weekly report Oct 31 - Nov 3

[Andrew Haley]

On 03/11/16 17:46, Anton Kozlov wrote:
> ARM32:

>   * 8819: first step of patching rework is done, approach taken
>   should be fastest according microbenchmarking it. Working on
>   loading from constant pool instead of in-instruction-
>   stream-constants. Microbenches seems aren't reliable here, adding
>   minor work to registers and/or ALU while loading significantly
>   changes results. Anyway, method under implementation should
>   produce better code. Core implementation is ready, need to clarify
>   strange situation with relocation on patched code and select
>   between approaches.

Interesting.  I'm looking forward to seeing what you come up with.

Andrew.