Why do you overclock?

[biostud]

I know this is an old test, but it shows that overclocking can be more efficient than stock.

http://www.tomshardware.com/reviews/core-i5-750-efficiency,2500-8.html

As you can see the overclocked 3.36/3.84 Ghz CPU is more efficient than the 2.8/3.2. The Watt-hours needed to complete a task is lower with the overclocked chip. And actually it's first when you reach 4.0+ Ghz the efficiency is lower than stock.

 

[tolis626]

Your logic is brilliant... If you don't modify the voltage and optimize the overclock then everything is fine... The fallacy in that argument, however, is that core frequency is voltage dependant and lower frequencies use lower voltages. Even a stock frequency, the chip can be undervolted so making the argument that voltage doesn't change with OC is really grasping at straws...

I can under-volt my 4770K to .95V and run it at 3.9GHz stable, the chip requires incremental voltage increases to reach higher frequencies. My 24/7 configuration is 4.6GHz at 1.196V. 4.0, 4.1, 4.2, 4.3, 4.4, and 4.5 all require unique voltages to run stable. Yes, I can leave the chip at stock voltages and claim that I am receiving "free performance" since I am able to clock 300-400MHz higher without increasing voltages. However, this is disingenuous as any individual who is overclocking should also be trying to obtain the lowest voltage for any specific frequency and, as such, should also explore under-volting the chip to determine the lowest possible voltage for stock performance. Only then can you analyze the difference in power consumption at different frequencies of a given chip.

Even the referenced article discussed this:

I'm coming from an overclocking background where perf/W is the biggest factor ; smartphones.

Currently I'm using a Galaxy S III.At stock,it ran at 1.4GHz with 1.275/1.3mV,depending on the binning of the chip.Mine was 1.3mV,and is now running at 1.6GHz with 1.225mV.The result?A nice bump in performance with actually LESS wattage drawn than at stock.I can easily do 1.7GHz but it exceeds stock voltage and destroys battery life.

An even better example was the Galaxy S 2 with its initial overheating problems.It ran at 1.2GHz with a rather high voltage (I think 1.3mV again).And believe me when I say that at times it would almost burn my finger when I touched the area around the CPU.After spending weeks tinkering with it,I had finely tuned voltages for each frequency and modified frequency scaling.The result was a phone running at 1.5/1.6GHz (I always tried to stay at the more sane 1.5GHz but again went back to 1.6GHz after a few days) with no overheating problems.

My point?If you really take the time to find the sweet spot for your particular CPU,you can make it work wonders.Having the optimal voltage for each frequency will help clock higher and consume less energy,which means lower temperatures too.Don't stick to the "+1GHz with +0.5V" methodology.Do it right or don't do it at all.

 

[zir_blazer]

As for the statement that 10 years ago it was easy to overclock, and now its much more complicated, the motherboards and BIOS are now a lot more complicated, so , yes, it may be more difficult. But what about the "easy OC " option of many motherboards nowadays ?You don't get a MAX OC, but its an easy one (if its stable)

When I did said that overclocking 10 years ago was easier? I never said that, I said it was cheaper, not easier. Actually, it was MUCH harder that currently is, specially since the UEFI GUIs takeover. The difference was, that if you know what parts to buy, you could overclock for cheap, something that right now you can't due to the fact that now they're charging you for it as an added feature on the top mainstream parts. And for people without deep wallets (Like during an economic crisis), budget overclocking could make hell of a difference.
As I stated in what you say is a "troll post", you could purchase a low end Processor, spend the difference on a higher quality Motherboard that allowed you to touch FSB, Multiplier and VCore (Value Motherboards at that era usually didn't, and some had crappy SiS Chipsets. And Chipset's Memory Controllers made a noticeable difference in performance, remember? And we had to learn to avoid PC Chips, too), with that, you could get more performance overclocking a low end Processor than if you had purchased a mainstream Processor with a mediocre Motherboard which didn't allowed you to overclock.

I'm against the AI overclocking features. Most of them overclock with unsafe overvolting without even telling the user - and for that reason they're even worse from a power efficiency perspective. Oh, and you're forgetting the fact that you STILL need a more expensive K series Processor with a Z series Chipset to use that.
My whole rant is about how overclocking was a life changer for budget users, I don't care about the ones that can afford mainstream or higher parts, so you're missing the point.


