Hmm, for micro-benchmark maybe, but this has to be measured once the JIT is ready. >From the stack VM, I can notice a slight slowdown w.r.t. 32 bits stack. But it doesn't really matter, stak is too slow compared to COG, so let's wait.
2015-12-21 21:24 GMT+01:00 stepharo <[email protected]>: > :) > So is it correct to think that a lot of number oriented computation will > go faster because not on large positive integers? > > Stef > > Le 20/12/15 13:39, Nicolas Cellier a écrit : > > SmallInteger maxVal highBit -> 60. > 3 bits reserved for immediate tags, 1 bit for sign, 60 remaining for > positive magnitude... > > 2015-12-20 11:56 GMT+01:00 stepharo <[email protected]>: > >> Super !!! >> Eliot what will be >> >> 1 class maxval >> then? >> >> Stef >> >> Le 17/12/15 20:12, Eliot Miranda a écrit : >> >>> Hi All, >>> >>> the real Cog 64-bit Spur x64 VM just evaluated 3+4 correctly on Mac >>> OS X: >>> >>> >> >> > >
