Is overclocking safe?
Author
Many people buy or build computers and raise the voltage to the motors in their fans and drives to get more speed out of their machine. Just as well, many car racers connect a container of dry ice to their engine intake to make the air colder and get more horsepower, but this in the end will KILL your engine.
So, with that analogy out of the way, I ask the question: In the end, is it safe to overclock your machine''s components? Does it damage your mobo or other parts? Is it really worth it?
Granted there are other ways to overclock your computer-- in MaximumPC magazine, they had one guy who stuck pipes into his tower. That''s a good idea, I think. Before that, they had a guy who built his computer inside a fridge. Not so good.
So what''s the safest way to overclock your computer, if it''s safe at all?
"Safely" overclocking your computer is an odd statement...
There really isn''t a "safe" way to overclock.
CPU manufactures will claim that any thing you do at all to increase speed will cause massive damage to your cpu. This isn''t neccisarily true.
The most common overclocking is to increase the FSB speed. Lower end celerons were popular for this, because they run at a 66mhz bus, and could be overclocked on a 100mhz bus.
Unfortunaly, the multiplies are high enough now, that this big of a jump would force the processor too fast. IE. a 400 mhz celeron, put on a 100mhz FSB, would run at 600mhz, this is a BIG jump.
Many brands of mobo''s now have a bios option to adjust the fsb by 1 mhz increases. This allows for smaller increments, but also causes PCI and AGP slots to overclock, which is more likely to cause damage than just overclocking the cpu (from my knowledge atleast)
I''ve heard it said that you can safely overclock by as much as 4mhz without worrying. I can''t verify this though.
Intel has often talked about excelerated electromigration (think that''s the word...) which is basicaly that when current is passed through the chip, the electical pathways can be altered.
There are some facts that are commonly circulated among pro overclockers:
Electromigration on a chip that is overclocked to X, should theoretically be the same as a chip clocked at X by Intel.
Electromigration takes many years, on the order of 5 to 10 i believe, to have an effect. Most people say they will have upgraded by then anyway so it doesn''t matter.
Cpu''s are made in large sheets, and a few from each sheet are effectivly torcher tested to see how fast they can run (better quality run faster), as all processors on the sheet will have approximatly the same quality they will be given a multiplier that is a conservative estimate as to the processors ability. So they should be capable of running faster.
There is a theory that companies like intel take some chips that could easily run at, say 800mhz, and lock the speed lower, say 600 mhz. Why this may seem insane to do, and intel has responed to these claims by saying so, there are valid reasons that I could explain, but this post is already long enough. If you want to know, say something.
There is other stuff, but this post is getting long so I''ll close with this:
Choosing to overclock is your own decision, but you need to take into consideration a lot of things. If you overclock and it causes something to go bad you''re stuck paying for it, as overclocking voids the warrenty. However, overclocking in the past has been a way of getting a faster processor out of a cheap chip (don''t expect miracles though...)
Drakonite
[Insert Witty Signature Here]
There really isn''t a "safe" way to overclock.
CPU manufactures will claim that any thing you do at all to increase speed will cause massive damage to your cpu. This isn''t neccisarily true.
The most common overclocking is to increase the FSB speed. Lower end celerons were popular for this, because they run at a 66mhz bus, and could be overclocked on a 100mhz bus.
Unfortunaly, the multiplies are high enough now, that this big of a jump would force the processor too fast. IE. a 400 mhz celeron, put on a 100mhz FSB, would run at 600mhz, this is a BIG jump.
Many brands of mobo''s now have a bios option to adjust the fsb by 1 mhz increases. This allows for smaller increments, but also causes PCI and AGP slots to overclock, which is more likely to cause damage than just overclocking the cpu (from my knowledge atleast)
I''ve heard it said that you can safely overclock by as much as 4mhz without worrying. I can''t verify this though.
Intel has often talked about excelerated electromigration (think that''s the word...) which is basicaly that when current is passed through the chip, the electical pathways can be altered.
There are some facts that are commonly circulated among pro overclockers:
Electromigration on a chip that is overclocked to X, should theoretically be the same as a chip clocked at X by Intel.
Electromigration takes many years, on the order of 5 to 10 i believe, to have an effect. Most people say they will have upgraded by then anyway so it doesn''t matter.
Cpu''s are made in large sheets, and a few from each sheet are effectivly torcher tested to see how fast they can run (better quality run faster), as all processors on the sheet will have approximatly the same quality they will be given a multiplier that is a conservative estimate as to the processors ability. So they should be capable of running faster.
