Re: [boost] upcoming Switch library review Jan 5th - Jan 9th

From: Stjepan Rajko (stipe_at_[hidden])
Date: 2008-01-06 18:43:02

On Jan 4, 2008 3:31 AM, Alexander Nasonov <alnsn_at_[hidden]> wrote:
> Tobias Schwinger <tschwinger <at> isonews2.com> writes:
> > We actually very much want a preprocessor-generated 'switch' statement
> > because it is a special hint for optimization and most compilers
> > generate very efficient code for it...
>
> BTW,
> switch_ doesn't implement fall-though and I was worried about performance of
> this important case (bzero with Duff's device optimization):
>
>   switch(n % 8)
>   {
>     case 7: buf[6] = 0;
>     case 6: buf[5] = 0;
>     case 5: buf[4] = 0;
>     case 4: buf[3] = 0;
>     case 3: buf[2] = 0;
>     case 2: buf[1] = 0;
>     case 1: buf[0] = 0;
>   }
>
> switch_ would generate this code:
>
>   switch(n % 8)
>   {
>     case 7:
>       buf[6] = 0; buf[5] = 0; buf[4] = 0; buf[3] = 0;
>       buf[2] = 0; buf[1] = 0; buf[0] = 0;
>       break;
>     case 6:
>       buf[5] = 0; buf[4] = 0; buf[3] = 0;
>       buf[2] = 0; buf[1] = 0; buf[0] = 0;
>       break;
>     case 5:
>       buf[4] = 0; buf[3] = 0; buf[2] = 0; buf[1] = 0; buf[0] = 0;
>       break;
>     case 4:
>       buf[3] = 0; buf[2] = 0; buf[1] = 0; buf[0] = 0;
>       break;
>     case 3:
>       buf[2] = 0; buf[1] = 0; buf[0] = 0;
>       break;
>     case 2:
>       buf[1] = 0; buf[0] = 0;
>       break;
>     case 1:
>       buf[0] = 0;
>       break;
>     default:
>       break;
>   }
>
> Below is a program that demonstates a difference of assembly code between hand-
> crafted switch and the switch_. The are identical on gcc 3.4.6 x86_64.
>
> [snip program]

Hi Alexander,

Thanks for the example and the assembly analysis.  Off the top of my
head, it doesn't seem like it would be too difficult to add support
for fall-through directly (e.g., by taking an (optional?) sequence of
MPL bool constants to specify whether a case should fall through or
break/return).  Although, the difficult part might be deciding how to
deal with return values in this case.

Steven, what do you think?

Stjepan