DES bitslice

DES Bitslice - http://dango.chu.jp/hiki/?Bitslice+DES#googtrans/ja/en

Bitslice DES is - make the throughput by parallel processing
DES depth discussion about implementing such software, please refer to the following sites.
Http://Ruffnex.oc.to/Kenji/Xrea/Des.txt (Aiko from Kenji's site)
Where DES is the most time eating the "encryption operation 16-stage" is represented by the loop. Processing time in the implementation is fully optimized for more than 90% of the total. UFC and libdes libraries, such as expanding the inverted S-Box · Gather references to conduct that reduce transposition tables, improving performance and reducing the number of memory references.
The opposite idea, but once that process in making a performance by processing multiple data in parallel Bitslice DES is slower. n n-bit processor machine assumes a 1-bit vector in parallel, run together. Currently, most modern CPU is 64bit or 32bit processor so, the 32 parallel, and can not run in parallel, or 64.
The problem in this parallel processing on Table S-BOX is called random. Reference to the table, each reconstructed block of 6 bits, since it is necessary to disassemble again.
S-BOX takes the input as 06-bit 64-bit output is 0 to 4 and 15. That would implement a combination of all input and output bit Boolean bit operations for this table is equivalent to reducing significantly the cost to reconstruct the sequence each time, DES register bit width of all the loop processing core not be implemented in parallel computing is a few minutes.
Cryptographer Eli Biham is an Israeli professor of this algorithm, DES for a modern high-speed RISC announced parallel implementations.
Improvements have been made repeatedly to reduce the number of bit logical operation, is currently implementing many of Mr. Matthew Kwan has said RISC processors as the best.

Application of SIMD instructions
For Intel x86 MMX / SSE * and PowerPC AltiVec (VMX) can be regarded as parallel machines and 128 not parallel 64. 64bit or 32bit general purpose registers that are only minutes parallelism, 64bit or 128bit SIMD can be operated in parallel or replace an order, you can calculate faster.
An Implementation of Bitsliced DES on the Pentium MMX Processor
http://www.engr.uconn.edu/ ~ zshi/course/cse5095/ref/yhilewitz_thesis.pdf

UNIX crypt and its application to parallel computing 2ch Trip
Since the UNIX crypt is an application of DES, Bitslice DES can still use the algorithms. UNIX crypt Bitslice DES algorithm was applied to the crack of the software as John the Ripper is well known.
Trip search program for Windows is also Tripcode Explorer Matthew Kwan's S-box based Bitslice DES algorithm is used.

Enhanced Bitslice DES - to support the implementation of logical operations-focused three-input
Moved to a separate page
*Cell SPU / Altivec for Enhanced Bitslice DES(http://dango.chu.jp/hiki/?Enhanced+Bitslice+DES#googtrans/ja/en)
*AVX + AMD XOP (VPCMOV) for Bitslice DES for AVX / XOP(http://dango.chu.jp/hiki/?Bitslice+XOP#googtrans/ja/en)
*AMD SSE5 (PCMOV) for DES Bitslice and SSE5(obsolete)(http://dango.chu.jp/hiki/?SSE5+and+Bitslice+DES#googtrans/ja/en)

No comments:

Post a Comment