amd, but should be the same, i hope)
another way to do it, is to start at stock everything, and without changing the clock speed, lower the vcore as much as possible, then overclock as though that value was the default vcore, lower starting point equals lower end point, so start overclocking like the day you first set your hands on that cpu, using that lower vcore as your staring point, slowly raising it as you go along, of course benchmarking at intervals......
the longer you do it, the better the results, but the improvement lessens after a while as you get closer to the absolute minimum vcore
Using CPU Stretcher (a program in O/c Bible) I saw +100MHz after burning...
The idea is simple:
My P4 Prescott (custom watercooled) achieved 3.5GHz (stock voltage 1.400Volts multimeter showed 1.385)...
Stability program was Prime95 (a couple of hours)....
I increase FSB and I run this program up to make this CPU 100% unstable... :D
That's why I used ZZZ=100 (read instructions) and one thread at lowest priority....
The MSF was 3650MHz! (Vcore was the same)....
After a couple of hours I tested the CPU at 3.6GHz and it passed Prime95!
Soft burnin not hard is my opinion...
As about AMD CPUs I have not noticed high gains (around 20MHz)...
Using CPU stretcher a friend of mine could boot his Prescott 3.8GHz with Vcore=1.6Volts (before could boot his CPU if Vcore > 1.550!) :woot:
before 1.550Quote:
Originally Posted by MrSeanKon
after 1.6????
umm, am i missing something? :stick:
what is the point of being able to run the cpu at a higher vcore?
No aja you are not missing something....
I have tested four P4s (one CeleronD and 2 Prescotts) and nobody could boot if Vcore is > 1.6V....
emnezia=my friend was running CPU stretcher for 10 days (usually 8-12 hours per day) and he said me this...
I tried this for my Prescott but still cannot boot at 1.6! :mad:
okay cool
good program, but its confused the bolx out of me,
my cpu is a Pentium, M its not on the list, does that matter ?
is there a way to force "prime" not to check for errors.
becuase it passed 10 minutes, if i then stop it, it crashs after about 1 min
you know guys that there is no way cpu burn in can change anything. The only thing that might happen is that if it runs hot the contact between cpu and heatsink will get better alowing for more heat transfer. I experinced that with AS5. After few hot sessions i got a bit better temps and was able to use a slightly higher OC.
And about p4 pressie not booting with 1.6v? Coman i booted mine with 1.775v.
The only reason you were not able to boot is beacuse of your temps @1.6 would prevent you to do so. ;)
Someone should do like a Mythbusters thing on CPU BURNING IN, maybe it's a 50% thing? works for some, not for others? Doesn't work for me... at all i find the sweet spot on the first try, where it won't go up, won't go down and 100% stable.
ot mobo / psu couldnt support 1.6v boot power requiredQuote:
Originally Posted by railer
Do you mean for CPU Stretcher? :rolleyes:Quote:
Originally Posted by ibby
No it does not.
But if you run one thread then CPU utilization is 50%.
You can run it for any CPU but you have to play in the beggining to find the optimal settings.
Be very careful if you choose Highest Priority and you must have a qualified PSU...
Ill read the instructions again :)
OK ibby but Intel CPUs are more sensitive than AMDs (in my opinion) if you run them with high Vcores.
That's why I don't run my Prescott with Vcores > 1.45 Volts...
Some of you guys are doing it completely wrong! DO NOT INCREASE VCORE OVER DEFAULT. Run the highest bootable freq. at default vcore or the lowest bootable vcore at default speed. Then burn it in. In some cases I've seen cpus needing more vcore after several max oc burn-ins with high vcore. So, do it right. MrSeanKon has the plan!
Edited this scentence: In some cases I've seen cpus needing more vcore after several max oc burn-ins with high vcore.
either way you do it it will work.. as long as you are lowering the vcore at the same frequencies
I agree mate. :toast:Quote:
Originally Posted by Kjaks
Well a sticky thread for burnin is here at Ocforums.
The idea of CPU Stretcher is similar like spring:
As more as you stretch a spring then it is possible for a permanent deformation.
Similarly with CPUs...
But you have to be patient.