Also, I have a better question for you:

As for unnoticeable performance, my dual 5639 right won't even get the bonus if not overclocked. so 7k vs 160k ppd is not noticeable ? that post is a troll post.

Mind you tell me what the hell you use those Dual Xeons for that defies common sense? The default Frequency of your Xeons is 2.13 GHz, you have them at 2.8 GHz, that's a 31% overclock. How on the earth you're getting 23 times the performance with a 31% overclock? At the very best scenario, which would be a totally CPU bound one, you could get a 31% performance boost assuming lineal performance increase, through chances are that it is a bit less due to diminishing returns. So your 7K vs 160K numbers are not realistic to me.

I know this is an old test, but it shows that overclocking can be more efficient than stock.

http://www.tomshardware.com/reviews/core-i5-750-efficiency,2500-8.html

As you can see the overclocked 3.36/3.84 Ghz CPU is more efficient than the 2.8/3.2. The Watt-hours needed to complete a task is lower with the overclocked chip. And actually it's first when you reach 4.0+ Ghz the efficiency is lower than stock.

Two things:

1 - Comparing default Frequency vs overclocked at the same Voltage (Check settings) is unfair. The increase in power consumption due to the extra Voltage is already accounted for in that scenario, you get just the lineal part of the extra Frequency as added power consumption. You should compare the lowest possible Voltage at default Frequency against the higher possible Frequency at default Voltage to make it fair.
2 - The reason why overclocking seems more power efficient is because all the other components of the computer are also on and using power, so the increase of the Processor's power consumption is less notorious. The Processor itself is ALWAYS less efficient, is just that the entire system can get to Idle earlier. You can say that the entire system is more power efficient, the Processor as a standalone device is not.

 

[biostud]

When I did said that overclocking 10 years ago was easier? I never said that, I said it was cheaper, not easier. Actually, it was MUCH harder that currently is, specially since the UEFI GUIs takeover. The difference was, that if you know what parts to buy, you could overclock for cheap, something that right now you can't due to the fact that now they're charging you for it as an added feature on the top mainstream parts. And for people without deep wallets (Like during an economic crisis), budget overclocking could make hell of a difference.
As I stated in what you say is a "troll post", you could purchase a low end Processor, spend the difference on a higher quality Motherboard that allowed you to touch FSB, Multiplier and VCore (Value Motherboards at that era usually didn't, and some had crappy SiS Chipsets. And Chipset's Memory Controllers made a noticeable difference in performance, remember? And we had to learn to avoid PC Chips, too), with that, you could get more performance overclocking a low end Processor than if you had purchased a mainstream Processor with a mediocre Motherboard which didn't allowed you to overclock.

I'm against the AI overclocking features. Most of them overclock with unsafe overvolting without even telling the user - and for that reason they're even worse from a power efficiency perspective. Oh, and you're forgetting the fact that you STILL need a more expensive K series Processor with a Z series Chipset to use that.
My whole rant is about how overclocking was a life changer for budget users, I don't care about the ones that can afford mainstream or higher parts, so you're missing the point.


Also, I have a better question for you:



Mind you tell me what the hell you use those Dual Xeons for that defies common sense? The default Frequency of your Xeons is 2.13 GHz, you have them at 2.8 GHz, that's a 31% overclock. How on the earth you're getting 23 times the performance with a 31% overclock? At the very best scenario, which would be a totally CPU bound one, you could get a 31% performance boost assuming lineal performance increase, through chances are that it is a bit less due to diminishing returns. So your 7K vs 160K numbers are not realistic to me.



Two things:

1 - Comparing default Frequency vs overclocked at the same Voltage (Check settings) is unfair. The increase in power consumption due to the extra Voltage is already accounted for in that scenario, you get just the lineal part of the extra Frequency as added power consumption. You should compare the lowest possible Voltage at default Frequency against the higher possible Frequency at default Voltage to make it fair.
2 - The reason why overclocking seems more power efficient is because all the other components of the computer are also on and using power, so the increase of the Processor's power consumption is less notorious. The Processor itself is ALWAYS less efficient, is just that the entire system can get to Idle earlier. You can say that the entire system is more power efficient, the Processor as a standalone device is not.

