This writer had looked forward to the day when the RISC versus CISC debate would be over. There are historically many good arguments on both sides since limited CPU resources had to be balanced with the human resources to program them (Brookshear, p85). But the convergence of the architectures (Hannibal, 1999), the emergence of such innovations as pipelining, superscaling and Very Long Instructions (Fischer, 1983), and the evolution of processors to such speeds that they are currently rarely the limiting factors of computing devices seemed to have rendered the dispute largely irrelevant (Wei, Huang & Naidu, 2001). Alas, now confronted with this very question I must be reminded of when I characterized X86 processors as being capable of 'millions of NOPs a second'.
As Brookshear notes: “The argument in favor of RISC is that such a machine is efficient and fast” while “CISC … is easier to program”. But Hannibal points out that these arguments date to a time when “Very Large Scale Integration (VLSI) yielded transistor densities … that couldn't fit too much functionality onto one chip”. So in a sense, as Wei et al assert, the “Debate [is] becoming moot”.
Many of today's commodity CPUs might be described as RISC cores with CISC interfaces. This makes perfect sense since microcode is only written once, when the architecture is designed, while higher level code will continue to be written for the entire useful lifetime of the CPU. However 'pure' RISC does still survive in some embedded and low power architectures (ARM, 2010).
At times this argument has evolved into ephemeral Superscalar versus VLIW (BDTI, 1999) discussions or devolved into PC versus Macintosh religious wars, but it often returns to this contrasting of rich and limited instruction sets. Granting that we “tend to be critical and disputatious” (Gislason, p19) this is not unexpected behavior. However I am now a little contrite with the astringency of my own arguments on the matter at one point. As anyone who has attempted to code assembly for the historical platforms can witness, Reduced Instruction Sets do have a somewhat greater tendency to align and therefore evidence fewer NOPs than Complex Instruction Sets.
Anon (2010) ARM: the architecture for the digital world [Online]. Available from: http://www.arm.com (Accessed 23, January 2010).
Brookshear, J. Glenn (2009) 'Computer Science: An Overview'. 10th Ed. Boston: Pearson Education Inc.
Fischer, Joseph A. (1983) 'Very Long Instruction Word architectures and the ELI-512', International Symposium on Computer Architecture [Online]. Abstract Available from: http://portal.acm.org/citation.cfm?doid=800046.801649 (Accessed 23, January 2010).
Gislason, Stephen J. (2005) 'Existence and the Human Mind'. Sechelt: Environmed Research, Inc.
Hannibal (1999) 'RISC vs. CISC: the Post-RISC Era', ArsTechnica [Online]. Available from: http://arstechnica.com/cpu/4q99/risc-cisc/rvc-1.html (Accessed 23, January 2010).
Wei, Y., Huang, B. & Naidu, A. K. (2001) RISC vs CISC [Online]. Available from