crtz各个平台实现效率不一致

[MacroModel]

麒麟：
~/.../0002.file_io_tsc/time $ ./iobuf_file_old output:0.070751563s input:0.133950521s ~/.../0002.file_io_tsc/time $ ./iobuf_file_new
output:0.072596354s
input:0.146643229s
~/.../0002.file_io_tsc/time $ ./iobuf_file_new
output:0.067580208s
input:0.151073958s
~/.../0002.file_io_tsc/time $ ./iobuf_file_old
output:0.070063542s
input:0.134553125s
~/.../0002.file_io_tsc/time $ ./iobuf_file_old
output:0.066207292s
input:0.136011979s
~/.../0002.file_io_tsc/time $ ./iobuf_file_new
output:0.070186459s
input:0.1534625s

麦金塔new比old快

x86需要测有无bmi1的性能
龙芯小鸡去测
wasm默认提供
全部搞完后对特定平台开发，sto算法完结