[Date Prev][Date Next][Thread Prev][Thread Next]
- Subject: Re: bit.lshift and performance - luabitop v.s. lua-5.2.0-work4
- From: KHMan <keinhong@...>
- Date: Wed, 13 Oct 2010 21:53:28 +0800
On 10/13/2010 9:33 PM, David Kastrup wrote:
Anyway most app developers do not need to code DCTs and FFTs
I find that C compilers tend to be actually astonishingly bad at that.
Coding the inner loops in assembly language can often halve the
execution time. Somehow the optimizers/register allocators don't seem
to like complex butterflies all too well.
This was why I chose *specifically* DCT and FFT as a
conversational gambit -- most widely-released mainstream
implementations are still optimized in assembly language.
Mortals like me consume them in ready-made libraries. Would be
interested to see what other scripting languages are doing in this
A few hundred million ops per second doing them on Lua seemed far
off enough that it seemed to me like a reasonable future target,
but not something that can be currently done. Now since I was
thinking only in terms of Lua, it seemed reasonable enough to me.
Obviously, I am a horrid dunce.
Kein-Hong Man (esq.)
Kuala Lumpur, Malaysia