Find a small program, masters enter

雷卡特

Use the sse instruction to quickly compare two floating-point functions, ready to be embedded in VC.

Experts please provide source code, thank you!

xsteel

Quickly compare two floating point numbers

// what to compare?

雷卡特

Compare float

if (floata-floatb> 0.0001)
{
   ...
}
This is slower

I hope that the expert provides a function comparefloat written by the sink
if (comparefloat (a, b) == 0)
{
   ...
}

Seems to use cmpdd instruction? I'm not familiar with compilation, I hope to help, thank you

chiang999

Personally, I do n’t need to use SSE because the SSE instruction set is used to process large amounts of data, such as 4 FLOATs at the same time.
If it ’s not processing massive amounts of data, it ’s very inefficient.
Floating-point instructions are used, without looking at the instruction cycle, I don't know if it is fast
float comp (float x, float y)
 {asm
   {
   FLD x; load X into st (0)
   // FCOM y; you can also use the comparison instruction to set the sign bit
   FSUB y st (0) -y results in st (0)
   FST x stores st (0) into X
   }
   return x;

  }

雷卡特

Thank you

firemars

Just a few lines of code, no improvement