AMD FX-8350 Linux Performance-Per-Watt

[devius]

I'm sorry but the performance per watt graphs are wrong

No need to apologize, since you are correct as most of the posts in this thread confirm. The one who should be giving an explanation is Michael, but I don't even know if he realizes the formulas he used to calculate performance per watt are wrong.

Numbers
- green took 31.45 seconds @ 220 Watts
- blue took 33.03 seconds @ 262 Watts
So in performance per watt, green is the obvious winner - less time to finish, less average power.

That's pretty obvious. It's not like Phoronix has a ton of credibility anyway so no real harm done.

[xception]

I'm sorry but the performance per watt graphs are wrong, that is assuming the power consumption graphs are correct, for example in the C-Ray test the blue line is both higher and longer than the green line, so the system in blue (AMD FX-8150 @ 4.00 Ghz) took more power at each moment in time and also took more time to complete the test than the system in green (AMD FX-8350) - yet the performance per watt says that more is better and green is lower than blue (which from the other graph should green should have more) ... so either less is better in the performance per watt graphs or the formula used for the calculation is wrong.

Numbers
- green took 31.45 seconds @ 220 Watts
- blue took 33.03 seconds @ 262 Watts
So in performance per watt, green is the obvious winner - less time to finish, less average power.

[curaga]

Would it be the first time power saving on Linux didn't work right?

It could just as well be 8150 power-saving not working well on Windows. Don't jump to conclusions.

[bug77]

Hm, right. There's still the discrepancy of the idle and load numbers though (at stock), in Win reviews 8350 was lower in both, here the other way around.

Would it be the first time power saving on Linux didn't work right?

[curaga]

Hm, right. There's still the discrepancy of the idle and load numbers though (at stock), in Win reviews 8350 was lower in both, here the other way around.

[oleid]

@Michael:
Reading the article, every test result my brain stretched and wondered what was wrong. Yet, as mcgreg already said, the only really valid number to compare the processors' efficiency is to calculate the energy which is needed to get the job done. And this, Michael, should be sone by calculating the integral of your W(t) curves... As you have discrete measurements with a fixed step size in time, dt, it's nothing more than a sum over all measurements, multiplied by dt.

dt * \sum_{i=0}^{N} W_i

If you know how many floating point operations this task needs, you can calculate the energy per floating point operation -- which is the same as the average number of Watts per Speed (measured in flop / sec).

[mcgreg]

Meh, is there anything I calculated correctly?

[abacaba]

also I'd calculate
297,8 * 79,25 = 23601 kJ for the fx 8150 @ 4,6ghz
264,3 * 72,98 = 19288 kJ for the fx 8350 @ 4,6 ghz

mcgreg

Correct, except "J", not "kJ".

Or, one can get inverse: 1/23601 = 0.000042371, which means ~42 microlinuxkernels (are compiled) per joule. Here, the more the number is the better.

[mcgreg]

Ahh yes, I used the wrong numbers, and you corrected it, thx.

So, the results is clear: the FX 8350 is more efficient in performance / watt and michaels results is wrong then - this is actually what most people were already thinking, we just proved it.

[droste]

Well yes, if calculate the number by doing what is done here (full run of the same workload) it can be compared.

but it's
8350@4,6ghz 0,276s / 72,89s = 0,379 % of the kernel finished = 1 watt
8150@4,6ghz 0,266s / 79,25s = 0,336 % of the kernel finished = 1 watt

If you calculate this by actually give the CPU only 1 watt and say compile the kernel and see what is done, you can not be sure that the same things were computed in this time, because they may start compiling different files. Not that someone would do this :-D. But still, context is needed (as in how did you come up with the number).