There is a theory that companies like intel take some chips that could easily run at, say 800mhz, and lock the speed lower, say 600 mhz. Why this may seem insane to do, and intel has responed to these claims by saying so, there are valid reasons that I could explain, but this post is already long enough. If you want to know, say something.
There is other stuff, but this post is getting long so I''ll close with this:
Choosing to overclock is your own decision, but you need to take into consideration a lot of things. If you overclock and it causes something to go bad you''re stuck paying for it, as overclocking voids the warrenty. However, overclocking in the past has been a way of getting a faster processor out of a cheap chip (don''t expect miracles though...)
Drakonite
[Insert Witty Signature Here]
I think I''d rather pay the extra few bucks and get the faster one, rather than risk my whole hardware...
What I did was o/c''ed my video card.. I strapped the fan off of an old 486 on my heat sink ( that thing with the lines sticking out of it on top of your cpu ) I got an extra 20 mhz out of my chip, and 15 on my ram. This gave me better frame rates, and if kept cool, I didn''t have an stability issues. The key factor is to o/c in small steps and watch for artifacts. If you get artifacts in your game ( small visual discreptinces that shouldn''t be their ) then step it back a little. Then run your machine doing something graphiclly intensive and make sure it doesn''t get too hot. If you take the proper steps and don''t go crazy it isn''t all that dangerous. Just make sure your hardware doesn''t get too hot.
Temperature is the main culprit in OCing. If you get a good case, with lots of venilation...then you should be able to OC a P3 1Ghz to 1.12 w/o issue.
An important thing to do when overclocking is burn-in. Basically run a highly proc-intensive App/Game (preferably in a non Win9x OS) if the computer locks or crashes after a couple days then step the speed back. If it doesn''t crap out...then you shouldn''t have problems running at that clock speed.
Also, a higher number isn''t always better...if compenents lose sync you can take a dramatic speed hit. Most of the time this happens with older harddrives...if you OC your FSB to high, your HDD can''t keep up...and your computer just crawls. So make sure that you run benchmarks to confirm that you''re actually recieving benifits from OCing.
Epolevne
An important thing to do when overclocking is burn-in. Basically run a highly proc-intensive App/Game (preferably in a non Win9x OS) if the computer locks or crashes after a couple days then step the speed back. If it doesn''t crap out...then you shouldn''t have problems running at that clock speed.
Also, a higher number isn''t always better...if compenents lose sync you can take a dramatic speed hit. Most of the time this happens with older harddrives...if you OC your FSB to high, your HDD can''t keep up...and your computer just crawls. So make sure that you run benchmarks to confirm that you''re actually recieving benifits from OCing.
Epolevne
Another thing to keep in mind are two of Intel''s marketing schemes:
Scheme 1) The Intel Celeron is just a normal Pentium with the math co-processor disabled. The original Celerons could actually have it re-enabled (if you knew what you were doing), causing the chip to function just like a Pentium. Intel has since started actually burning out the co-processor to prevent this. Why do any of this? It saves money by not having to engineer/develop/produce a new main die, which costs millions, and 2, it gives them a chance to sell "a new chip". Moral of the story - Just because Intel says it''s not capable of something doesn''t mean it''s necessarily true.
Scheme 2) They produce chips of certain capabilities, then disable some of these capabilities, such as multipliers. They then enable them with new releases and call them new chips, even though the circuitry is the same. This was with mainly with the older pentiums, though. Most of the new pIII''s and IV''s are multiplied to their max. Moral of the story - Read the documentation to make sure what you''re hearing is correct.
Thank you for your bandwidth.
-- Succinct
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Scheme 1) The Intel Celeron is just a normal Pentium with the math co-processor disabled. The original Celerons could actually have it re-enabled (if you knew what you were doing), causing the chip to function just like a Pentium. Intel has since started actually burning out the co-processor to prevent this. Why do any of this? It saves money by not having to engineer/develop/produce a new main die, which costs millions, and 2, it gives them a chance to sell "a new chip". Moral of the story - Just because Intel says it''s not capable of something doesn''t mean it''s necessarily true.
Scheme 2) They produce chips of certain capabilities, then disable some of these capabilities, such as multipliers. They then enable them with new releases and call them new chips, even though the circuitry is the same. This was with mainly with the older pentiums, though. Most of the new pIII''s and IV''s are multiplied to their max. Moral of the story - Read the documentation to make sure what you''re hearing is correct.
