Intel actually more efficient for gaming?

[henson0115]

Even Einstein wasnt 100% correct all of the time 99.9% will do

 

[CuriousTomCat]

I'm curious if this is correct, or if 100% of the consumed power is converted into heat.

While the CPU is operational, approximately 99.9999999% of the energy is converted to electromagnetic radiation. Heat and light are both wavelengths of electromagnetic radiation. The other 0.0000001% is added to the rest mass of the CPU, heatsink, and all the other components that heat up. The amount of mass gained is calculated using E2=(pc)2+(mc2)2. However, the value is too small for any current measuring device to detect.

When the computer is turned off and the temperature returns to that of the surrounding area, the rest mass will return to its original value.

No energy is ever used up. It`s just converted into another form.

 

[Ryan-3]

I like how we're discussing output energy of a CPU. I mean, energy can be output to transform the state of something chemically (potential energy), provide a motion/friction (kinetic energy) or just converting it to a wave of a given length (sound, light, heat).

Everything in a CPU eventually dissapates away as heat output. If you can hear it squeal or it fires out like a thrown piston... you've got bigger issues than paying a few more quid a year in electric bills.

 

[0h_no_not_again]

Urgh, why am I getting drawn into this stupid discussion.

I haven't seen a correct response here to absolute efficiency. It has nothing to do with E=MC^2 or anything like that.

The CPU has inputs from the motherboard in the form of power and signals, and outputs to the motherboard in the form of signals.

The efficiency is the (power of the outputs) / (power of the inputs), where the power of the outputs is tiny low voltage signals that carry information.

For a very particular task, I could design an FPGA that produced the same output as the CPU for a specific set of inputs using 100th of the power, i.e. 100x more efficient. However, If I were to design an FPGA based processor capable of general and complex computation would consume 100x more power, i.e. 100x less efficient. The point is CPUs are terribly inefficient but incredibly useful for a broad range of tasks.

So efficiency in this situation is only useful as a relative measurement: performance per watt. Cars are benchmarked the same way miles/gallon, litres/100km, despite the manufacturers having the absolute efficiencies of their power plants over a defined cycle (maybe 30%).

 

[alec]

power of the outputs is tiny low voltage signals that carry information

Is it really the useful output of a CPU? I would guess that power of output signal pins of a cpu has remained relatively constant since maybe Pentium era. But the work completed, computation, has grown couple orders of magnitude.
As in your example, you could design a CPU withsame performance but output signals 100x less powerful, wouldn't make it more efficient.

Thermodynamic discussions have no place in CPU efficiency measurement
Computational performance per watt (and as in op article, watts to reach some level of performance ie playable fps) is where it is at.