Any point to speedstep when overclocking?

[Maximilian]

I mean the cpu keeps the same voltage anyway no matter what the mhz is according to cpu-z so its using the same amount of power no?

 

[Kenmitch]

That would depend on how much you overclock the cpu. Depending on the chip and mb you may be able to get a nice overclock with the voltage set to normal. With voltage set to normal and the speedstep enabled the voltage will drop with the multiplier. Under load the voltage rises with multiplier.

An example would be a e5200 which most of the time will go 3.33-3.66ghz or so stabily on voltage set to normal....Well at least the couple that I've tested out anyways.

Your cpu will use less energy at idle than it does on load. Your cpu will use less energy at idle @2ghz than it would at idle @4ghz this is one of the reasons for speedstep. The other one would be I guess heat output. Your cpu will put out less heat the slower it is running.

 

[Maximilian]

That would depend on how much you overclock the cpu. Depending on the chip and mb you may be able to get a nice overclock with the voltage set to normal. With voltage set to normal and the speedstep enabled the voltage will drop with the multiplier. Under load the voltage rises with multiplier.

An example would be a e5200 which most of the time will go 3.33-3.66ghz or so stabily on voltage set to normal....Well at least the couple that I've tested out anyways.

Your cpu will use less energy at idle than it does on load. Your cpu will use less energy at idle @2ghz than it would at idle @4ghz this is one of the reasons for speedstep. The other one would be I guess heat output. Your cpu will put out less heat the slower it is running.

Even if the voltage remains the same? I cant set it on normal as the mobo gives it too much juice for my liking at 3.6ghz itll use 1.28v, it runs prime stable at that speed with just 1.175v so i wanna keep it like that manually.

 

[Kenmitch]

Even if the voltage remains the same? I cant set it on normal as the mobo gives it too much juice for my liking at 3.6ghz itll use 1.28v, it runs prime stable at that speed with just 1.175v so i wanna keep it like that manually.

Do you mean auto setting on voltage? Whats the stock vcore of your chip?

The highlighted would be true....Even if you leave the voltage lock'd to 1.175 the core speed also plays a part in the power consumption and heat output....Not just the voltage.

 

[IntelUser2000]

Of course it does. Power usage is usually linear with frequency changes. Reduce frequency by 10%, reduce power by 10%. Voltage has greater effect, especially coupled with frequency changes. Switching all those transistors aren't free you know.

 

[Maximilian]

Of course it does. Power usage is usually linear with frequency changes. Reduce frequency by 10%, reduce power by 10%. Voltage has greater effect, especially coupled with frequency changes. Switching all those transistors aren't free you know.

Right cool thats what i wanted to know.

 

[drizek]

I don't know if there is something similar for intel(im sure there is... something like RMClock) but with my AMD setup I set custom volatages for it, so it is actually undervolted at idle and overclocked and overvolted on load. In the BIOS I set up the highest stable OC on stock volts, which allows me to run it OCd in OS X or Linux while still having CnQ.

 

[soccerballtux]

Of course it does. Power usage is usually linear with frequency changes. Reduce frequency by 10%, reduce power by 10%. Voltage has greater effect, especially coupled with frequency changes. Switching all those transistors aren't free you know.

AMD transistors are made better than Intels and can switch with less power. Power scales down with the square of reduced frequency on the AMDs.

 

[soccerballtux]

I don't know if there is something similar for intel(im sure there is... something like RMClock) but with my AMD setup I set custom volatages for it, so it is actually undervolted at idle and overclocked and overvolted on load. In the BIOS I set up the highest stable OC on stock volts, which allows me to run it OCd in OS X or Linux while still having CnQ.

This is an AMD only feature-- Intel processors you cannot control the voltage manually. For any enthusiastic user, this is a big drawback to using Intel processors, especially because of voltage droop.

 

[Munky]

This is an AMD only feature-- Intel processors you cannot control the voltage manually. For any enthusiastic user, this is a big drawback to using Intel processors, especially because of voltage droop.

I don't know what you're talking about, on my i7 rig I set the voltage as an offset value, and keep speedstep enabled. It automatically drops the clocks and voltage when idling, and still uses my OC when needed. No reason to run the cpu full throttle when surfing the web.

 

[Munky]

dp

 

[Ben90]

This is a motherboard feature. AMD processors are the most powerful processors ever! My 720BE finds 15 Mersenne Primes for GIMPS every minute. They also do your laundry from time to time, blow up terrorists, and have started a secret colony on mars. Oh yea Intel sucks

fixed

Im glad you pointed out that processors cannot create power for itself and therefor must rely on the motherboard to deliver power. Its people like you soccerballtux that keep this forum top notch with accurate information.

However I do not think you emphasized how amazing AMD processors are compared to spIntels. Let me break out the news for these poor unaware Intel-using folks:

First and foremost, Intel is absolutely evil in its existance. Its name being so long compared to AMDs is evidence right there. It takes exactly 5 bytes of ASCII to store the letters i-n-t-e-l, while AMDs name is a streamlined 3. What this means for us end users is that it takes more processing power to display "Intel" than "AMD", thus making spIntel users upgrade their computers more often. AMD processors are also more powerful, but the dreaded Intel Compiler hampers performance. Once the unreleased AMD compiler, codename: "Jesus's-compiler", comes out even single core Sempron processors will be outrunning overclocked i7-965s by around 7000%.