Yes, the CPU in itself is note more efficient, but if you want to calculate efficiency and wattage you need to use the whole system, as it is very difficult to run the CPU without the rest of the computer.

So If you do a lot of rendering and you can turn off your computer sooner and therefore save money.

Underclocking and tweaking voltage is IMO also a form for "overclocking", just with another goal than reaching highest frequency. So to me the question is stock vs every kind of tweaking.

So it's still possible to get a more efficient system vs a stock system by doing some over clocking.

http://www.tomshardware.com/reviews/core-i5-3570-low-power,3204-14.html

Same here where they compare efficiency runs of different intel 37xx processors. So while the computer uses more power under load it completes the task faster on the faster CPU's.

 

[TerryMathews]

How on the earth you're getting 23 times the performance with a 31% overclock? At the very best scenario, which would be a totally CPU bound one, you could get a 31% performance boost assuming lineal performance increase, through chances are that it is a bit less due to diminishing returns. So your 7K vs 160K numbers are not realistic to me.

Its fine that you dont know how DC works, but you look like a fool for calling mark out when its your lack of knowledge. Go look up bonus PPD and you'll understand. Hint: you have to be producing a minimum amount of work to qualify.

2 - The reason why overclocking seems more power efficient is because all the other components of the computer are also on and using power, so the increase of the Processor's power consumption is less notorious. The Processor itself is ALWAYS less efficient, is just that the entire system can get to Idle earlier. You can say that the entire system is more power efficient, the Processor as a standalone device is not.

Mathematically false. It is extremely unlikely nee impossible that in this complex system the processor is already operating at the most efficient speeds for the specific workload in question. If power used is the area under the curve of current draw/time, manipulating idle clock and turbo clock and the amount of time spent in each state can have a dramatic effect of the area under the curve.

But again, who cares? 99.99% of people just rely on the automatic power controls which still function normally if you are overclocking.

 

[Markfw]

Not sure what you are trying to get at here... You really are trying too hard to drive home this fallacy of "free performance"

If you want to continue to propagate this myth, at least provide some substantiated evidence to support your position...

Its not a fallacy, and I already provided an example. Mild OC, with no vcore increase=free performance. There is another poster here that is overclocked with sub-standard vcore, so he is saving money AND getting a free performance boost (not typical, grant you), so other posters have already backed me up here.

Knock off your crap, like the words fallacy, you are trolling, and I will have another mod infract you if you keep it up. (not allowed to infract in a thread I have posted in)

TO: zir blazer
As for the 23 time performance, I never said that. I said PPD. If it takes you 34 minutes per step on an 8101 unit, you get the 200k (approx) point bonus for finishing on time. If it takes you 35 minutes, you don't get the bonus. 20k vs 220k is a big point difference. an 8104 unit (I am doing right now) is 17k base credit. The bonus credit is 177k. I am doing a step in 16 minutes. If I took 26 minutes, I would not get the bonus. The better example are 8101 units, as they are tight. I have 24 threads, so it does OK. My non-overclocked dual E5570's@2.93 ghz miss the bonus time for 8101's since they only have 16 threads. Right now my dual Opteron 6234's (24 cores) are doing an 8101@31 minutes per frame and will get 215k credit.

You guys need to give up vs the experts here. OCing CAN pay off big and and other times be free.

And as for DC, look here at the stats, this is why I need so much hardware running 24/7@100%
http://folding.extremeoverclocking.com/user_list.php?s=&t=198

And by the way, that's a donation of computer time to research curing cancer.....

 

[Slomo4shO]

I know this is an old test, but it shows that overclocking can be more efficient than stock.

http://www.tomshardware.com/reviews/core-i5-750-efficiency,2500-8.html

As you can see the overclocked 3.36/3.84 Ghz CPU is more efficient than the 2.8/3.2. The Watt-hours needed to complete a task is lower with the overclocked chip. And actually it's first when you reach 4.0+ Ghz the efficiency is lower than stock.

