[Patch libfortran/28452] random intrinsics for large reals

[Steve Kargl]

On Mon, Jul 31, 2006 at 10:00:30PM -0400, Jack Howarth wrote:
>    Unfortunately the long string of zeros is still present with the
> tonights gcc trunk and the random-patch-5.diff and the change...

(snip)

> As you requested I have added...
> 
> print *, x4(1:3) ; print *, x8(1:3); print*, x10(1:3)
> 
> ...which I placed before the bin summaries are printed out.
> The results are...
> 
   0.3090169
   0.309016994505426
   0.3090169945054261466733706525294548

   0.3524981    
   0.352498068323475     
   0.3524980683234748589735270183948808      

   0.8038055
   0.803805520479522
   0.8038055204795220681406659687936414 

This looks good.  Of course, doing some statistics on the 
last 18 digits would be interesting.

> 195974 196081 196285 196146 195718 196224 196325 196221 196183 195869
> 195974 196079 196287 196145 195717 196226 196325 196221 196182 195870
> 195974 196079 196287 196145 195717 196226 196325 196221 196182 195870

This looks good.

>    46.47314    
>    46.4762618000000     
>    46.47626180000000000000000000000068      

It could be the way I'm computing the chi-squared value or even it's
simply definition.

Thanks.

Thomas, I think is ready to commit.
-- 
Steve