I don't understand why Microsoft went to so much effort to use two instructions to load a 32 bit value instead of placing it in RAM and loading it. Was the RAM

I don't understand why Microsoft went to so much effort to use two instructions to load a 32 bit value instead of placing it in RAM and loading it. Was the RAM load so much of a delay in the pipeline that this hack was worthwhile?