• 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.

Effect of Temperature on Power-Consumption with the i7-2600K

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Idontcare said:
^ I just had an idea of how to run a test to capture the loss of efficiency as a function of the PSU...I need to insulate it in a box along with a remote temperature probe and then monitor power at the wall while leaving everything else constant on the rig.

I'm going away on vacation for the next 9-10 days and will be away from keyboard, but when I get back I will run some of those tests.

How hot can I get my PSU and still be safe? (I don't want to irreversibly damage my PSU by getting it too hot)
Click to expand...

I don't know the specifics about AX-850, but SPCR tested the X-400 here. They truly tortured the thing, running it at 100% (rated) load, without a fan, for hours and hours on end (15 hours straight). Component temperatures were running close to ~80C there. Seeing as how the AX and the X-400 are the same platform and manufacturer, I would think similar temperatures would not cause any ill effect.
 
kevinsbane said:
I don't know the specifics about AX-850, but SPCR tested the X-400 here. They truly tortured the thing, running it at 100% (rated) load, without a fan, for hours and hours on end (15 hours straight). Component temperatures were running close to ~80C there. Seeing as how the AX and the X-400 are the same platform and manufacturer, I would think similar temperatures would not cause any ill effect.
Click to expand...
If they are the same then the excess 450W is capped to enable fanless operation. It would be no different than to run the AX-850 at 50% load without a fan. It could be a different case if the load is tested at 100% where heat only adds to a list of factors that decreases the efficiency.
 
dma0991 said:
If they are the same then the excess 450W is capped to enable fanless operation. It would be no different than to run the AX-850 at 50% load without a fan. It could be a different case if the load is tested at 100% where heat only adds to a list of factors that decreases the efficiency.
Click to expand...

They aren't the same power supply, just the same platform. It is highly not recommended to run an AX-850 at 400W without a fan! The X-400 has larger heatsinks, more perforations in the casing, and did I mention more holes in the case? The thing is more holes than casing! I was just commenting on how safe it might be to run at temperatures of ~80C. The X-400 also has an overtemperature protection circuit, so I would not be surprised to see the AX-850 protect itself against excessive temperatures as well.
 
Nicely done!
what was your ambient room temps at the time of your experiment?
How much variance in cpu temps with respect to room temps, is what I'd like to know.
 
Nice work, glad to see theoretical power consumption still matches closely to reality.

While you never will find out the answers, I am a little curious to see how well your dynamic : static power ratio matches what we see in our databases. 🙂
 
Awesome data! Always nice when the measurements fit the physics 🙂

Obviously the next experiment is to plot the maximum frequency against temperature. And repeat all the studies with multiple chips to demonstrate die-to-die variation. And a give everyone a pony 😉

It'd be great if your results could be turned into an article to give overclockers a better idea of what's going on and how the various factors affect power and performance. Those of us who already have real-world data usually can't share it, and your methodology is fantastic for someone not using million-dollar machines. Anand should seriously consider this as front-page material!
 
TuxDave said:
Nice work, glad to see theoretical power consumption still matches closely to reality.

While you never will find out the answers, I am a little curious to see how well your dynamic : static power ratio matches what we see in our databases. 🙂
Click to expand...

Going by this equation of state:

PtotalVccTGHzSymbolicEquation.png


Taking a ratio of the Dynamic power to the Static Power results in:

DynamictoStaticEquation.png


Adding in the experimentally derived parameter values arrives at:

DynamictoStaticFormula.png


As expected, the ratio is dependent on the clockspeed, the operating voltage, and the temperature.

You can use that equation above to check the ratio results based on the specific test conditions for which your database values are documented (I don't know the specific temperature used for your in-house tests for instance).

But I can show you what I get with my CPU:

DynamictoStaticGraph.png


^ Understand these numbers are "messy" because the temperature is not held constant across these tests, nor is the voltage held constant. (but the operating Vcc and Temp for each point were taken into account when computing the relative static and dynamic power with which the ratio itself was then computed)

Perhaps unsurprising, the ratio appears to reach a maximum (app dependent too, this if for IBT) right near the spec'ed stock clockspeed for the 2600K...suggesting the clockspeed for the Sandy Bridge family was intended to operate at or near this maximum value.
 
Haven't even read the OP yet. Saw the thread name, crossed my fingers and saw it was started by IDC. My mind shouted "YESSSSSSSSSS!!!!!".

The hardware scene has been slightly less thrilling for me recently. Despite all the commotion and frenzy over BD, I was actually more excited seeing this thread than seeing "The Bulldozer Review: AMD FX-8150 Tested" on the front page of AT.

I really do know testing stuff like this is tedious. Hopefully you are having a great time doing this in the name of science. I dug this up and edited it.
idc.png
 
Sticky quality material here and I believe all AT members would agree.

Unfortunately a few paragraphs have me lost and I've even come back twice to it. I need to clear my head and try again.
 
Last edited:
I'm curious if we can discover any connection between it's power draw at temps, and the fact that seemingly despite every other piece of hardware I've owned over the last two years it don't actually benefit from it from a OC standpoint.

For instance my 1090T air vs cold water resulted in a 500MHz gain using nearly the same voltage, my 470's air vs water is about 100MHz core and 200MHz shader, and the power difference between air stock clocks with auto fan and stock clocks on cold water is insane. Also the voltage per clock with my 1090T and 470 are considerably improved by lower temps, meaning they required less voltage for the same clocks simply because of cooler operating temps.

Same thing with 1366 chips, i7's benefited greatly from higher end water loops...

Yet my i5-2500k does nothing extra despite the fact that I can run it around 30C full load at 5.3GHz.... Aside from the fact that it won't cold boot, stupid chip 🙁
 
Wow. Apparently, I shouldn't have withdrawn from cpu/overclocking forums after we went to only one (major) cpu manufacturer.

IDC, this actually isn't the article I was referencing, this is much much better than that! Great stuff.
 
:$ LOL, thanks for the kind words folks 🙂

It is rewarding to hear from so many people that they read and enjoyed this thread. :thumbsup:

You guys keep this up and I'm going to run out of excuses for not buying an IB to compare 32nm SB to 22nm IB in the same way :hmm:
 
Ahem, cough cough. It sure would be interesting to see how these 3d transistors handle when pushed to the limits. Especially with all the rumors of outrageous temperatures and whatnot. In the name of science of course.
 
I love idontcare threads. I don't get everything but I enjoy watching someone so passionate do good work. Thanks for the interesting numbers!
 
BallaTheFeared said:
Yet my i5-2500k does nothing extra despite the fact that I can run it around 30C full load at 5.3GHz.... Aside from the fact that it won't cold boot, stupid chip 🙁
Click to expand...

What voltage is that at? What's your ambient? Regular water cooling or chilled water?

Also, great read IDC :thumbup:
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top