Overclock.net banner
1 - 20 of 49 Posts

·
Registered
Joined
·
1,037 Posts
Discussion Starter · #14 ·
VonDutch: I would not put a lot of faith in the power consumption number that any software displays. Intel added some registers to the CPU and their documentation shows a method to calculate power consumption. All software seems to use this method. This data is generated to control the Intel Turbo Boost feature. It is based on the CPU VID voltage. It is not based on the actual voltage the CPU is using. If you go into the bios and manually set a core voltage that is completely different than the VID voltage then the power consumption displayed becomes a meaningless number. The only accurate way to measure CPU power consumption is to tap into the CPU socket and measure actual current flowing through the CPU.

Thanks to Dufus for setting me straight.

Maybe this power consumption number is useful when the CPU is idle. I have no idea. You would need to test and overclock a hundred CPUs to see if there is any correlation between a low power consumption number and a CPU that is a good overclocker. No matter how I have my 3570K set up, it shows about 21W at idle and it doesn't seem to be a great overclocker but I haven't had time to wring its neck yet.
smile.gif


Maybe someone could start a new thread that compares idle W to overclock speed. That might be interesting. You could also include Prime95 Small FFT power consumption during the first test just so the load is fairly consistent.

Your 3770K looks a lot better than my 3570K sucking 93W.

http://img138.imageshack.us/img138/2300/realtempaisuite.png

Do some stress testing with Prime95 version 27.7 and make sure the AVX instructions are being used to create some more heat.

Thanks for the feedback.
 

·
Registered
Joined
·
1,620 Posts
interesting, thanks. I will look into W consumption numbers as well on my end. I have an APC UPS battery with current wattage draw on display, so I can actually see how much power the computer draws at idle (16x multi, lowered voltage using offsets) vs with cpu under full load while other conditions stay same ... well MB power draw probably rises too when CPU goes under load, but either way I should be able to compare the numbers roughly in spare time.
 

·
Registered
Joined
·
2,767 Posts
Quote:
Originally Posted by unclewebb View Post

VonDutch: I would not put a lot of faith in the power consumption number that any software displays. Intel added some registers to the CPU and their documentation shows a method to calculate power consumption. All software seems to use this method. This data is generated to control the Intel Turbo Boost feature. It is based on the CPU VID voltage. It is not based on the actual voltage the CPU is using. If you go into the bios and manually set a core voltage that is completely different than the VID voltage then the power consumption displayed becomes a meaningless number. The only accurate way to measure CPU power consumption is to tap into the CPU socket and measure actual current flowing through the CPU.
Thanks to Dufus for setting me straight.
Do some stress testing with Prime95 version 27.7 and make sure the AVX instructions are being used to create some more heat.
Thanks for the feedback.
i just let prime run 30 min orso, to get to the small fft 8k, they create the most/alot heat i think

went up to about 84-85 W, with peaks to 86 W, 7gb ram used
 

·
Registered
Joined
·
1,037 Posts
Discussion Starter · #17 ·
feniks: In my testing with a Kill-a-Watt meter, I found best idle power consumption by enabling the C3/C6 core sleep states. On my Asus board, using the C3/C6 package C states actually increased idle power consumption. Not sure why. It might be something on the board that causes that.

The RealTemp C States window is kind of handy when trying to eliminate useless background tasks on your computer. I can get my 3570K cores to spend about 99.5% of the time in C6. If you do this, the multiplier and voltage don't seem to make any significant difference to power consumption at idle because the CPU is mostly asleep. I would definitely be interested in seeing any power consumption data from your UPS. No hurries. It takes me forever to get any testing done these days too.
smile.gif
 

·
Premium Member
Joined
·
10,558 Posts
I like the new version. The VID read looks like it could be handy, checked 4 x 3770ks, looking at the VID it was pretty accurate for picking the best overclocking chip. Lower VID = better overclocking.
As stated the watts need to be taken with a grain of salt, overclocking the cpu makes the watts on the watt meter jump, where the watts in realtemp did not change, still no way to accurately read the cpu watts with software.
 

·
Registered
Joined
·
1,037 Posts
Discussion Starter · #20 ·
Quote:
Originally Posted by FtW 420 View Post

The VID read looks like it could be handy, checked 4 x 3770ks, looking at the VID it was pretty accurate for picking the best overclocking chip. Lower VID = better overclocking.
Low VID was usually a good thing to look for when overclocking Core 2 CPUs. The Ivy Bridge CPUs have a lot more VID voltage steps so for a fair comparison between CPUs, you need to make sure the CPU load is as close to identical as possible.
 
1 - 20 of 49 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top