• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Voltage drop with w/c....

P

plywood99

[H]ard|Gawd
Joined
Dec 25, 2004
Messages
1,666
Since I switched to water cooling, I figured I would try dropping the voltage to my cpu. I was thinking lower temps with w/c maybe means that not so much voltage is needed. So I dropped my volts to 1.8 from 1.9. By the way my system was not stable below 1.9 on air.

Well, it works!! :) :D

Load temp dropped 5 degrees celcius and processor is still stable..

Is this normal? Can voltage be dropped more if cpu is runnning cooler??
 
Markland556 said "well isnt that just kinda making the cpu not work as much since its not going as fast."

I dropped the voltage to the cpu, not the speed at which it runs. I just found it interesting that I can drop the voltage but still maintain current speed. I couldn't do this when using my old Thermaltake Volcano 11. :rolleyes:
 
Hey plywood99 Thanks
I forgot the first rule of racing
"go as slow as you can and still be in the lead at the finish line !"
Same true with Cpu
I got as far as I could go and left my volts at 1.75. Read you Noobe post and said "hey this Noob is on to something"
So I lowered my volts and the best I can do is 1.7 v
Well I guess I can get another five years outta this sucker :D

All hale the Noobe !
Thanks and welcome
 
imagine point A and B in your cpu:

there is a time-limit which electrons have to go from point A to B
(that time limit is 1second / Mhz_of_cpu)
when the electrons dont make it in time.. shit hits the fan... you crash or data gets corrupted ... etc...

the hotter the cpu gets, the further point A and B are from each other, due to expansion and so on.
voltage is a pretty term for "the speed of the electrons"
so when the electrons dont have as far to travel in the cpu, they dont have to go as fast... therefore voltage doesn't have to be as high

same concept for ram, northbridge and everything electronic with a clock cycle
 
james.m.flood said:
imagine point A and B in your cpu:

there is a time-limit which electrons have to go from point A to B
(that time limit is 1second / Mhz_of_cpu)
when the electrons dont make it in time.. shit hits the fan... you crash or data gets corrupted ... etc...

the hotter the cpu gets, the further point A and B are from each other, due to expansion and so on.
voltage is a pretty term for "the speed of the electrons"
so when the electrons dont have as far to travel in the cpu, they dont have to go as fast... therefore voltage doesn't have to be as high

same concept for ram, northbridge and everything electronic with a clock cycle
Click to expand...


What kind of evidence do you have to back up this theory?
Voltage, in relation to electrons, is a driving force for them to move. Electron speed is virtually constant, regardless of voltage.
 
not theory: fact...

the only evidence i have is being in 3rd year physics and comp-sci, and what i'm about to explain: (and i could probably back it up with some text-book quotes but i dont feel like proving myself that badly)

it is true that electron speed is constant and defined.. but this speed is not the speed at which they move from one atom to another..
if you remember from high school chemistry there are 'exited states' and 'valence electrons'
conductivity of metals is basicaly the valence electrons are 'exited', through 'voltage', and 'jump' from atom to atom...
this is such a Crude description but it gets the idea across :)

when in 'rest' (power off) there is very little movement of electrons from atom to atom.. allthough the electrons within the atoms ARE moving at the constant electron speed you mentioned...
when voltage is applied what is basically happening now is there is a large 'ground' (which you can think of as a thing that sucks valence electrons), the sucktion-power of the ground is what we call voltage and the stronger it is, the faster the valence electrons travel towards the grounds source jumping from atom to atom.
(the electrons being sucked away are replaced by ones coming from the other prong of the plug going into your wall... and that's what you pay for every 2 months)
so yes.. electron speed is constant.. but the speed at which they move from atom to atom is WAY slower then the speed at which they revolve around an atoms nucleus...

i hope this helps.. if you dont understand or get it then i can find some text-book references for you
 
so in answer to the original thread:

you can drop the voltage if your cpu is running cooler.. to a certain extent...
it is always dependent on how high your clock cycle is.

this is nice because the lower the voltage, the less heat the cpu generates.. and the lower you can drop it again...