I could go on, but everyone knows the facts already. If you are just joining AnandTech and do not know how amazing AMD processors are I suggest researching for yourself. Good Google searches include: "intel sucks" "intel is so slow" and "amdzone.com"

 

[drizek]

I don't know what you're talking about, on my i7 rig I set the voltage as an offset value, and keep speedstep enabled. It automatically drops the clocks and voltage when idling, and still uses my OC when needed. No reason to run the cpu full throttle when surfing the web.

Ya...

I had the same setup on my Pentium M laptop. I am actually interested in what apps people use for this nowadays so that I can play around with undervolting my incoming i5 laptop.

 

[Munky]

Ya...

I had the same setup on my Pentium M laptop. I am actually interested in what apps people use for this nowadays so that I can play around with undervolting my incoming i5 laptop.

Maybe it's specific to certain motherboards, in my case it's a bios feature.

 

[drizek]

I seriously doubt I will be able to do anything in the BIOS on a laptop. Back in the day I used a program called Notebook Hardware Control that did a really good job.

RMClock used to do the trick as well, but that hans't been updated in years.

 

[nyker96]

If your cpu/mb can remain stable under s.s. turn it on it saves power. if not just bump the base voltage for cpu up a bit usually that also raises s.s. voltage when throttled back eventually you should get a stable system while still retaining s.s.

 

[Maximilian]

My mobo locks the voltage when its set manually so even speedstep cant change it, which i gather from googling around is a good thing, speedstep does influence frequency though. Its undervolted a bit anyway.

 

[soccerballtux]

fixed

Im glad you pointed out that processors cannot create power for itself and therefor must rely on the motherboard to deliver power. Its people like you soccerballtux that keep this forum top notch with accurate information.

However I do not think you emphasized how amazing AMD processors are compared to spIntels. Let me break out the news for these poor unaware Intel-using folks:

First and foremost, Intel is absolutely evil in its existance. Its name being so long compared to AMDs is evidence right there. It takes exactly 5 bytes of ASCII to store the letters i-n-t-e-l, while AMDs name is a streamlined 3. What this means for us end users is that it takes more processing power to display "Intel" than "AMD", thus making spIntel users upgrade their computers more often. AMD processors are also more powerful, but the dreaded Intel Compiler hampers performance. Once the unreleased AMD compiler, codename: "Jesus's-compiler", comes out even single core Sempron processors will be outrunning overclocked i7-965s by around 7000%.

I could go on, but everyone knows the facts already. If you are just joining AnandTech and do not know how amazing AMD processors are I suggest researching for yourself. Good Google searches include: "intel sucks" "intel is so slow" and "amdzone.com"

I'm smiling really big right now

 

[exar333]

AMD transistors are made better than Intels and can switch with less power. Power scales down with the square of reduced frequency on the AMDs.

Uninformed fanboy comment here; everyone please isgnore this grossly inaccurate comment.

 

[aigomorla]

AMD transistors are made better than Intels and can switch with less power. Power scales down with the square of reduced frequency on the AMDs.

So Since when did AMD get intel's patents on the Hi-K gates?

http://download.intel.com/technology/silicon/HighK-MetalGate-PressFoils-final.pdf

Superior? Dont make me laugh.... seriously..

 

[Ben90]

Yea right Aigomorla, you just linked a whole bag of spin from spIntel.com. Everyone knows AMD is far superior from spIntel. The facts are that every single reviewer ever is sponsored by spIntel, and.... and... Intel sucks!

In all seriousness though, the cpu has absolutely no control over the voltage fed into it. All it can do is suggest to the motherboard that it wants 1.225 or 1.2 or whatever, but the motherboard doesn't have to listen to the processor. That's the whole concept of overclocking. The processor suggests 133 bclock or 200 HTbus, but we tell the processor to screw off and force it higher through the mobo

Oh yea, and for the transistors being lower power:

 

[Hyperlite]

AMD transistors are made better than Intels and can switch with less power. Power scales down with the square of reduced frequency on the AMDs.

i wish we had that crazy laughing ROFL emoticon on here.

 

[soccerballtux]

i wish we had that crazy laughing ROFL emoticon on here.

 

[MJinZ]

I mean the cpu keeps the same voltage anyway no matter what the mhz is according to cpu-z so its using the same amount of power no?

That's not true. You can set your CPU to super high voltages but speed step will still drop it down when it's on its lowest multiplier.

 

[RussianSensation]

That's not true. You can set your CPU to super high voltages but speed step will still drop it down when it's on its lowest multiplier.

That's not exactly true. My UD3R took until Bios F3n to get this feature (dynamically changing Vcore) to work correctly. Otherwise I was cruising at ~ 1.7ghz at the same load voltage of 1.312V that I need to get my 3.9ghz overclock. The motherboard has to be able to support Dynamic CPU Vcore (in overclocking) for the cpu voltage to vary between idle and load frequencies. While most new motherboards support dynamic Vcore at nominal cpu frequencies, not all new motherboards support this feature once you start overclocking.