Furmark is an appropriate tool to test power consumption?

[Nevk]

Do AMD/NV will lower clock or loading while run this software?

Is it appropriate to observe power consumption ?

thanks.

 

[xorbe]

Not really useful, unless you play Furmark all day.

 

[AsRock]

I would have to say no as it over stresses way over the top imo, i personally use OCCT.

 

[Nevk]

So it is not accurate test to test card power consumption in Furmark?

Or it is more appropriate to see the power consumption in games or 3Dmark?

 

[MxPhenom 216]

Power consumption of what? Full system or just GPU?

 

[Vlada011]

I don't use Furmark at all. I do not care for Furmark or OCCT at all.
For Processors Prime95, for GPU EVGA Precision and OC Scanner.

 

[Jetster]

You can use Furmark 2013 to stress your system and check the power consumption. Its fairly accurate and test things that Prme will not. Its just a little temperamental sometimes and takes time. Intel Burn test will also work

 

[RejZoR]

Best way to test is to use the most advanced and demanding game. Because from my experience, even the most demanding game ever is never as demanding as Furmark. That's why i also never test overclocks with Furmark. It starts to lockup or artefact with it, but even with most demanding games, it never does.

 

[Octopuss]

Best way to test is to use the most advanced and demanding game. Because from my experience, even the most demanding game ever is never as demanding as Furmark. That's why i also never test overclocks with Furmark. It starts to lockup or artefact with it, but even with most demanding games, it never does.

But you could apply the same wrong logic to CPU testing. Prime95 makes a CPU sweat like nothing else, too, and yet we do consider it valid testing method (ideally along with IntelBurnTest).

 

[Mussels]

nvidia drivers throttle in furmark, AMD might too.


best bet: fire up a CPU stress test (i use OCCT linpack) and something like 3Dmark or a game with Vsync off at the same time, and get a peak result that way.

then do a game with Vsync on, for gaming results.

then do just linpack for 'CPU intensive' (video encoding) results.


i do this with every build i make, and measure them at the wall. lets me know if a PSU can handle the rig too, most store bought systems cant handle these kind of tests for long.

 

[AsRock]

So it is not accurate test to test card power consumption in Furmark?

Or it is more appropriate to see the power consumption in games or 3Dmark?

The program runs GPU's at extreme power usage which is way past what any game does.

 

[BarbaricSoul]

I don't understand why you guys are saying furmark is not good for measuring power consumption. I mean sure, furmark pushes your GPU harder than any game, but if your system can handle furmark, it can handle any game, right?

 

[Mussels]

I don't understand why you guys are saying furmark is not good for measuring power consumption. I mean sure, furmark pushes your GPU harder than any game, but if your system can handle furmark, it can handle any game, right?

no, because it tests it wrong. it maxes out SOME parts of the GPU and its VRM's, but misses other parts.


then theres the you know, entire point: the drivers throttle it to prevent it killing cards.

Its never going to show you power readings for any situation other than furmark.

 

[xorbe]

The way things are designed these days, it's impossible to find any one program that tests thoroughly, due to dynamic voltages and clocks, and throttling of specific programs, etc.

 

[Solaris17]

I would have to say no as it over stresses way over the top imo, i personally use OCCT.

I always use OCCT and tell it to run everything If I want a full system load test

 

[Octopuss]

no, because it tests it wrong. it maxes out SOME parts of the GPU and its VRM's, but misses other parts.

That's interesting. Is there any backup to this claim? Not saying it's not true, same thing seems to apply to Prime95 (which is why I recently started using IBT along with it), but this is the first time I see someone say that.

 

[erixx]

at least it is good to test ventilators or fans as some call them
and heat the room....

 

[Sasqui]

the drivers throttle it to prevent it killing cards.

How do you know this? ...serious question.

 

[AsRock]

You can use Furmark 2013 to stress your system and check the power consumption. Its fairly accurate and test things that Prme will not. Its just a little temperamental sometimes and takes time. Intel Burn test will also work

OMG no frigging chance, the most of any game or any app i have used on my system it just took it a extra 80-100w and that was in the 1st 2% of it running.

I did not see Furmark 2013 but found Furmark 1.11.0

Great way to kill your gear screw that crap. and if i was using vsync it's a easy 100w over the top. Watching thee wattage fly upwards like crazy like never seen before on my Zalman fan controller which monitors power usage b4 PSU.

They call it over kill and it does kill.

I always use OCCT and tell it to run everything If I want a full system load test

Best way without over doing it. Only other i would test with would be 3Dmark for usage and making sure DX is ok for the most part.

 

[erocker]

Or it is more appropriate to see the power consumption in games or 3Dmark?

Yes.

 

[xorbe]

How do you know this? ...serious question.

Go read a graphics card boost explanation. You can call it "boost under light load" or "downclock under heavy load". Also, we used to be able to rename the furmark.exe binary, and the down-clock would stop. Since it's all hardware monitoring based now, I doubt that has any effect.

I really don't like the word throttle, because many of us associate that with a -95% throttle when CPUs try to save themselves when the HSF fell off or partial contact. Extreme performance impact of course. (These days I think they just power off.) Now the graphics card clocks wander around 5% as the dynamic boost does it's thing, people bust out the T word and have a mental breakdown over 1 fps.