Performance per watt, is it a good measure for CPUs?

[Prate_k]

Hi

When comparing cpus, are the performance per watt values suitable in considering a cpu (intel cpus). Of course other attributes such as the frequency, TDP and cache are important.. but sometimes, some cpus are too similar. In these cases are the performance per watt measures good to make decisions?

Thanks in advance

 

[datguy20]

I would only use performance/watt if you were considering a mobile CPU that needs to be efficient. People whine all the time about extra power usage between AMD & Intel, but if an extra 10W a day makes you go homeless, then a PC should be the least of your worries.

The best way to compare CPU's is with a real-world benchmark concerning whatever it is you want the CPU to do (so gaming/editing/etc).

 

[gangrel]

Performance per watt is what I'd call a final measure. Base measures would be those you note: cache size, CPU speed, and the like. Then you get the performance benchmarks, that should be correlated to the base measures. Performance per watt obviously is the performance benchmarks coupled with power requirements. So I'd look at the performance benchmarks first; given 2 options where the important benchmarks are pretty much equivalent, then it's a no-brainer to me to prefer the lower power draw.

Performance per watt is more important than raw performance when you're dealing with parallel systems, or in cases where power itself is a larger concern (laptops and tablets).

 

[Prate_k]

Sorry i did not state more info on my question. I was referring to notebook cpus, some of them are quite too similar to compare. But thank you for clearing my doubts on the issue and the quick replies. much appreciated.

 

[Prate_k]

just one more question. Would selecting a laptop with say i7 4712MQ processor heat up the laptop quickly? I mostly use my laptop for programing and gaming.
I'm looking at this laptop config specifically:
lenovo thinkpad edge 540c (i7-4712MQ/8Gb/1TB/nVidia GT 740M). would the GPU limit the performance of games (batman, crysis, assassins creed, league of legends etc..).

 

[Manish Tewari]

Hi

When comparing cpus, are the performance per watt values suitable in considering a cpu (intel cpus). Of course other attributes such as the frequency, TDP and cache are important.. but sometimes, some cpus are too similar. In these cases are the performance per watt measures good to make decisions?

Thanks in advance

Not strictly going by p=i*f/n; it depends on number of other factors viz power consumption, processing speed
channel capacity, latency, throughput, band width, rel efficiency, scalability etc

Hence, i would say that PPW is rather the energy efficiency measure - what you are looking for is perhaps floating point ops/ sec/ watt or F.L.O.P.S.

Hope that helps

 

[charlesNguy]

performace per watt is more related to your energy bill,

 

[teknobug]

I would only use performance/watt if you were considering a mobile CPU that needs to be efficient. People whine all the time about extra power usage between AMD & Intel, but if an extra 10W a day makes you go homeless, then a PC should be the least of your worries.

The best way to compare CPU's is with a real-world benchmark concerning whatever it is you want the CPU to do (so gaming/editing/etc).

Performance/watt is a big deal, my i5 3550 under load draws 35W-40W at stock (not total system) while my slower and older FX 4350 draws 110-112W. Why this wouldn't matter to someone is beyond me.

For that matter in mobile CPU's, it's about the battery life. My A4 5000 (15W) lasts 6 hours on battery while my older i3 4000M (37W) only barely clears 2 hours.