On Mon, Nov 29, 2010 at 3:58 AM, Michael Hope <span dir="ltr">&lt;<a href="mailto:michael.hope@linaro.org">michael.hope@linaro.org</a>&gt;</span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
I sat down and measured the power consumption of the NEON unit on an<br>
OMAP3.  Method and results are here:<br>
 <a href="https://wiki.linaro.org/MichaelHope/Sandbox/NEONPower" target="_blank">https://wiki.linaro.org/MichaelHope/Sandbox/NEONPower</a><br>
<br>
The board takes 2.37 W and the NEON unit adds an extra 120 mW.<br>
Assuming the core takes 1 W, then the code needs to run 12 % faster<br>
with NEON on to be a net power win.<br>
<br>
Note that the results are inaccurate but valid enough.<br><br></blockquote><div> </div><div>Thanks for doing this Michael. Very interesting numbers. Some questions/comments:</div><div><br></div><div>1. What is the load on the system when running your loop?</div>
<div>2. 2.73W for the board certainly means no PM. So the actual cost of NEON will be higher when you run a kernel with working PM.</div><div>3. What percentage of the packages in the main repo of Ubuntu generate NEON?</div>
<div><br></div><div>Regards,</div><div>Amit</div></div>