(and no i'm not saying that voltage is the ONLY thing that generates heat :)
 
I'm gonna go ahead and backup James on this one....as I've also proven this FACT myself......with GFX cards. Simply adding a waterblock to a card like my X800XT-PE typically does increase overclockability (if the cards has the extra headroom) at the same voltage....and for me it gave me extra room to work at the same, already raised voltages. Now that my card is sitting at 702Mhz and still at a load temp of 30-40C depending on where in the house i have it.....i'm gonna be looking at a 172W Pelt for it, because I've seen temp drops boost OCs FAR to often not to give it a go this time. ;)

Another case where the lower temps helped with a CPU is my 3.4E. On regular h2O sitting in my room I can't bench at anything higher then 240FSB....for a clock of 4.08GHz, and even afterawhile that gets unstable due to the gradual temperature rise in the water with the XT-PE on the same loop. Moving my comp upstairs and dropping teh CPU temp by 10-15C by placing it near an open window gets me all the way up to 4.3GHz.......now imagine what I'll be able to do with a Vapo running at -40C. :cool:
 
that's pretty sweet man...

i've been toying with the idea of pelts in my system...
but it doesnt make sence because for the same price of setting one up.. i could just buy a faster cpu/mobo :-D
so unless i've actually got the fastest cpu out there.. overclocked as high as it can go.. i dont see myself moving into that realm yet :)

that and i'm very poor... its amazing that i have the system i have now... that was luck
 
james.m.flood said:
that's pretty sweet man...

i've been toying with the idea of pelts in my system...
but it doesnt make sence because for the same price of setting one up.. i could just buy a faster cpu/mobo :-D
so unless i've actually got the fastest cpu out there.. overclocked as high as it can go.. i dont see myself moving into that realm yet :)

that and i'm very poor... its amazing that i have the system i have now... that was luck
Click to expand...

So you decided to put money into watercooling your FX5200 that sees next to no gains from overclocking whatsoever? I used to have one of those "things" as a temp card strapped in with my 2.8C @ 3.5GHz....and even with a very nice overclock there were VERY small gains in FPS in any game I tried, and my 3DMarks barely moved. :p
 
i agree with you entirely

but the reason i have it is not because i wanted to get extra frames or overclock my card at all...
i actually didnt pay any money for it.. it was given to me by a friend who got a lower profile block

and the reason i was very happy to take it is because the card crashes without a fan on it... and the fan is loud....

so this way my system is one fan quieter...

if it had not been given to me i would not have bought it :)
i would have put the money towards a better video card.. or maybe even something crazy like: food ;-)
 
james.m.flood said:
i agree with you entirely

but the reason i have it is not because i wanted to get extra frames or overclock my card at all...
i actually didnt pay any money for it.. it was given to me by a friend who got a lower profile block

and the reason i was very happy to take it is because the card crashes without a fan on it... and the fan is loud....

so this way my system is one fan quieter...

if it had not been given to me i would not have bought it :)
i would have put the money towards a better video card.. or maybe even something crazy like: food ;-)
Click to expand...

Good stuff, that's kinda what I was hoping had gone down. :p
 
cornelious0_0 said:
I'm gonna go ahead and backup James on this one....as I've also proven this FACT myself......with GFX cards. Simply adding a waterblock to a card like my X800XT-PE typically does increase overclockability (if the cards has the extra headroom) at the same voltage....and for me it gave me extra room to work at the same, already raised voltages. Now that my card is sitting at 702Mhz and still at a load temp of 30-40C depending on where in the house i have it.....i'm gonna be looking at a 172W Pelt for it, because I've seen temp drops boost OCs FAR to often not to give it a go this time. ;)

Another case where the lower temps helped with a CPU is my 3.4E. On regular h2O sitting in my room I can't bench at anything higher then 240FSB....for a clock of 4.08GHz, and even afterawhile that gets unstable due to the gradual temperature rise in the water with the XT-PE on the same loop. Moving my comp upstairs and dropping teh CPU temp by 10-15C by placing it near an open window gets me all the way up to 4.3GHz.......now imagine what I'll be able to do with a Vapo running at -40C. :cool:
Click to expand...

Actually, for every 10c drop in cpu load temp, you should be able to increase your speed 2 % - 3%. Say you ran 4ghz @ 50c. At 30c you should be able to run around 4240. This is with no increase in voltage.
 
plywood99 said:
Actually, for every 10c drop in cpu load temp, you should be able to increase your speed 2 % - 3%. Say you ran 4ghz @ 50c. At 30c you should be able to run around 4240. This is with no increase in voltage.
Click to expand...