Like CPU burnin works RAM burnin (it is more possible to gain MSFs). :)
So do u guys have any recommended programs for using burn in???
P.S. Would be nice if you al so give me the link to download :)
http://www.d2ol.com/dOL/download_instructions.html
Thats one of the best. 100% load all the time, plus u can help out xs. If your not into that Prime95 is probably the best for CPU:(http://www.mersenne.org/freesoft.htm) and memtest is probably one of the best for memory (http://www.memtest.org/#downiso)
occt, s&m, dual prime, systool, stay away from sandra unless you are doing long term stuff....
also give oc bible a try, link in mrsean's sig......
yep, what moddo sed, those are really great as well, especially when you run cpu and memory testing stuf simultaneously.....
Memtest is a program that burns in in bios, isn't it?Quote:
Originally Posted by moddolicous
Like the one in DFI Mainboard??
But if we do with other MOBOs how to do it???
nono, memtest is run from windows, or you can create a boot disk and run it from there, like a recovery cd.......it doesnt seem to work in xp 64bit, so just use the boot disk one........
Overclocker's must have programs are here and hereQuote:
Originally Posted by Pekkie
As aja said O/c Bible is available following my sign. :)
MemTest is a program for testing if your RAM is stable (runs OK without errors) but some people believe that it burns also RAM.
Keep in mind that burn-in is quite different than stability testing.
Thanks a lots man!!!Quote:
Originally Posted by MrSeanKon
you guys have any recommend waterblock for 775
Toast works great! Especially in combination with ClockGen/anoter fsb adjusting tool. You just start Toast in normal priority, then adjust the fsb upwards to max with ClockGen. Then Toast will automaticly rise the clockspeed in very small steps. Im using it myself.. I found the oc bible a bit to complicated and old, but I can see it works with MrSeanKon.
Regards Pekkie! :toast:
I believe that safe and stable o/c can be achieve using BIOS settings.Quote:
Originally Posted by Kjaks
This is my opinion.
I know that O/c Bible is a complicated program.Quote:
Originally Posted by Kjaks
OK I am the creator of this package; I have written the source code; I know anything for this but I don't use all programs. :)
Sometimes I use AMD Crash temperature other times other tools etc...
In addition it covers many things for o/c which you cannot find to other programs.
It needs patience. :)
Your feedback/questions etc for this program should be in Bible's thread.Quote:
Originally Posted by Kjaks
That's why I opened an extra thread and give the link in sign.
I just read that your supposed to keep the frequency at stock while lowering the vcore? o_O
What about maxing out the frequency and then lowering the voltage to a point where prime gives no errors? Ive been doing for about 8 hours already. <<;
I found this thread for Intel burn-in.
Useful I think.
Reznik Akime :rolleyes: I followed this idea (two years ago with 2 Bartons) and no significant gains...
I don't remember exact values but I think MSF was 2450 MHz (Barton 2800+) with Vcore=1.925 and after this process my CPU could pass at 2463MHz with 1.900 Volts.
Multiplier was 12.5X (locked CPU).
Ok you may ask me "Man this thread is for Intel burn-in" but this process similar for Intels and AMDs but you should avoid high Vcores cos Intels are very sensitive.
My current CeleronD and previous P4 Prescott became more unstable after CPU burn-in.
Vcore was 1.584 for both CPUs.
So at stock speed I just drop vcore as low as possible. Now for the actual burn in, do I run two instances of prime or just one?
Of course if you have Intel HT CPU or any dual core CPU you should run two instances of Prime95.
If you don't know click here
In my opinion Prime95+StressPrime2004 are not for burn-in both of them are stability programs. :)
Also forget SuperPi for Intel stability tester is not sensitive cos poor Intel FPU. :)
Yeah, I am running D2OL now. Just realized this burn in time should not go wasted. Vcore at 1.18 :D I think this is good for now. I will let D2OL run for about 5 hours before I shoot for lower voltage. Then tomorrow evening I will start increase FSB and voltage slowly.
p.s. just to clarify, this is x2 3800+ :woot:
windows would freeze at 2450 at 1.28v. Anything I can do before I increase voltage?