Answering msg from Paul Edwards to rowan_crowe,
on Wednesday January 03 1996 at 00:10

 r>> I don't think you'll see much of a speed improvement. On the old
 r>> 8088s the string commands such as LODSB, STOSB, MOVSB were much
 r>> faster, but now "hand rolling" them discretely as
above can be just
 r>> as fast, sometimes faster.

 PE> As you will have seen before you get this message, I calculate
 PE> both the assembler combo and the C loop to come to to 8n, meaning
 PE> it is pointless to even convert it to assembler.  I had a quick
 PE> look at the 8088 timings, and it looked like the situation was
 PE> the same to me, ie the C loop is the same speed as the string
 PE> commands.

While calculating the number of clocks is fun, in the real world it might
be quite different. Especially with the Pentium which has nifty things
called AGI stalls, which cost a cycle or two, if the instructions are not
correctly ordered. (At least that's how I understand it).

I suggest you actually profile the code to see which one fares better.

---