Not that I do or don't believe you, but do you have any links that can back you up here, just so ppl dont start to think you're talking out your a$$. ;)

EDIT: Just so you know, I'm not really doubting you here, as that's actually REALLY close to the gains I saw when I drop my load temps by 15-20C by nearly having my comp out the front door right now.....about 5%.

If you can provide some proof to back this up then I might think twice about switching to my A64 rig so soon.....as if I do get this Vapo LS at the end of the month I'd be dropping my load temps enough to warrant me hitting about 4.7Ghz.....going by that 10C = 2-3% calculation which so far has held up for me. ;)
 
cornelious0_0 said:
Not that I do or don't believe you, but do you have any links that can back you up here, just so ppl dont start to think you're talking out your a$$.
Click to expand...

It's called Thermic Acceleration. You can read about it on Asetek's website. I was going to link it for you, but there sight is not showing. I get this message, "The page cannot be displayed. There is a problem with the page you are trying to reach and it cannot be displayed."
Don't know what's up with there site.

Anyway, the only time I talk out my a$$ is when i fart.... :D
 
Another good example for how temperature increases overclockability

my 2.4c would do 3.2 on Air @ default volt
the same chip also did 3.6 once it was moved to a wcing system(not mine, i gave the chip away) and once again at stock voltage

both cases were using a p4c800 Deluxe
 
BigTaf said:
Another good example for how temperature increases overclockability

my 2.4c would do 3.2 on Air @ default volt
the same chip also did 3.6 once it was moved to a wcing system(not mine, i gave the chip away) and once again at stock voltage

both cases were using a p4c800 Deluxe
Click to expand...

While i have seen cooling changes make that much of a change.....had you previously tested that your board could DO 300fsb? Or do you know now.....
 
plywood99 said:
It's called Thermic Acceleration. You can read about it on Asetek's website. I was going to link it for you, but there sight is not showing. I get this message, "The page cannot be displayed. There is a problem with the page you are trying to reach and it cannot be displayed."
Don't know what's up with there site.

Anyway, the only time I talk out my a$$ is when i fart.... :D
Click to expand...


As far as I can gather, "thermic acceleration" is just a fancy term that Asetek uses for "overclocking" aided by cpu cooling... No scientific theory behind the term.
 
zer0signal667 said:
As far as I can gather, "thermic acceleration" is just a fancy term that Asetek uses for "overclocking" aided by cpu cooling... No scientific theory behind the term.
Click to expand...

Which is just what I was thinking.....especially if that's something you saw on AseTek's site.....I'm still open to a link with scentific proof behind this, but something from AseTeks mouth alone about how lower your load temp 10C will give you 2% more clock speed is preferred....just so we know we're getting facts. ;)
 
cornelious0_0 said:
Which is just what I was thinking.....especially if that's something you saw on AseTek's site.....I'm still open to a link with scentific proof behind this, but something from AseTeks mouth alone about how lower your load temp 10C will give you 2% more clock speed is preferred....just so we know we're getting facts. ;)
Click to expand...


Hehehe, I was unaware that advertising had anything to do with fact ;)
 
I'm sure you all have heard of superconductivity right? Well here's the definition.
"The flow of electric current without resistance in certain metals, alloys, and ceramics at temperatures near absolute zero, and in some cases at temperatures hundreds of degrees above absolute zero."
Notice that is says basically with lower temperature come less resistance to the flow of electric current. Resistance to electrical flow is the problem, creating heat. But if one decreases temperature, resistance to electric current is less. Now, since resistance is less, current can be increased. So, the more you lower temps, the more current can be applied...aka overclocking.
The statement on Asetek's site 10C decrees in temps equals 2% - 3% increase in speed is true. Cornelious it is not just coincidence that you have experinced this yourself in your overclocking escapades.
Yeh the term "Thermic Acceleration" may be a nice catch phrase Asetek came up with, be the principles behind it are fact.
 
I agree that yes, they did make it sound rather gimicky....but it does draw the eye.

I guess I'll see if this holds true, as it's "only" 3 weeks until I find out if I'm gonna be getting this Vapo LS or not....and if I do then there'll sure be some fun times coming. :D
 
plywood99 said:
I'm sure you all have heard of superconductivity right? Well here's the definition.
"The flow of electric current without resistance in certain metals, alloys, and ceramics at temperatures near absolute zero, and in some cases at temperatures hundreds of degrees above absolute zero."
Notice that is says basically with lower temperature come less resistance to the flow of electric current. Resistance to electrical flow is the problem, creating heat. But if one decreases temperature, resistance to electric current is less. Now, since resistance is less, current can be increased. So, the more you lower temps, the more current can be applied...aka overclocking.
The statement on Asetek's site 10C decrees in temps equals 2% - 3% increase in speed is true. Cornelious it is not just coincidence that you have experinced this yourself in your overclocking escapades.
Yeh the term "Thermic Acceleration" may be a nice catch phrase Asetek came up with, be the principles behind it are fact.
Click to expand...


