c++ - SSE FPU parallel -
i wondering if possible use sse in parallel x87. consider following pseudo code,
1: sse_insn 2: x87_insn
would pipeline execute 1 , 2 in parallel assuming can executed in parallel?
in modern (and older) processors, x87 , sse instructions use same execution units, it's unlikely benefit sort of code. there may special cases can trick processor running example x87 divide in parallel sse add, or that, if doing big loop of similar operations, there no benefit.
Comments
Post a Comment