NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/programm/2071

Text File | 1992-07-22 | 3.3 KB | 69 lines

Newsgroups: comp.programming
Path: sparky!uunet!s5!is1.is.morgan.com!is.morgan.com!berlin
From: [e-mail] (Alexander Berlin)
Subject: Re: Does "400% faster" make sense? (Was: Re: finding 1st one in integer)
Message-ID: <[e-mail]>
Sender: [e-mail]
Nntp-Posting-Host: chico
Organization: Morgan Stanley - IS
References: <[e-mail]> <[e-mail]>
Date: Wed, 22 Jul 1992 21:23:37 GMT
Lines: 56

In article <[e-mail]>, [e-mail] (Andy Lowry) writes:
|> In article <[e-mail]> [e-mail] (Alexander Berlin) writes:
|>  > Just a side note: it's impossible to have code that runs 400% faster. Twice
|>  > as fast means 50% faster. 100% faster means in NO TIME.
|>  > 400% slower makes sense, 400% faster doesn't.
|>  > 4 times faster means 75% faster.
|>
|> Hmm... did somebody issue an edict on this of which I wasn't aware (or
|> was this the edict itself :-)?  I know these are confusing terms that

Must be your math teacher :)

|> are vague because they are used differently by different people, but I
|> can make a perfectly good case for saying that "400% faster" should
|> mean "at 400% the speed," i.e. "at four times the speed."  Or maybe
|> I'd prefer to say that "400% faster" means "at a speed that is 400%
|> greater" which means "at five times the speed."  This is also in line
|> with what I would expect a "400% speedup" to mean.  I certainly would
|> think it confusing if making something 75% faster were to result in a
|> 300% speedup!
|>
|> Your rule appears to be that if measured time is cut to x% of the
|> original, then you call it "(100-x)% faster" if x < 100 and (I'm
|> guessing) "x% slower" if x > 100.  But "fast" is a word that I
|> associate with speed, not time, so even if the world were to vote on
|> what these phrases were supposed to mean, I don't think I'd vote for
|> your definitions!
|>
|> And if the world doesn't vote and start using consistent definitions,
|> then your side note is pretty much without value (or perhaps of
|> negative value because of its misleading statements), since it doesn't
|> help me one iota in reading literature that makes use of such phrases.
|> I still have to rely on either a precise definition made up front by
|> the author, or on being able to deduce the intended meaning from
|> context, because I certainly can't assume that the author is following
|> your conventions!
|>
|> It's much better to point out the inherent vagueness in these terms as
|> they are used in practice, and encourage people to include precise
|> definitions when needed, than to claim that there's some universally
|> accepted precise definition when there isn't.

If some job takes me 100 seconds and you can do it in 25 seconds:
I am 300% slower then you are.              (100-25)/25*100
You are 75% faster than I am(not 300)       (100-25)/100*100
Your performance is 400% of mine.           (1/25)/(1/100)*100
My performance is 25% of yours.             (1/100)/(1/25)*100

And yes, to do the job 75% faster (in 75% less time) you need to increase your
performance by 300%. Vote is not required here.
|> --
|> Andy Lowry, [e-mail], [phone]
|> IBM Research, P.O. Box 704, Yorktown Heights, NY  10598
---
Alex Berlin
[e-mail]