enk@sj.ate.slb.com (Edan Kabatchnik) (10/21/89)
Net, Recently there has been a thread of discussion comparing $/MIPS or MIPS/$ depending on the article. My question for you is "Why is this a reasonable means of comparing computing power for the money?" The reason why MIPS seems like a problematic unit of comparison to me is that different machines have different instruction sets. Some instruction sets are far more complex than others. For example, what a Symbolics 3600 and a VAX workstation can do in a single instruction might require half a dozen instructions or so on a RISC workstation. Could someone shed some light on this for me? +---------------------------------------------------+-------------------------+ | "The pain was enough to make a shy, bald buddhist | Edan Kabatchnik | | reflect and plan a mass murder." | enk@slcs.slb.com | | - The Smiths | enk@wheaties.ai.mit.edu | +---------------------------------------------------+-------------------------+
rcfische@polyslo.CalPoly.EDU (Raymond C. Fischer) (10/21/89)
In article <1989Oct20.190436.22949@sj.ate.slb.com> enk@slcs.slb.com (Edan Kabatchnik) writes: > Recently there has been a thread of discussion comparing $/MIPS or MIPS/$ >depending on the article. My question for you is "Why is this a reasonable >means of comparing computing power for the money?" The reason why MIPS seems >like a problematic unit of comparison to me is that different machines have >different instruction sets. Some instruction sets are far more complex than >others. For example, what a Symbolics 3600 and a VAX workstation can do in a >single instruction might require half a dozen instructions or so on a RISC >workstation. > > Could someone shed some light on this for me? Yup. 90% marketing hype. MIPS is usually used to impress people who think it actually means something. To be fair, there are a few cases where it is meaningful. The first is when your comparing two computers that use the same instruction set (as in two different models of VAXes) or when you're approximating to a standard instruction set (as in 4 VAX MIPS). You may notice that MIPS is often used when talking about RISC processors. Guess which kind this usage of MIPS is? Ray Fischer rcfische@polyslo.calpoly.edu
dce@sony.com (David Elliott) (10/22/89)
In article <1989Oct21.051814.13919@polyslo.CalPoly.EDU> rcfische@polyslo.CalPoly.EDU (Raymond C. Fischer) writes: >In article <1989Oct20.190436.22949@sj.ate.slb.com> enk@slcs.slb.com (Edan Kabatchnik) writes: >> Could someone shed some light on this for me? >where it is meaningful. The first is when your comparing two computers >that use the same instruction set (as in two different models of VAXes) >or when you're approximating to a standard instruction set (as in 4 VAX MIPS). >You may notice that MIPS is often used when talking about RISC processors. >Guess which kind this usage of MIPS is? Unfair. Different companies use different standards. MIPS (the company) has always tried to be quite fair, saying that a machine is 4 MIPS if it can run a large set of benchmarks 4 times as fast as a given configuration of a Vax 11/780. Other companies have, for various reasons, chosen to compare with different architectures, or use the "Millions of Instructions Per Second" numbers. For a while, there was talk of changing over to using VUP (Vax Unit of Processing) as a standard, but companies with inflated numbers didn't seem to join in very fast. This doesn't help the customer any, which is why SPEC was formed. SPEC is an organization started by a number of RISC vendors who wanted to come up with a standard set of benchmarks and comparison methodologies. Assuming that SPEC works as advertized, you will be able to obtain a report stating how various aspects of a machine compare. -- David Elliott dce@sony.com | ...!{uunet,mips}!sonyusa!dce
marti@ethz.UUCP (Robert Marti) (10/23/89)
In article <1989Oct20.190436.22949@sj.ate.slb.com>, enk@sj.ate.slb.com (Edan Kabatchnik) writes: > Recently there has been a thread of discussion comparing $/MIPS or MIPS/$ > depending on the article. My question for you is "Why is this a reasonable > means of comparing computing power for the money?" It isn't. After all, MIPS stands for "Meaningless Indicator of Processor Speed". Sorry, I just couldn't resist ;-) -- Robert Marti Phone: +41 1 256 52 36 Institut fur Informationssysteme ETH-Zentrum CSNET/ARPA: marti%inf.ethz.ch@relay.cs.net CH-8092 Zurich, Switzerland UUCP: ...uunet!mcvax!ethz!marti
pepke@loligo (Eric Pepke) (10/23/89)
MIPS is an acronym for Meaningless Information from Pushy Salescreature. Eric Pepke INTERNET: pepke@gw.scri.fsu.edu Supercomputer Computations Research Institute MFENET: pepke@fsu Florida State University SPAN: scri::pepke Tallahassee, FL 32306-4052 BITNET: pepke@fsu Disclaimer: My employers seldom even LISTEN to my opinions. Meta-disclaimer: Any society that needs disclaimers has too many lawyers.
bobd@hpuxa.ircc.ohio-state.edu (Bob DeBula) (10/24/89)
Actually, I've heard that MIPS stands for Mythical Instructions Per Second. Likewise, I couldn't resist :-) -=- ========================================================================== Bob DeBula | Internet: bobd@hpuxa.ircc.ohio-state.edu The Ohio State University | Disclaimer: These are my views, not the U's Davros sez: When my Daleks compute they use X-TER-MI-NALS!
alms@cambridge.apple.com (Andrew L. M. Shalit) (10/24/89)
In article <300@fsu.scri.fsu.edu> pepke@loligo (Eric Pepke) writes:
MIPS is an acronym for Meaningless Information from Pushy Salescreature.
No, no. It stands for "Mythical Index of Performance Something".
-andrew