/arch:SSE prefers to use fmul?

Lychy 1

Here's some sample code:
int n = 200;
float x = 0.2;
double z = n * x;

If I were to compile using /arch:SSE, the assembler code uses fmul,
but compiling with /arch:SSE2 results in it using mulss,

How come /arch:SSE doesn't use mulss despite it being an SSE1 instruction?

Minxin Yu 13,506 Reputation points Microsoft External Staff

2022-04-18T07:36:04.783+00:00

For your reference: SSE allows computations with vectors of up to four single-precision floating-point values. SSE2 allows the use of single and double precision. LINK: x87 FPU and SSE instructions.

Your answer