km@mathcs.emory.edu (Ken Mandelberg) (11/16/90)
Exactly how do you calculate the MIPS rating of a processor? Is it derived directly from the instruction set and cycle timings, or is it more empirical? -- Ken Mandelberg | km@mathcs.emory.edu PREFERRED Emory University | {rutgers,gatech}!emory!km UUCP Dept of Math and CS | km@emory.bitnet NON-DOMAIN BITNET Atlanta, GA 30322 | Phone: (404) 727-7963
de5@ornl.gov (Dave Sill) (11/16/90)
In article <1990Nov16.080815@mathcs.emory.edu>, km@mathcs.emory.edu (Ken Mandelberg) writes: >Exactly how do you calculate the MIPS rating of a processor? Is it derived >directly from the instruction set and cycle timings, or is it more empirical? It's worse than either of those approaches. What passes for a MIPS these days is really a "VAX dhrystone equivalent". If you assume that a 780 is a 1 MIPS machine, and run a dhrystone benchmark on it, you get a number, e.g., 1750 dhrystones/sec. (Ignoring the variations due to compiler, OS, dhrystone versions, system load and configuration) The you run (presumably) the same dhrystone benchmark on your system and divide its result by the VAX result. What you get is "MIPS". And Eugene thinks SPECmarks are bad... -- Dave Sill (de5@ornl.gov) Martin Marietta Energy Systems Workstation Support
rnovak@mips.com (Robert E. Novak) (11/17/90)
In article <1990Nov16.080815@mathcs.emory.edu>, km@mathcs.emory.edu (Ken Mandelberg) writes: > Exactly how do you calculate the MIPS rating of a processor? Is it derived > directly from the instruction set and cycle timings, or is it more empirical? > > -- This is exactly why SPEC was formed. There was no consistent method for measuring "mips" ratings. Do you use Dhrystone 1.1 "mips", do you count instruction counts (CISC manufacturers cried foul when RISC did this), do you take a suite of UNIX applications and measure their relative performance to a VAX (which VAX, what suite?)? This lead to a great number of claims and counterclaims about who had the best "mips" rating. With SPEC, at least the counterclaims have stopped. We now have definitive (if not perfect) answers. SPEC will continue to add applications to the SPEC suite. Stay tuned for more information. --- Robert E. Novak Mail Stop 5-10, MIPS Computer Systems, Inc. {ames,decwrl,pyramid}!mips!rnovak 950 DeGuigne Drive, Sunnyvale, CA 94086 rnovak@mips.COM (rnovak%mips.COM@ames.arc.nasa.gov) +1 408 524-7183