[PATCH v2.1] Use long for mantissa for generic mp code

[Richard Henderson]

On 2013-03-15 09:06, Richard Henderson wrote:
> On 2013-03-15 08:09, Richard Henderson wrote:
>> On 2013-03-15 08:05, Siddhesh Poyarekar wrote:
>>> Had a bit of a brain-fart there - I somehow thought this was a
>>> discussion on the benchmark patch.  The before/after is obviously the
>>> long mantissa patch.  Does the testsuite pass?  I might have missed
>>> some overflow case where a result needs to be stored in int64_t and
>>> not just long.
>>
>> Testsuite is still running.  I'll let you know what happens there.
>
> I'm going to say no, it doesn't.
>
> I saw that test-double.out had more than an hour of cpu time -- the whole
> testsuite should have run in 30 minutes or so.

I believe this can be replicated on an i686 build -- test-double.out now has 10 
minutes of cpu time here...


r~