On 14 June 2011 09:36, Alexander Gavrilov <angavrilov@...> wrote:
> That's very old and incomplete. My cl-simd is actually
> based on that, but improves some things:
*blush* You know, despite talking about cl-simd with you any number of
times last year, I'd forgotten that you also had the SBCL branch.
Can you summarize the state of your SSE patches? Is there something
still on your TODO list? Do you think the branch is shipshape to be
merged or does it need to be cleaned up first?
From: Alexander Gavrilov <angavrilov@gm...> - 2011-06-18 13:10:15
> > https://github.com/angavrilov/sbcl/commits/sse/
> > https://github.com/angavrilov/cl-simd
> Can you summarize the state of your SSE patches? Is there something
> still on your TODO list? Do you think the branch is shipshape to be
> merged or does it need to be cleaned up first?
Well, to tell the truth I haven't worked on it at all
since last year, except by rebasing occasionally. They
definitely need a review from somebody who knows the
internals enough to point out potential issues.
I have rebased them again today, and it still seems to
work, to the extent I tested it...
These first three patches only fix errors in instruction
definitions, so I think they can be considered separately
as a sort of a bug fix: