On Wed, 29 Aug 2018, Jan Hubicka wrote:
> Hi,
> this patch reduces memory use of sreal by converting m_sig from int64_t to
> int32_t. This requires me to reduce SREAL_PART_BITS from 32 to 31 so there is
> space for sign. There is comment "SREAL_PART_BITS has to be an even number."
> but I found n
Hi,
this patch reduces memory use of sreal by converting m_sig from int64_t to
int32_t. This requires me to reduce SREAL_PART_BITS from 32 to 31 so there is
space for sign. There is comment "SREAL_PART_BITS has to be an even number."
but I found no reason for that comment. It is possible that I mi
