Jump to content

Talk:Polytope model

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

not sure about definition of manifested loop. loop bounds can be symbolic value or non-affine expressions which can be treated as unknown global variable in loop analysis. Chunchen


The "simple example" code is invalid, since it's evaluating a negative index when i==0: a[i-1][j]. Maybe it should start from 1? Dgutson (talk)

Correction needed for color sequence diagram

[edit]

Can someone please change the graphic that relies on a color sequence to one that relies on a numeric sequence (probably numerals in circles)? The current design is difficult for both normally-sighted users (the sequence of colors is both difficult to use, requiring either that you immediately recognize the strangely-truncated EGA color sequence or that you waste time looking back and forth, back and forth between the color listing and the diagram) and colorblind users (for some, figuring it out on their own could be impossible). — 172.58.145.100 (talk) 18:30, 10 February 2016 (UTC)[reply]

Example unclear

[edit]

`Why is the skewed version of this loop in the example is better than the non-skewed one? bungalo (talk) 04:30, 28 April 2010 (UTC)[reply]

-> The inner loop (p) can be executed in parallel in the skewed version, since there are no dependencies between these iterations. Vincent.loechner (talk) 16:22, 19 May 2010 (UTC)[reply]

You're right. But the article doesn't explain why the skewed version is more efficient. It just says that it's an optimization technique and gives an example. It doesn't mention parallelism at all. bungalo (talk) 20:03, 24 August 2010 (UTC)[reply]
It was clear before that change. Abradoks (talk) 06:55, 28 November 2011 (UTC)[reply]

Image error

[edit]

There is an error in the second image, t = 2j+i, NOT (1/2)j +i. giandrea 22:42, 30 October 2011 (UTC)[reply]

Can you find an image to replace it, or possibly correct it ?Beefcake6412 (talk) 23:07, 30 October 2011 (UTC)[reply]