>The results can still be useful, taken with a grain of salt or two.
or three, when you consider that a cache miss in present day RISC
processors can "stall" you for eight or ten clock cycles. Doesn't take
much to halve your effective performance. Got bit with this already ...
From: Ed Falis[SMTP:[log in to unmask]]
Sent: Wednesday, 12 February, 1997 9:38 AM
To: [log in to unmask]
Subject: Re: RE> sloc to bytes of code
Mike Brenner writes:
> There are two main reasons people on the web claim to measure lines of
> The first is to approximate the volume of work involved in developing
> software. The second is to approximate the volume of work involved in
> maintaining software.
There's a third reason - sizing resource requirements for embedded
applications. I think it's reasonable in an Ada context state the
following limitations: "excluding generic instantiations and calls to
runtime routines" as a way of making the estimate of code expansion more
meaningful. One then measures expansion for some set of code sequences
that approximate the application code, estimates runtime system
contribution under various conditions, and comes out with something
semi-meaningful, while remaining aware that the standard deviation for the
first measurement can be equal to the mean. The results can still be
useful, taken with a grain of salt or two.
Aonix (617) 221-7341