SCL execution speed

I need a special sorting algorithm, known as algorithm that solves unbounded knapsack problem. The proposed algorithm I choose is Martello-Toth MTU2:

http://www.or.deis.unibo.it/staff_pages/martello/cvitae.html

The author wrote this algorithm in FORTRAN, now first I have to translate this in SCL. But what execution times I can expect?

The author tested the algorithm on HP9000/850:

HP 9000/850 Cheetah (also HP 9000/850S):

• 27.5 MHz NS-1 processor
• 128 KB cache (combined I/D)
• 4096-entry TLB
• CTBs run at 9.16 MHz
• CIO I/O bus
• Maximum RAM of 128 MB with one memory controller (MC0) and 256 MB with two memory controllers (MC0, MC1) [it could be the 256 MB/two MCs were only supported on the HP 3000 equivalents]
• Performance of about 14 MIPS
• Price of about US $200,000 at time of introduction

Now what would be a good candidate of S7 CPU with SCL to compute the result maximaly 10 times slower?

Any good advice how to solve unbounded knapsack problem with S7, for small dataset (n=20)?