Date   : Tue, 31 Jan 2017 16:34:15 +0100
From   : kortink@... (John Kortink)
Subject: Pi-based second processor - fast, flexible,

On Tue, 31 Jan 2017 16:19:48 +0100, I wrote:

>[...]
>
>I'm not sure if a test like 'ClockSp' is a proper indicator
>at these speeds. It may be introducing accuracy problems.
>
>[...]

Indeed, if you look at the progress reported on the forum, it
is telling that, at the start, e.g. 'repeat' is 69 and 'trig'
160 (160 / 69 = 2.31), while at the end it is 273 vs. 286
(= 1.04).

So you're clearly starting to measure the overhead there, not
the operations themselves. The length of the loops measured
should be increased by factors to get a more accurate result.


John Kortink