|
Mike. The loop you see is a real loop as defined by a user. It is I think the only way to have a bwd jump.
On Jul 21, 2011 9:43 AM, "Benjamin Segovia" <segovia.benjamin@gmail.com> wrote:
Mike. After a second thout, I cannot transform the code with if then else. That changes the semantic of my program and I need to keep it intact since it is a simulator.
Remember that the code is not scalar so with simd, even a if then else (like if then else in cuda or ocl) is not a real if then else. They are really consecutive fwd jumps.
This version was just a quick test using closure but quicly replacing the closures by repeat ... until true (as I did initially) is the real issue.
The problem I see is that I just was unable to implement a fast forward jump luajit.
Anyway, I guess I am on my own now.
Sorry guys to disrupt your threads.
Cheers,
Ben
>
> On Jul 21, 2011 8:06 AM, "Javier Guerra Giraldez" <javier@guerrag.com> wrote:
>
> On Thu, Jul 21, 2011 at 7:04 AM, Alex Bradbury <asb@asbradbury.org> wrote:
> On 21 July 2011 12:47, ...
> Well I for one am very interested in this ...
>
> >> Maybe move this discussion to personal emails now.
> >
>
> +1
>
> these are interesting issues. even if my code is unlikely to hit
> those same cases, it...