• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Intel® Power Gadget (for SandyBridge and newer)

Power Gadget working? What does it say your i7-2700k is using?
 
Last edited:
Burpo said:
Power Gadget working? What does it say your i7-2700k is using?
Click to expand...

Depending on how I set the Windows power plan;

Balanced - Max processor state at 50%, the gadget goes from under 10 watts to maybe 20 something watts max.

High performance - Max processor state at 100%, gadget goes 30 watts to 40 watts

BIOS set to all cores multi. 44, hyper-threading on, all power mgmt. stuff on, C6 state enabled, etc.
 
I thought this gadget wasn't working properly but my killawatt meter gave me the same results.

4.3GHz or 800MHz idle produces the same power consumption on a haswell i5. 😕

What's the point of dropping the clock speed at idle then? Does it just increase the longevity of the chip?
 
Durp said:
I thought this gadget wasn't working properly but my killawatt meter gave me the same results.

4.3GHz or 800MHz idle produces the same power consumption on a haswell i5. 😕

What's the point of dropping the clock speed at idle then? Does it just increase the longevity of the chip?
Click to expand...

Lowering the clock frequency should have a pretty substantial effect on the power usage of the CPU. You are saying that that the gadget gives the same measurement no matter what clock frequency you are running?
 
pm said:
Lowering the clock frequency should have a pretty substantial effect on the power usage of the CPU. You are saying that that the gadget gives the same measurement no matter what clock frequency you are running?
Click to expand...

Well there's a tiny, tiny difference. I guess it's c6/7 that's keeping the consumption low?

800MHz at idle
Gadget Package Pwr = 8.5w
Killawatt reading total system = 57w

4300MHz at idle
Gadget Package Pwr = 8.7w
Killawatt reading total system = 57w

Same settings between the two. I just opened the power options and set minimum processor state to 100%. The Intel gadget and other programs such as hwmonitor shows all cores of the processor staying at 4.3GHz when I do this.

😵
 
Haswells are very efficient. Mine is running fixed voltage and there is ~1w delta going from 800mhz to non-turbo 3.5Ghz. Those CPUs spend most of their time in C3+ state, so no wonder that power usage is dominated by those few hundred Mhz of unhalted cycles.
So 3.5Ghz is using more, but sleeping more as well, since it can complete the race to sleep way faster.
 
Durp said:
Well there's a tiny, tiny difference. I guess it's c6/7 that's keeping the consumption low?

800MHz at idle
Gadget Package Pwr = 8.5w
Killawatt reading total system = 57w

4300MHz at idle
Gadget Package Pwr = 8.7w
Killawatt reading total system = 57w

Same settings between the two. I just opened the power options and set minimum processor state to 100%. The Intel gadget and other programs such as hwmonitor shows all cores of the processor staying at 4.3GHz when I do this.
Click to expand...

I have no idea why yours is like that. I just tested mine - which is a Core i7 4770K with all 4 cores overclocked to 4.3GHz and I have an ATI 7970 video card, 1TB SSD, and a whole lot of fans.

800MHz at idle
Gadget Package Pwr = 15.2w
Killawatt reading total system = 71w

4300MHz at idle
Gadget Package Pwr = 26.3w
Killawatt reading total system = 83w

To get it to idle at 4300MHz, I went to Windows -> Control Panel -> Power Options and chose "High Performance". To get it to idle at 800MHz, I chose "Balanced".

You might say "hey, 26.3-15.2 isn't equal to 83-71" and I can't explain that. I just read the numbers. But I'd guess my Killawatt rounds the numbers and that's why.

But I will say that a ~11W delta is about what I would have expected. Maybe even a little better than I would have expected.
 
Check how your C-States are being used. The difference in KW reading is probably just efficiency loss when converting from mains power to power provided to the CPU.
 
Dufus said:
Check how your C-States are being used. The difference in KW reading is probably just efficiency loss when converting from mains power to power provided to the CPU.
Click to expand...

Can you clarify on the C-states? Check them how? And what am I looking for? And why am I checking them? 15W idle on a slightly overvoltaged 4770K doesn't seem too bad to me...
 
C-States are idle states which exist for cores and package.

For instance C0 is when the core/package is active, i.e. clocks are running.
C1 is when the core clock is halted, C3 and C6 go further to reduce power. The longer you spend in C6 the more power you save.

There are some softwares out there for checking C-State residency, IIRC realtemp / throttlestop and some softwares from Intel but I do not remember their names.

Your looking at how much time is spent resident in each C-State.

Why are you checking them, because I thought you were curious as to why Durp was seeing less power used. I was not suggesting you change anything, how you have your CPU set up is entirely up to you.
 
Hi, newbie playing with Intel Power Gadget on a i7-4930K Win7 64, all stock settings at the moment in the BIOS. Water block installed but no overclock. CPU power 74 watts at 49 deg C running Prime95 In-Place Large FFTs max heat.

I know this CPU is spec'd TDP at 135 watts, although I've measured more on overclock. So I'm wondering why Power gadget shows me a package power limit of a kilowatt! Does it calculate this dynamically from the actual operating core temperature or is it reading back (wrongly) a fixed byte chip power identifer in the processor?

Any replies appreciated - Thanks
 
Thanks Burpo - That makes sense. Strange how Intel software writers allow a numeric string like that, when they could have just trapped the inquiry response as 'unlocked' in the App or displayed a set of dashes. Didn't find that in the pdf manual either.
 
@voxmagna It's read from your PL1 limit set by BIOS. If you would prefer 150W say, then change your PL1 (long time) limit in the BIOS to 150W or to your maximum usable power usage, whatever that may be.
 
Last edited:
Took me a while to find it in the Asus P9X79 bios, where it is under Turbo Boost settings. Remove 'Auto' put in 150 (watts) and Power Gadget shows the new realistic upper threshold.

I have done more watts, but the problem as always is the thermal resistance losses between the substrate and the cooler block. I don't OC at core temperatures higher than 70 deg. C and the cores (6) are never at exactly the same temperature anyway, so I take the worst case. I guess Power Gadget is giving one average across all 6 cores.

Thanks for solving my problem!:thumbsup:
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top