You cannot compare superconductors to anything used in modern day semiconductors. Superconductors are mainly specially fabricated materials that are insulators at ALL but near absolute zero temps, and under special magnetic conditions. If you know anything about semiconductor and materials properties, you should know that there is a temperature below which semiconductors simply fail to be semiconductors- they become insulators. Even within acceptable temperature ranges, lower temps don't necessarily "allow" you to increase voltage OR current due to lower resistance. The change in conductor resistance per temp. change is very slight, and semiconductors work in very different manners than conductors. Increasing clock speed is not a privelege allowed by lowered resistance.
 
zer0signal667 said:
You cannot compare superconductors to anything used in modern day semiconductors. Superconductors are mainly specially fabricated materials that are insulators at ALL but near absolute zero temps, and under special magnetic conditions. If you know anything about semiconductor and materials properties, you should know that there is a temperature below which semiconductors simply fail to be semiconductors- they become insulators. Even within acceptable temperature ranges, lower temps don't necessarily "allow" you to increase voltage OR current due to lower resistance. The change in conductor resistance per temp. change is very slight, and semiconductors work in very different manners than conductors. Increasing clock speed is not a privelege allowed by lowered resistance.
Click to expand...

I'm not saying cpu's are superconductors, just that some of the principles can be applied. Yes there is a point where extreme cooling no longer benifits modern cpu's, but MOST of the people around here will never experience it...AKA cooling with liquid Nitrogen. Tom's Hardware and several others have went well over 5ghz and Team Akiba hit 6ghz using extreme cooling. http://www.xtremesystems.org/forums/showthread.php?t=42655&page=1&pp=25
How did they do this? There is less heat affecting the resistance of the flow of electricity in the processor.

Say your running a 15amp miter saw off a 100' extension cord made from 18gage wire. After a while the wire gets hot and may even melt the plastic jacket around it, causing a short, or sometimes a fire. Why does this happen? Because of the amperage being drawn through the small wire, too much electrical flow. Now, if you take that wire and use it in the winter on a nice cold day, there is less chance of the wire shorting out because of the cold dissipating the heat, that is to say removing resistance to the flow of electricity.

I'm sure almost every person on this forum who overclocks has experienced greater success keeping their processor at a colder temp.

Once again, I did not say processors are superconductors, but the principles CAN be applied to them.
 
cornelious0_0 said:
Isn't it fun trying to explain electronics? :rolleyes: :p
Click to expand...
HaHa, Yeah it's a blast, a real knee slapper.... :(

Try to help out and the mafia comes down on you. :D
 
plywood99 said:
Now, if you take that wire and use it in the winter on a nice cold day, there is less chance of the wire shorting out because of the cold dissipating the heat, that is to say removing resistance to the flow of electricity.
Click to expand...
:) You're Right!! but for the wrong reasons.. but YOur Right! :)

**i bet ya anything HeThatKnows is goina make me calculate this!**
 
Since I'm not learned enough in this area I'm just gonna sit back and listen....but the end result of this discussion should be interesting.
 
james.m.flood said:
**i bet ya anything HeThatKnows is goina make me calculate this!**
Click to expand...
LOL! Even I'm not that cruel. Not that you wouldn't get extra credit for a paragraph or two on the 'mean free path' of conductance-band electrons... :)
 
HeThatKnows said:
LOL! Even I'm not that cruel. Not that you wouldn't get extra credit for a paragraph or two on the 'mean free path' of conductance-band electrons... :)
Click to expand...

You just lost me.....again. :rolleyes:
 
HeThatKnows said:
LOL! Even I'm not that cruel. Not that you wouldn't get extra credit for a paragraph or two on the 'mean free path' of conductance-band electrons... :)
Click to expand...
I'll make you a deal, i'll do my best, but you gimme some [H]ardware in exchange...
enough of this poor student labour for free ;)
 
HeThatKnows said:
LOL! Even I'm not that cruel. Not that you wouldn't get extra credit for a paragraph or two on the 'mean free path' of conductance-band electrons... :)
Click to expand...


