Re: question about 4.2 Multiplication in AES document(FIPS 197)

From: Tom St Denis (tomstdenis_at_iahu.ca)
Date: 05/30/03 

Date: Fri, 30 May 2003 12:58:57 GMT

Brian Gladman wrote:
> Your are right - what I should have said was and[a && b]

The whole point was to avoid a branch [though in retrospect it may be
faster on an 8051 since a branch only takes 2 cycles].

>>>I also think that the table sizes needed are one 512 and two 256 byte
>>>tables.
>>
>>That's what I originally specified...
>
>
> When I last did multiplication, 9x8 was 128 and 8x8 was 64 :-)

When you describe tables in terms of hardware its common practice to
give the input and output dimensions. E.g. serpent has 4x4, DES 6x4,
AES 8x8, CAST 8x32, etc...

Tom

Relevant Pages