As was foretold, we've added advertisements to the forums! If you have questions, or if you encounter any bugs, please visit this thread: https://forums.penny-arcade.com/discussion/240191/forum-advertisement-faq-and-reports-thread/
Options
Power usage: testing methodology
Hiravaxis
Registered User regular
Registered User regular
I read a review on hardware canucks for the new lynnfield processors.
Part of the review spoke on how good the power usage was on the i5.
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/22151-intel-lynnfield-core-i5-750-core-i7-870-processor-review-20.html
But the testing methodology was: measured the peak wattage through our UPM EM100 power meter.
What the hell good does that do me?
Is this a valid testing methodology at all or what?
If not, can someone link a review with a quality assessment of the i5's power usage?
Part of the review spoke on how good the power usage was on the i5.
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/22151-intel-lynnfield-core-i5-750-core-i7-870-processor-review-20.html
But the testing methodology was: measured the peak wattage through our UPM EM100 power meter.
What the hell good does that do me?
Is this a valid testing methodology at all or what?
If not, can someone link a review with a quality assessment of the i5's power usage?
Hiravaxis on
0
Posts
Can trade TF2 items or whatever else you're interested in. PM me.
Don't be a dick.
I did that and I know what peak wattage is.
In layman's terms it's the highest point on the curve of power draw over a period of time.
So I know that the overall power draw will be something less than that.
But that doesn't let me make any comparisons! Does it?
My question is: Is using peak wattage a valid testing methodology?
My intuition says "No".
Am I missing something?
They can only run so many tests though. HTPC-oriented gear usually gets a "Power during MPEG4/H.264/VC-1 Playback", but don't expect to see a "Surfing For Porn Power Draw" "Dicking around in TF2 Power Draw" or "Applying Windows Update Power Draw."
Can trade TF2 items or whatever else you're interested in. PM me.
I would have thought that a kWh measurement would have been more useful.
But I supposes since the test is based on an Idle CPU and a Fully Loaded CPU, that the average rate of power consumption would be very close to the peak wattage measurement.
kWh would be useful to determine cost of running the system over time, but you can get that by running some numbers based on peak wattage and length of time you're going to use it during the day vs. length of time it will be on and idle (or off, if you shut down while you're not using it) - but what kWh can't tell you is "just how big of a power supply do I need?" Peak wattage can get you more in the right direction for that. It would also require a fairly large assumption of how the system will be used - for example, it's going to suck up a lot less energy browsing the PA forums, reading and posting than it will watching HD video on Hulu (processor intensive) and that in turn will be less than playing TF2 (processor and GPU intensive). And it depends on how much of each activity you do in an hour. So an instantaneous measure is easier to extrapolate on, I suppose.
Can trade TF2 items or whatever else you're interested in. PM me.
Unless you are gaming non-stop or running a distributed computing program your CPU very rarely runs fully loaded. Watching Youtube, general surfing, Word, etc will use about 5-30%. Windows has some logging tools to monitor your CPU usage, but I am pretty confident most people average well below 50% most of the time. Not to mention that with multicore systems, chances are you aren't pegging all the cores.
To clarify, the peak wattage measurement found in the test would also likely be very close to the actual wattage used and any point in time because the CPU remains in a constant state throughout the test.
In other words, due to the nature of the test the lowest wattage is also very close to the peak wattage.
@Peregrine
Energy savings are what the 5i is being touted for, so I simply expected the testing to be based around that.
I wasn't looking at the test results expecting to see anything about the size of PSU I'd need.
And they are oriented around that - the instantaneous "peak wattage" is just easier to interpret and extrapolate on for the different types of users (casual office use, occasional peak use, hardcore gamer, F@H e-Peener) depending on how much time they'll spend at load vs idle.
Plus, it makes easier graphs to say "Core i5 uses 85W at peak. Core i7 uses 125W." rather than "Core i5 uses 100/200/400/800kWh (5% usage per hour, 25% usage per hour, 80% usage per hour, 100% usage per hour), Core i7 uses i'm too bored"
Can trade TF2 items or whatever else you're interested in. PM me.
They could just as easily have had the numbers be in kWh used over the course of the test.
It's still the same test they ran, just capturing different data.
Tho I'm not sure if you can easily start/stop the kWh data with the equipment they had, but I do know it does record that data. The logistics of running the test in that manner may have made for a less accurate test.