Oooh, does that go for me too?? :D
 
plywood99 said:
I'm not saying cpu's are superconductors, just that some of the principles can be applied. Yes there is a point where extreme cooling no longer benifits modern cpu's, but MOST of the people around here will never experience it...AKA cooling with liquid Nitrogen. Tom's Hardware and several others have went well over 5ghz and Team Akiba hit 6ghz using extreme cooling. http://www.xtremesystems.org/forums/showthread.php?t=42655&page=1&pp=25
How did they do this? There is less heat affecting the resistance of the flow of electricity in the processor.

Say your running a 15amp miter saw off a 100' extension cord made from 18gage wire. After a while the wire gets hot and may even melt the plastic jacket around it, causing a short, or sometimes a fire. Why does this happen? Because of the amperage being drawn through the small wire, too much electrical flow. Now, if you take that wire and use it in the winter on a nice cold day, there is less chance of the wire shorting out because of the cold dissipating the heat, that is to say removing resistance to the flow of electricity.

I'm sure almost every person on this forum who overclocks has experienced greater success keeping their processor at a colder temp.

Once again, I did not say processors are superconductors, but the principles CAN be applied to them.
Click to expand...


Let's get back to what should be the root of this conversation. What do you suggest is the cause for overclocking failure? Don't tell me heat, tell me what specific failure in some part of the processor is caused by heat. Don't tell me "too much speed", tell me why the processor can't handle signal the frequency. If we determine this, then we can more easily see what effects heat has on the whole scenario.
 
james.m.flood said:
:) You're Right!! but for the wrong reasons.. but YOur Right! :)

**i bet ya anything HeThatKnows is goina make me calculate this!**
Click to expand...
James, I've seen many of your posts in this forum and you seem to be pretty edumacated. :D Could you PLEASE enlighten us to the true reason why colder temps can help increase cpu speed? :confused:
 
zer0signal667 said:
Let's get back to what should be the root of this conversation. What do you suggest is the cause for overclocking failure? Don't tell me heat, tell me what specific failure in some part of the processor is caused by heat. Don't tell me "too much speed", tell me why the processor can't handle signal the frequency. If we determine this, then we can more easily see what effects heat has on the whole scenario.
Click to expand...
Zero, I'm a carpenter, as you can tell from my tag-name, not a bleeding rocket scientist. Ask me something about wood, miter joints or crown molding angles and I'll talk till I'm blue in the face.

Based off what I've experienced fooling with computers, heat seems to be and I believe is ONE of the factors limiting speed increase in cpu's. I never said heat is the end-all be-all reason for limiting processor clocks. If it was then we would see speeds well in excess of what Team Akiba achieved.

I'ved read somewhere about electrons jumping their "paths" inside cpu's causing stability problems and plenty of other things I've forgot. I believe I also heard that cpu achitecture plays an important role in limiting processor speed, as in the AthlonXP. Nevertheless in my experience heat does seem to be one of the problems.
 
zer0signal667 said:
Oooh, does that go for me too?? :D
Click to expand...
You're not in school, so extra credit doesn't apply. But I could give you a cookie! :p

The two big things affected by temperature are: resistance of the interconnects (with related transmission line effects), and transistor switching speed.

Without getting into all the nasty details, the transistors in your CPU work by allowing or excluding charge carriers (electrons and 'holes') from a semiconductor channel that separates the 'source' and the 'drain'. You do this by adding/removing electrical charge from the 'gate' electrode, which sets up an electrical field that repels the charge carriers (so that electricity can't flow, nothing to carry it) or allows them to pass.

Now, we like to think of electricy as moving really fast, and the electrical fields do. But the charge carriers themselves move rather slow. In fact their movement due to the electrical field (their 'drift velocity') is much slower than their random movement from thermal energy. So when we want to shut off a transistor, the charge carriers are pushed out of the source to drain channel at the drift velocity, but are also zipping in-and-out due to thermal energy. The less thermal movement, the faster the channel can be sufficiently cleared and the faster a valid output is available.

Of course, before the transistor does anything, we need to get the gate charged (or discharged, either way).The higher the resistance of the interconnect wires, the long it takes for a useful amount of charge to come or go. Lower the resistance (by cooling) or raise the voltage, and the gate charges quicker.
 
You must log in or register to reply here.
Back
Top