Not sure what this proves since all the relevant speeds are at stock voltages...

our Core i5-750 did not operate higher than about 3.7 GHz without a voltage increase. As a result, we found that a 160 MHz base clock was the maximum stable setting we could use with Turbo Boost enabled and without increasing voltage.

Considering that the lower frequencies could have been run at lower voltages than the stock voltage the above graph only illustrates the illusion of efficiency considering that stock voltages are not optimized for the stock frequency at a chip specific level. The graph does show that efficiency does decrease as the chip is pushed past 3.8GHz since these are the only settings where voltage was actually modified...

I'm coming from an overclocking background where perf/W is the biggest factor ; smartphones.

Currently I'm using a Galaxy S III.At stock,it ran at 1.4GHz with 1.275/1.3mV,depending on the binning of the chip.Mine was 1.3mV,and is now running at 1.6GHz with 1.225mV.The result?A nice bump in performance with actually LESS wattage drawn than at stock.I can easily do 1.7GHz but it exceeds stock voltage and destroys battery life.

An even better example was the Galaxy S 2 with its initial overheating problems.It ran at 1.2GHz with a rather high voltage (I think 1.3mV again).And believe me when I say that at times it would almost burn my finger when I touched the area around the CPU.After spending weeks tinkering with it,I had finely tuned voltages for each frequency and modified frequency scaling.The result was a phone running at 1.5/1.6GHz (I always tried to stay at the more sane 1.5GHz but again went back to 1.6GHz after a few days) with no overheating problems.

My point?If you really take the time to find the sweet spot for your particular CPU,you can make it work wonders.Having the optimal voltage for each frequency will help clock higher and consume less energy,which means lower temperatures too.Don't stick to the "+1GHz with +0.5V" methodology.Do it right or don't do it at all.

I am unsure whether you are trying to refute or support my argument... Stock voltages are not optimal and there is always efficiency to be gained by lowering the stock voltage to determine the best optimal voltage at a specific frequency. And yes, it is typically possible to lower the stock voltage and still obtain an overclock. This, however, doesn't contradict the fact that performance per watt drops as frequency increases.

 

[TerryMathews]

(Bah, I wouldn't call AM3+ current)

And no, I'm not trolling, regardless of what you think. I'm comparing how overclocking made a big difference to budget users that know what they were doing in everyday task a decade ago or so, compared to now where budget users can't even do it, and on mainstream is makes less sense due to the fact that you have a faster part to begin with.

That comment right there shows you are either trolling or being intellectually dishonest. Also, AMD has a number of overclockable chips on the FM2+ platform. Do they not count either?

 

[Slomo4shO]

Its not a fallacy, and I already provided an example. Mild OC, with no vcore increase=free performance.

Knock off your crap, like the words fallacy, you are trolling, and I will have another mod infract you if you keep it up. (not allowed to infract in a thread I have posted in)

And I have shown that the mild OC without vcore increase is not a free performance since the stock frequency can be obtained at a much lower voltage than the typically inflated stock voltage. Again, this idea of "free performance" is a fallacy.

So now I am being labeled a troll because you can't formulate a constructive rebuttal? I guess if you can't win the argument through debate then by all means bully your way out of it...

As for the 23 time performance, I never said that. I said PPD.

Our discussion started like this:

t's fun and hard to turn down a free performance boost.

This.

It is fun but there is no such thing as "free performance".

Yes, there is... Almost no matter the setup, OC to the max, then take half that for a stable "free" overclock.

Our discussions were never about PPD, overclocking obviously increases PPD when just discussing just the chip costs.

Also, I don't fold so I can't commend on how the reward system works for folding...

 

[TerryMathews]

And I have shown that the mild OC without vcore increase is not a free performance since the stock frequency can be obtained at a much lower voltage than the typically inflated stock voltage. Again, this idea of "free performance" is a fallacy.

So now I am being labeled a troll because you can't formulate a constructive rebuttal? I guess if you can't win the argument through debate then by all means bully your way out of it...

So now we are to "overclocking is inefficient because you can undervolt further at stock."

Can you please let me know when you're done moving the goalposts? Kthx

 

[Slomo4shO]

So now we are to "overclocking is inefficient because you can undervolt further at stock."

