Subject: Re: pointer size in [CC99]
From: Craig Chambers (chambers@cs.washington.edu)
Date: Tue May 02 2000 - 12:23:28 PDT
After more checking, this isn't really a bug, as all space data is in units of
words. So the space data I report in the paper should have units of (4-byte)
words, not bytes.
-- Craig Chambers
Mayur Naik wrote:
>
> Hello
>
> >> The experimental assessment in [CC99] considers the space cost of a pointer to be 1 byte. Wouldn't 4 bytes be more realistic?
> >
> >Why do you say it's only a byte? I intended it to be 4. Actual SPARC data.
> >
>
> The last para of sec. 5.2 in [CC99] says:
>
> In the dynamically typed versions, all expressions corresponding to predicate class boolean tests have only 2 possible classes, while all formal expressions have all 861 classes.
>
> However, the minimum and maximum size (in bytes) for the pure [] implementation in the dynamically typed version in Table 3 is specified as 8 and 867 respectively. The former is the sum of 2 bytes for pointers to the 2 possible classes plus 6 bytes for the array lookup instruction, and the latter is the sum of 861 bytes for pointers to the 861 possible classes plus 6 bytes for the array lookup instruction.
>
> I inferred that you assume a pointer to be of 1 byte from these results. Please correct me if I'm wrong.
>
> Thanks for your time.
>
> -- Mayur
>
> HotBot - Search smarter.
> http://www.hotbot.com
This archive was generated by hypermail 2b25 : Tue Oct 03 2000 - 15:21:33 PDT