aglew@urbsdc.Urbana.Gould.COM (08/14/88)
>Someone a while ago asked what a "balanced system" is. I propose >the following definition for debate/flaming: > >"A balanced system is one where an improvement in the performance >of any single part would not increase the overall performance of >the system, and where the degrading of any single part would de- >crease the overall performance." > > Hubert Matthews Well, I asked the original question, and my response to this proposed definition is that it may well be a good definition - but I would seldom plan to build such a "balanced" system. This sort of balanced=saturated system would mean that, to improve the performance of the system, I would have to improve the performance of all components simultaneously. That is an expensive thing to do. Maybe incremental cost should enter the definition? Ie. improving the performance of any single part would not increase the overall performance of the system _at_ _an_ _acceptable_ _price_/_performance_ _ratio_. This gives flexibility to the defintion - if I can focus on one component, and find a way to reduce the incremental cost of improving its performance, I can improve the system changing only one component. Andy "Krazy" Glew. Gould CSD-Urbana. 1101 E. University, Urbana, IL 61801 aglew@gould.com - preferred, if you have MX records aglew@xenurus.gould.com - if you don't ...!ihnp4!uiucuxc!ccvaxa!aglew - paths may still be the only way My opinions are my own, and are not the opinions of my employer, or any other organisation. I indicate my company only so that the reader may account for any possible bias I may have towards our products.
davidsen@steinmetz.ge.com (William E. Davidsen Jr) (08/16/88)
In article <28200188@urbsdc> aglew@urbsdc.Urbana.Gould.COM writes: | | >Someone a while ago asked what a "balanced system" is. I propose | >the following definition for debate/flaming: | > | >"A balanced system is one where an improvement in the performance | >of any single part would not increase the overall performance of | >the system, and where the degrading of any single part would de- | >crease the overall performance." | > | > Hubert Matthews | This sort of balanced=saturated system would mean that, to improve | the performance of the system, I would have to improve the performance | of all components simultaneously. That is an expensive thing to do. Sounds like a great definition to me. I suspect that what you mean is "a system with equal bottlenecks in all portions, such that doubling the performance of any one part will produce an equal improvement in performance." A balanced system (matthews definition, not mine) is a very *inexpensive* implementation, in that no part has unused capacity which has added expense without adding performance. -- bill davidsen (wedu@ge-crd.arpa) {uunet | philabs | seismo}!steinmetz!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me
aglew@urbsdc.Urbana.Gould.COM (08/16/88)
> A balanced system (matthews definition, not mine) is a very >*inexpensive* implementation, in that no part has unused capacity which >has added expense without adding performance. > > bill davidsen (wedu@ge-crd.arpa) (1) I was referring to expense on the part of the manufacturer, not customer - the manufacturer would have to build the whole thing over again for version N+1. It often happens that designing in extra capacity in one subsystem costs nothing at all in design -- and may end up saving the customer money in the long run. How many people swapped up processors on their IBM PC systems? (2) Cost functions are strongly non-linear.
rik@june.cs.washington.edu (Rik Littlefield) (08/16/88)
In article <11857@steinmetz.ge.com>, davidsen@steinmetz.ge.com (William E. Davidsen Jr) writes: > ... I suspect that what you mean is > "a system with equal bottlenecks in all portions, such that doubling the > performance of any one part will produce an equal improvement in > performance." > > A balanced system (matthews definition, not mine) is a very > *inexpensive* implementation, in that no part has unused capacity which > has added expense without adding performance. A bit of apples and oranges here. The first paragraph relates improved *performance* in a part to improved performance of the whole. The second paragraph relates increased *investment* in a part to performance of the whole. I think most people implicitly use investment/performance. (Bangs per buck, right?) If I can get 20% net improvement by sinking $50 into part A or $30 into part B, you can bet I'll choose B regardless of what the individual performance hikes are. --Rik Littlefield #include std.disclaimer