On 4/13/2015 5:16 PM, Josh Grams wrote:
That is helpful, but my basic objection still stands: you're computing
with *times* while the claim in the article was about *speeds*, I think.
He says "50% more work using the same amount of CPU cycles", which I
read as work/time. So you need to take the reciprocal of all your
values. Don't you?
The only numbers provided are cycles which are units of time. There is
no quantification of speed to analyze. Speed is not the inverse of time.
[Aside: cycles are not equivalent to work or to work/time - they are
simply time. Consider that "idle" cycles do not advance the computation. ]
The computation here is cycles/cycles so #1) the reciprocal does not
have a different meaning WRT preservation of units. More below.
The 2 claims made by the text are:
A) /"*3**.8.7 alpha version is **50% faster than the 3.7.17 release*"/
B) claim *A* is equivalent to "*/[3.8.7] does /**/50% more work
[than 3.7.17] using the same number of CPU cycles/*".
paraphrasing mine.
There isn't actually a claim of more work, but rather a claim that
faster = more work / unit time. I know that is a subtle distinction,
but the difference is important to judging the validity of the claim.
Leaving that aside ...
The operative terms of the claims respectively are "faster" and "more" -
both of which in English are expressions of the difference of like
units. [At least if properly used.] Difference is not a reciprocal
relationship, it is a reverse (opposite) relationship, so #2) although
the difference percent is expressed as a ratio, it is the motion vector
rather than the units that determines which of the end points should be
the base.
George
--
You received this message because you are subscribed to the Google Groups "Racket
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to racket-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
