macdonaldk@watt.ccs.tuns.ca (04/30/91)
The dhrystone benchmark is a synthetic benchmark and does not truely reflect the operations used by an arcade game. Arcade games usually do not require 32 bit operations, especially multiply and divide. Much of the time is spent drawing objects on the screen. The c64 has hardware support for this; the i386 does not--the dhrystone benchmark does not reflect this fact. When used for arcade games, the i386 manipulates video memory, which is much slower than main memory--again, the dhrystone does not take this fact into consideration. the c64 can load an 8 bit quantity from video memory in 2 uSec. A 386/33 would take approximately 0.3 uSec, hardly 300 times faster. I am not arguing that the c64 is superior to the i386 for arcade games, but it is a misuse of the dhrystone benchmark to use it as a metric for arcade performance. Also, the c64's C compilers are not very good; whereas, those for the i386 are very good. Kevin Macdonald Technical University of Nova Scotia Halifax, NOva Scotia