Comments on: Low Level Optimization Decays With Time http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/ my weblog. Thu, 17 May 2012 17:13:27 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Vincent Gable » Retest Your (Low Level) Optimizations http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/comment-page-1/#comment-541 Vincent Gable » Retest Your (Low Level) Optimizations Wed, 04 Mar 2009 07:45:33 +0000 http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/#comment-541 [...] written before about the decay of machine-specific optimization. Even if your code isn’t run by a VM, I think it’s reasonable to expect that (at least [...] [...] written before about the decay of machine-specific optimization. Even if your code isn’t run by a VM, I think it’s reasonable to expect that (at least [...]

]]> By: Vincent Gable » Blog Archive » memcopy, memmove, and Speed over Safety http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/comment-page-1/#comment-62 Vincent Gable » Blog Archive » memcopy, memmove, and Speed over Safety Sat, 24 May 2008 19:42:42 +0000 http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/#comment-62 [...] less value then changing the algorithm that’s requiring all that duplication. It may even suddenly become slower when hardware is [...] [...] less value then changing the algorithm that’s requiring all that duplication. It may even suddenly become slower when hardware is [...]

]]> By: Vincent Gable http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/comment-page-1/#comment-3 Vincent Gable Fri, 07 Mar 2008 23:59:31 +0000 http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/#comment-3 Yeah, the dividing line between "low level" and "algorithmic" optimizations isn't always clear. Sometimes hardware drives what algorithm is most appropriate. Vectored processors can make a hash/look-up-table based approaches a lot more appealing, for example. Yeah, the dividing line between “low level” and “algorithmic” optimizations isn’t always clear. Sometimes hardware drives what algorithm is most appropriate. Vectored processors can make a hash/look-up-table based approaches a lot more appealing, for example.

]]> By: Zach Beane http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/comment-page-1/#comment-2 Zach Beane Fri, 07 Mar 2008 14:03:50 +0000 http://vgable.com/blog/2008/03/07/low-level-optimization-decays-with-time/#comment-2 Unfortunately, many hand-tuned Lisp programs must work with the limitations of the implementation today, such as the native machine word size or the performance of various built-in operators. For example, I wrote a CRC32 routine that had to split up work into 16-bit halves to deal with the tiny fixnums of LispWorks (they've since gotten bigger), and I had to write my own routine to copy parts octet-arrays because REPLACE was just too slow by default. If the implementation improves, I have to revisit the code... Unfortunately, many hand-tuned Lisp programs must work with the limitations of the implementation today, such as the native machine word size or the performance of various built-in operators.

For example, I wrote a CRC32 routine that had to split up work into 16-bit halves to deal with the tiny fixnums of LispWorks (they’ve since gotten bigger), and I had to write my own routine to copy parts octet-arrays because REPLACE was just too slow by default.

If the implementation improves, I have to revisit the code…

]]>