Thank you for your bandwidth.
-- Succinct
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The most important thing when overclocking a processor is to have a really good heatsink and fan, as a faster processor Mhz invariable causes the processor to generate more heat (which is bad 
).
The total speed in Mhz of the processor is calculated by multiplying the FSB speed with the processor multiplier (e.g. 100Mhz FSB * 8.5 multiplier = 850Mhz processor etc.)
Changing the multiplier is generally quite safe on modern processors. If you set it to high the processor will simply become unstable and either it won''t POST at all, or your computer will crash a lot. Lowering the multiplier to the default value will (generally) solve the problem. There is seldom any permanent damage to the processor. Note, that I say *seldom*. Neither me nor anyone I know that has overclocked has ever managed to kill a processor, though I guess it is possible.
Changing the speed of the FSB (Front Side Bus) is not as good an idea, since that will change the PCI and AGP speeds to, and many expansion cards don''t like that (as someone mentioned above). Besides, the FSB is pretty fast to begin with on modern computers.
If you raise the multiplier and/or FSB clock you might want to increase the voltage for the processor to increase the processors stability. Beware though, this can be a bit risky, and might destroy your processor if you set it too high, or at least shorten it''s lifespan considerably.
Generally though, the processors of today are pretty sturdy. Once I even ran my Athlon/Thunderbird without the fan connected (by accident
), it lasted for about 10 seconds, and then it completely hung. The heatsink was literally too hot to touch and I thought that I had fried it for sure, but as soon as it had cooled down it worked like a charm (and has worked ever since), though I probably took away a couple of years of it''s lifespan
I suggest reading as many overclocking articles and stuff that you can find about your processor before trying anything, and remember that you''re overclocking at your own risk...
Check out The Overclocking Guide at Tom''s Hardware Guide.
).The total speed in Mhz of the processor is calculated by multiplying the FSB speed with the processor multiplier (e.g. 100Mhz FSB * 8.5 multiplier = 850Mhz processor etc.)
Changing the multiplier is generally quite safe on modern processors. If you set it to high the processor will simply become unstable and either it won''t POST at all, or your computer will crash a lot. Lowering the multiplier to the default value will (generally) solve the problem. There is seldom any permanent damage to the processor. Note, that I say *seldom*. Neither me nor anyone I know that has overclocked has ever managed to kill a processor, though I guess it is possible.
Changing the speed of the FSB (Front Side Bus) is not as good an idea, since that will change the PCI and AGP speeds to, and many expansion cards don''t like that (as someone mentioned above). Besides, the FSB is pretty fast to begin with on modern computers.
If you raise the multiplier and/or FSB clock you might want to increase the voltage for the processor to increase the processors stability. Beware though, this can be a bit risky, and might destroy your processor if you set it too high, or at least shorten it''s lifespan considerably.
Generally though, the processors of today are pretty sturdy. Once I even ran my Athlon/Thunderbird without the fan connected (by accident
), it lasted for about 10 seconds, and then it completely hung. The heatsink was literally too hot to touch and I thought that I had fried it for sure, but as soon as it had cooled down it worked like a charm (and has worked ever since), though I probably took away a couple of years of it''s lifespan I suggest reading as many overclocking articles and stuff that you can find about your processor before trying anything, and remember that you''re overclocking at your own risk...
Check out The Overclocking Guide at Tom''s Hardware Guide.
Succinct: Math co-processor?
???
The Celeron is a Pentium without the cache (or, with less cache).
~~~~~~~~~~
Martee
???
The Celeron is a Pentium without the cache (or, with less cache).
~~~~~~~~~~
Martee
Author
So Intel is out to make a profit by selling "new chips" that are really older chips with capabilities that now enabled, which were disabled in the original. Makes sense. I''d do it if I were the CEO of Intel.
But I run an AMD Athlon-- do they do any of that schematic crap? And please let''s not turn this into a useless flame war over Intel. I have better things to do than read my posts as they become badger-fests.
And thanx for the link, Dactylos.
And if anyone is having a hard time understanding this techie talk: http://www.howstuffworks.com/ You can read up on this there.
But I run an AMD Athlon-- do they do any of that schematic crap? And please let''s not turn this into a useless flame war over Intel. I have better things to do than read my posts as they become badger-fests.
And thanx for the link, Dactylos.
And if anyone is having a hard time understanding this techie talk: http://www.howstuffworks.com/ You can read up on this there.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
