Why? It's easy enough to divide the numbers directly.
Say, machine A has N MB/sec, and machine B M MB/sec. Thus, A is N/M times as fast as B.
"normalizing" would gain absolutely nothing. Actually, you would loose relevance in the numbers, since we are trying to optimize a specific task, normalizing to some 'standard' machine would make it way harder to know how fast the hashing really is.
/ Per Hedbor ()
Previous text:
2004-02-05 21:31: Subject: Nettle
It is comparing between machines that is difficult. Typically you would like to have a good baseline benchmark to normalize the rest with, like the dB scale for signal strength. hash-speed/memxor-speed or so.
/ Martin Nilsson (saturator)