Re: Is this a joke?
> -----Original Message-----
> From: Mike The Spike
>
<snip>
> A Translator for DJGPP too now?
> let's test it...
> WTF????
>
> OK last time when I said that my translated programs ran
> SLOWER than the interpretted versions, I had the excuse of my
> Watcom version being Warezed and not the real thing, but now
> with DJGPP, wich has worked perfect for YEARS, IT'S STILL SLOWER!
> What the hell is going onhere?
> sequence.ex, shell.ex, sieve.ex, etc are all SLOWER once
> translated to C! About 10 TIMES SLOWER! WTF?
> Is this a goddamn joke?
> The slowness is visible, too.
> Things take longer to complete, the respoonse is slower, etc.
> So it's not just time() fucking up somehow (BTW ec.exe for
> DJGPP misses time.c or doesn't write it out to disk). Wha?
I ran sieve.ex, and it was about 3X faster after being compiled
(coincidentally, this is about what Rob was reporting--imagine that, a
software company that actually delivers on its promises :):
Compiled: 12203/sec
Interpreted: 4353/sec
Ratio: 2.8
I also tested my Matheval library, and it was over 10X faster for some ops
(eg, calculating (x+1)^32). So I'd guess that either your version of DJGPP
is out of date, or you didn't actually run anything. Let's see your source
code for DJGPP for sieve (or any of the other benchmark programs included
with Eu), and then we can compare the mythical U4IA++.
In short, I'm very impressed. I look forward to the Win32 version, and
putting up my $29.
Has anyone tried compiling anything that uses Neil? I assume this should
work, since it doesn't use Eu's built in graphics, right (not sure, since
I've never used it)?
Matt Lewis
|
Not Categorized, Please Help
|
|