Can you please let me know when you're done moving the goalposts? Kthx

When was the goal post moved? My first reply to Mark included a link to an article from Anandtech which discussed both undervolting and overclocking...

In case you missed it:

Spoiler

So you don't pay for the electricity you use?

Performance per watt decreases as you overclock, there is always a cost. No such thing as a free lunch.

 

[biostud]

Not sure what this proves since all the relevant speeds are at stock voltages...

Considering that the lower frequencies could have been run at lower voltages than the stock voltage the above graph only illustrates the illusion of efficiency considering that stock voltages are not optimized for the stock frequency at a chip specific level. The graph does show that efficiency does decrease as the chip is pushed past 3.8GHz since these are the only settings where voltage was actually modified...



I am unsure whether you are trying to refute or support my argument... Stock voltages are not optimal and there is always efficiency to be gained by lowering the stock voltage to determine the best optimal voltage at a specific frequency. And yes, it is typically possible to lower the stock voltage and still obtain an overclock. This, however, doesn't contradict the fact that performance per watt drops as frequency increases.

Who says you need to change voltages for overclocking? As I posted in an earlier post for me I interpret "overclocking" as all kind of tweaking whether it's change of frequency or voltage, both in positive and negative direction. You could split it into over/under clocking and over/under volting, but for me it's all the "same", as it is the same you do, just with different goals. And I like every type of tweak equally, I just think running @ stock is boring.

 

[Slomo4shO]

Who says you need to change voltages for overclocking? As I posted in an earlier post for me I interpret "overclocking" as all kind of tweaking whether it's change of frequency or voltage, both in positive and negative direction. You could split it into over/under clocking and over/under volting, but for me it's all the "same", as it is the same you do, just with different goals. And I like every type of tweak equally, I just think running @ stock is boring.

Fair enough Sir, I was discussing the merits of the efficiency claims in the graph and wasn't concerned with the semantics of overclocking :thumbsup:

 

[biostud]

When they increase the frequency from 3.3 to 4.7 @ 1.25V (~42 percent) power draw rises from ~160W to ~185W (15.6 percent) so you get 40 percent more speed for 15 percent more power consumption.

 

[Slomo4shO]

When they increase the frequency from 3.3 to 4.7 @ 1.25V (~42 percent) power draw rises from ~160W to ~185W (15.6 percent) so you get 40 percent more speed for 15 percent more power consumption.

At stock voltages. You could undervolt the chip at 3.3 to obtain greater efficiency.

3.7GHz was the maximum at stock voltage, see how there is exponential growth in power consumption once voltage is added at 3.8GHz and above? The linear progression between 3.2 and 3.6GHz would not have been linear if the stock voltages were changes to optimal voltages at said frequency.

 

[Markfw]

any change in stock vcore should be out of the question in the free performance argument. If you DON'T overclock, you won't touch the stock setting, period, as it could cause instability.

If you do overclock, and don't touch vcore, then its free performance, if you do up vcore, then its simply more performance.

You just keep trying to twist the facts to support you position. And so far every poster other than you recently has disagreed with you, and provided facts. You just keep trying to twist the facts to make your argument work, and its annoying.

 

[biostud]

Fair enough Sir, I was discussing the merits of the efficiency claims in the graph and wasn't concerned with the semantics of overclocking :thumbsup:

I was just thinking of the OP's argument not to o/c

I actually have more reasons to not OC:

1. More reliable computing experience, less BSOD, etc
2. More green to environment, less power
3. No need to tinker with BIOS, settings, etc.
4. Not much benefit for more than 90% of computer usage which is browsing, youtube, email, etc.

When you undervolt/underclock you run into the same problems as overclocking (except for heat) regarding to stability and hassle. Which for me is equally good as I like the hassle.

 

[TerryMathews]

You just keep trying to twist the facts to make your argument work, and its annoying.

Still wishing this was bolded...

FINISH HIM!

 

[Slomo4shO]

any change in stock vcore should be out of the question in the free performance argument. If you DON'T overclock, you won't touch the stock setting, period, as it could cause instability.

