1 Montgomery Modular Multiplication in Hard- ware
1 Montgomery Modular Multiplication in Hard- ware
1 Montgomery Modular Multiplication in Hard- ware
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
FEI KEMT<br />
Nios processor that is powerful enough to control the process and reduces the<br />
resources usage to reasonable 1275 LEs.<br />
The MMM coprocessor is based on a 16-bit (w = 16) CSA PE with 6 (n = 6)<br />
pipel<strong>in</strong>ed stages and occupies 1290 LEs. The total area occupation of the<br />
second, mixed hard<strong>ware</strong>-soft<strong>ware</strong> solution is comparable to the purely soft<strong>ware</strong><br />
solution. The processor has been clocked at 50 MHz and the MMM coprocessor<br />
at 150 MHz. Times necessary for MMM and squar<strong>in</strong>g are presented <strong>in</strong> Table 2 –<br />
6.<br />
Table 2 – 6 Execution times of mixed hard<strong>ware</strong>-soft<strong>ware</strong> implementation of MMM on Altera Nios<br />
development board (with APEX EP20K200) for the CSA PE<br />
Length Method <strong>Multiplication</strong> Squar<strong>in</strong>g<br />
(e × w) (ms) (ms)<br />
1024 = 64 × 16 MWR2MM CSA 0.073 0.073<br />
2048 = 128 × 16 MWR2MM CSA 0.291 0.291<br />
3. The third design we analyse is based on the same system architecture as the<br />
one <strong>in</strong>troduced <strong>in</strong> the second po<strong>in</strong>t. This time the MMM coprocessor <strong>in</strong>cludes<br />
the 16-bit (w = 16) CPA PE with 9 (n = 9) pipel<strong>in</strong>ed stages. The parameters<br />
were chosen with purpose to get the occupied area size comparable to the<br />
other two design variations. The processor has been clocked at 50 MHz and<br />
the MMM coprocessor at 100 MHz. The results obta<strong>in</strong>ed for this configuration<br />
are presented <strong>in</strong> Table 2 – 7.<br />
Table 2 – 7 Execution times of mixed hard<strong>ware</strong>-soft<strong>ware</strong> implementation of the MMM on Altera<br />
Nios development board (with APEX EP20K200) for the CPA PE<br />
Length Method <strong>Multiplication</strong> Squar<strong>in</strong>g<br />
(e × w) (ms) (ms)<br />
1024 = 64 × 16 MWR2MM CPA 0.069 0.069<br />
2048 = 128 × 16 MWR2MM CPA 0.278 0.278<br />
41