Converting binary integers of an arbitrary size to ascii decimal

[RuiLoureiro]

Dave,
       I said it is useless to me. And is. A 6 DWORD sized number was an example
       because FormalProof talking about numbers in some DWORDS and you
       about bignum

My reply was to caution you about trying to print a big number like 2^65536-1.
...

        I used the word "print" to give an idea, because you said
        "is not expressed in decimal".
       

And why not by a 64 bits value 10^N (or a 128 bits value...)

This could be done, but only if you have a system
that currently supports 64 bit registers.

        I suggested 64 because i was thinking in buffers with many qwords
        and a quicker method (div spends too many cycles). Instead of 100
        divs, 50 divs. It was the idea.
        To count digits in EAX (=4294967295) with div i got 621 cycles!

I don't think MMX is suitable for 64 bit multiplies or adds, I may be wrong about this.

        For what i saw, i think so and I may be wrong too,
        but i think we could use FPU (but i dont want to do it)

        EDIT: i think i never used the windows calculator till now !
Rui

[KeepingRealBusy]

Rui,

I just went to the AMD spec to see what was available. XMD divides are only for floating point. There are some interesting multiply options that seem to work on unsigned.

Dave.