If you do overclock, and don't touch vcore, then its free performance, if you do up vcore, then its simply more performance.

You just keep trying to twist the facts to support you position.

Please do advise which facts were twisted to support the argument? Why should changes in stock vcore be out of the question? If you run at stock settings and never touched the bios settings of your rig then you would obviously be running at less that ideal efficiency. Most people are content with this.


However, if you are going to argue that overclocking poses the capacity to provide "free performance" then you must have an honest discussion and consider that the individual who is willing to increase the frequency also has the capacity to change the voltages.

And so far every poster other than you recently has disagreed with you, and provided facts.

You mean you and Terry?

 

[biostud]

Please do advise which facts were twisted to support the argument? Why should changes in stock vcore be out of the question? If you run at stock settings and never touched the bios settings of your rig then you would obviously be running at less that ideal efficiency. Most people are content with this.

However, if you are going to argue that overclocking poses the capacity to provide "free performance" then you must have an honest discussion and consider that the individual who is willing to increase the frequency also has the capacity to change the voltages.

It's free because a stock system runs at a less than ideal setting and you can change that by either undervolting or overclocking.

And you could probably also undervolt a processor slightly and do mild overclock. In that case you would get a faster processor using less power = free performance.

You want to compare an optimally voltage tuned processor at stock to a perfectly tuned overclocked processor.

The rest of us compare a processor running at stock voltages and frequency with an overclocked processor.

 

[Markfw]

Please do advise which facts were twisted to support the argument? Why should changes in stock vcore be out of the question? If you run at stock settings and never touched the bios settings of your rig then you would obviously be running at less that ideal efficiency. Most people are content with this.

Exactly, Intel has not configured every chip to run at maximum efficiency. so stock is less than perfect, so an overclocker gets the free performance that Intel did not give them.

HENCE FREE PERFORMANCE, YOU JUST PROVED MY POINT.

definition: free is more than what Intel gave you (or AMD, sorry) for no extra cost.

As for who disagrees with you, its at least, me, Terry and biostud. Nobody else has posted recently.

 

[zir_blazer]

Its fine that you dont know how DC works, but you look like a fool for calling mark out when its your lack of knowledge. Go look up bonus PPD and you'll understand. Hint: you have to be producing a minimum amount of work to qualify.

Then his usage is a special case scenario due to that artficially imposed limit to qualify, the rest of the world that deals with real life applications gets diminishing returns the higher you go.
For someone that called me a troll for saying than the extra performance when overclocking a mainstream part isn't notorious, it still is like that for the vast majority of people, that 99.99% than you say.

Mathematically false. It is extremely unlikely nee impossible that in this complex system the processor is already operating at the most efficient speeds for the specific workload in question. If power used is the area under the curve of current draw/time, manipulating idle clock and turbo clock and the amount of time spent in each state can have a dramatic effect of the area under the curve.

I can't comment on turbo clocks, but for a sustained, continuous workload, it is.

The rest of the system is worth 130W, so the Processor itself consumes:

2 GHz @ 0.820V = 22W (Take this as baseline)
3 GHz @ 0.972V = 47W (110% more power consumption for 50% more Frequency than previous)
4 GHz @ 1.165V = 96W (100% more power consumption for 25% more Frequency than previous)
5 GHz @ 1.488V = 224W (133% more power consumption for 20% more Frequency than previous)

The higher you go, the less efficient it is on the same workload. The issue here is the rest of the system power consumption.

If I could assume that a given task (For example, a render) takes 24 hours at 2 GHz, and it scales perfectly, it should take 16 hours @ 3 GHz, 12 hours @ 4 GHz, and 9h 36m @ 5 GHz. The Processor itself would consume:

24 hours * 22W @ 2 GHz = 528 W-h
16 hours * 47W @ 3 GHz = 752 W-h
12 hours * 96W @ 4 GHz = 1152 W-h
9.6 hours * 224W @ 5 GHz = 2150.4 W-h

If I were to take the static 130W for having the rest of the system turned on, I would have to add it, too...

