On 12/24/15 23:52, Chen Gang wrote: > On 12/24/15 07:07, Richard Henderson wrote: > >> Moreover, I thought we agreed to do away with that CALC bit. >>
After check again, I guess, we can stil reserve CALC bit: - Then we can remove float32_to_sfmt (use high 32-bit to save float32 directly). And in helper_fsingle_pack2, for CALC, we can return high 32-bit directly. Only for NCALC, we need process it in details. - I guess, most cases are for CALC, so it will let the performance a little better (need float32_to_sfmt, then sfmt_to_float32 again). Then we can only focus on NCAL in helper_fsingle_pack2. Thanks. > > OK, I will try, next. > > I will copy and reconstruct related code from qemu fpu implementation > instead of (u)int32/64_to_float32/64 functions (just like you said, I > guess). > > Hope I can finish within 2015-12-31. > > Thanks. > -- Chen Gang (陈刚) Open, share, and attitude like air, water, and life which God blessed
