'Maximum processor state' vs. undervolting

I've recently discovered the possibility of changing the "maximum processor state" in Windows' advanced power management. Could some competent person explain to me the advantages and disadvantages with doing this as opposed to undervolting, when one's objective is a cool and quiet notebook. Thanks

 

Well as far as I know it just limits your clock speed.

Undervolting reduces voltage, which may create data errors just like overclocking. Stability has nothing to do with it-- you could have 100% stability, while the CPU is throwing out computational errors and screwing up your data.

 

I never talked about stability. As said, this is all in regards to achieving a cool and quiet machine. Will it do comparable good? Any comments?

 

Think of maximium Processor state as limiting the number of gears in a car to conserve energy(but it is silly as you are limiting the max speed), but undervolting is modifying the engine so it will need the least amount of fuel to do max speed.

 

It approaches the problem from different sides. Limiting the maximum state means your CPU won't run at full speed, so that when you put your machine under load, it won't get as hot because it won't be running flat out, as it were. Undervolting means that no matter what speed it's running at, it uses less power, and thus won't get as hot either. The difference will be mostly at the extremes; Under full load, an undervolted processor may run hotter than a limited maximum state processor (depending on how limited you make it), due to the fact that it's running faster. Contrariwise, at idle, the undervolted processor will run cooler than the limited maximum state processor, since they'll both be running at the same speed, but the undervolted one will be using less power.

 

I'm just telling you something that is relevant to your question because it's a huge difference between undervolting and limiting max processor state.

Undervolting can create calculation errors. Do not do it without testing extensively with a stress test like Orthos.

 

Undervolting isn't that serious... just do a an Orthos stress test for at least 10 mins after each major increment. Worst consequence is data loss or a BSOD, and that's really only if you haven't properly observed how low your particular processor can safely go (each one has different tolerances) with regards to voltage. Unlike limiting the max processor state, there is zero performance loss to undervolting, think of undervolting as sort of a fuel mixture, you adjust the air intake or carb so that the engine gets just the right amount of gas and none is unecessarily injected. Intel sets a standard voltage for each of their processors to run on, although this voltage is safe for every one they produce, it is unecessary power/heat for most and you make it so that just the right amount of power that is required at each multiplier is going in, rather than unecessary power.

 

10 minutes isn't enough. It's not just stressing the processor, but using different data, so you should let it go through a lot of different calculations. I would test for at least 8 hours.

Losing data is pretty serious. So is corrupting your OS. Once you undervolt too low and see errors in Orthos or get a bluescreen, there's a chance your OS is corrupted and needs to be reinstalled.

 

Probably a very very very small chance then, because it's never happened to me (and I've BSOD'd more times than I care to remember).

Back to the question, undervolting is probably a better option to consider if you want to decrease power consumption and heat, because you don't lose performance. Keep in mind though that you won't be able to overclock as high as you could have at stock voltage.

 

I seriously doubt that. I'm running fine undervolted with my T7700, never had a BSOD as a result of it, and no data loss isn't really that serious unless you physically destroy your HDD, in most cases all can be recovered.
Also i don't see how calculations are related to how much power is being fed... a calculation is a calculation, CPU's use Boolean algebra, "different" calculations should not use more power or somehow require a higher voltage that others. Therefore i do not see the logic behind your thought why we should run 8 hour tests everytime we undervolt a CPU, i'd rather fill that up with 7 hours and 50 minutes of gaming...

 

So, how can you manage to do either? I'm using Win7 if that helps
____________________________
Lenovo 3000 N100
T2600 @ 2.16 GHz
3 GB DDR II 667
GeForce 7300 Go

 

To be perfectly safe, I'd do at least twenty-five minutes. As you stress the card longer, it'll heat up, and as it heats up there's more of a chance for errors to crop up. If you can 100% stress it for twenty-five minutes it'll be a fair indicator that your CPU is perfectly stable. Also: I recommend OCCT Linpack test. If you don't BSOD or get errors after twenty minutes of that, your CPU is a-ok.


As for OS corruption, I've undervolted, overclocked, and gotten more BSODs than you'd care to think about. I used the built-in system file checker and the only error I've gotten was because I've patched the Windows Basic theme.

Now, I don't know if "sfc /scannow" is really reliable, so someone verfiy.

 

You only need chkdsk to check for filesystem consistency.
sfc /scannow is for malware attacks which modified important system files.

 

25 minutes isn't even close to long enough in orthos. 20 runs of Intel Burn Test or LinX are the only true *QUICK* ways of stress testing your CPU for stability. ie, my i5 @ 4GHz is stable at 1.296v for 8 hours of Prime95/24 hours of orthos but will fail Prime95 @12 hours and Intel Burn Test after the 15th run. ITB takes less than an hour to do 20 runs @ 2GB of ram usage.

 

Oh i see, you want to be all official and stress it for hours upon hours just to make sure, i don't see what this accomplishes.. the chances of crashing because of too low voltages are slim if you haven't adjusted the unrealistically.

Edit: can you justify one reason to run a test for a whole bloody day?

 

Unless your work is mission critical and you expect to keep your processor pegged out after undervolting, I really don't see the point.