24 hours * 130W + 528 W-h = 3648 W-h
16 hours * 130W + 752 W-h = 2832 W-h
12 hours * 130W + 1152 W-h = 2712 W-h
9.6 hours * 130W + 2150.4 W-h = 3398.4 W-h

4 GHz is the most efficient with the entire system taken care of, with 3 GHz coming quite close. However, the rest of the system power consumption is actually very high at 130W. If it were lower, like 50W (Realistic without discrete GPU), 3 GHz would win. The lower the rest of the system power consumption, the more efficient it is if the Processor has to work even for longer period of time.
In my case, as I have my computer on 24/7, running at lower Frequencies is more efficient if I can let it do something during the night because the rest of the system power consumption is always wasted anyways.

That comment right there shows you are either trolling or being intellectually dishonest. Also, AMD has a number of overclockable chips on the FM2+ platform. Do they not count either?

I saw relatively few low end APUs, none in enthusiast hands. Most want gaming performance, and end up with something like a Celeron/Pentium + discrete Video Card instead of trying to get the most out of an APU.

Exactly, Intel has not configured every chip to run at maximum efficiency. so stock is less than perfect, so an overclocker gets the free performance that Intel did not give them.

HENCE FREE PERFORMANCE, YOU JUST PROVED MY POINT.

definition: free is more than what Intel gave you (or AMD, sorry) for no extra cost.

As for who disagrees with you, its at least, me, Terry and biostud. Nobody else has posted recently.

You can take two paths: Highest possible overclock at stock Voltage (More performance for slighty higher power consumption, as the default Voltage is already higher than needed), or lowest possible Voltage at stock Frequency. Comparing overclocked vs stock is unfair, I said that earlier.

By that Thread results, things are like this:

Stock: 3.4 GHz @ 1.164V = 84W
Lowest Voltage: 3.4 GHz @ 1.038V = 65W
Highest Frequency: 4.2 GHz @ 1.163V = 99W

It is NOT FREE. Is simply that you're looking at it unidimensionally. The actual jump is 34W, a whole 50%. And considering that its pretty much the sweet spot in power efficiency if you take into account system power consumption, I would prefer to run at 3.4 GHz @ 1.038V and put the Fan at minimum to reduce noise.

 

[Slomo4shO]

Exactly, Intel has not configured every chip to run at maximum efficiency. so stock is less than perfect, so an overclocker gets the free performance that Intel did not give them.

HENCE FREE PERFORMANCE, YOU JUST PROVED MY POINT.

definition: free is more than what Intel gave you (or AMD, sorry) for no extra cost.

As for who disagrees with you, its at least, me, Terry and biostud. Nobody else has posted recently.

Your definition of free appears to differ from mine.

From an economics stance and utilization of opportunity costs, If I were overclocking and discovered that my 4770K can run at 1V at stock frequencies or can be bumped up to 4.3GHz at the stock voltage of 1.125V (arbitrary figures only) but the ~10% performance increase yields an increase in energy consumption greater than 10% then I can conclude that there is a cost associated with the performance gain. Therefore, the performance gain is not "free"

This sums up my stance:

I guess this is what I get for discussing economics with engineers :whiste:

Definitions:

free
adjective, fre·er, fre·est.
36.without cost, payment, or charge.


opportunity cost
noun Economics
the money or other benefits lost when pursuing a particular course of action instead of a mutually-exclusive alternative

 

[Markfw]

Since you 2 continue to ignore my comments, and change the rules, I am done with this thread...................No sense in arguing with someone with a closed mind.

 

[tolis626]

I am unsure whether you are trying to refute or support my argument... Stock voltages are not optimal and there is always efficiency to be gained by lowering the stock voltage to determine the best optimal voltage at a specific frequency. And yes, it is typically possible to lower the stock voltage and still obtain an overclock. This, however, doesn't contradict the fact that performance per watt drops as frequency increases.

I didn't try to contradict you nor agree with you.I merely stated that quick overclocks usually don't work.If you take your time with it and find the optimal voltage for each frequency step (The part on which I agree with you),you will either achive maximum efficiency or at least mitigate the higher power draw because of the overclocking.Unfortunately,there is no 1+1=2 scenario here.Anyone claiming so must think twice.It all comes down to making the decision between higher performance or lower consumption than stock.Or staying at stock,for that fact.