Scott Putterman wrote:
>
> FYI, there's talk of changing the target to be a 300MHz machine
> instead of a 133. I think we might change the memory to 128MB as
> well.
On a faster machine, the numbers will all be smaller. This makes the
margin of error for timing by hand a much larger percentage. The slower
the machine, the less of a factor the error margin is. Since all of the
target speeds are faster times than Netscape 4.76, it doesn't matter
what speed the machine is. Just compare Mozilla speed to Netscape
4.76... which is easier to do on a slower machine.
Also, the bar graph, while nice, doesn't reflect the numbers from the
table above.
<http://www.mozilla.org/mailnews/performance/win_performance_results.html>
According to the graph, the time for "Message Reply Speed" improved from
1/24 to 1/31, but according to the table, it worsened. The graph also
shows the "Delete 5 Messages" time for 1/24 as >25s, but the table says
5.62s.
IMHO, it would be easier to read the graph if it were a line graph with
time horizontal axis (and the target as a straight horizontal line).
Each test should be a colored line moving across the graph.
The Amigo
