Messages in this thread | | | From | Scott McDermott <> | Date | Fri, 5 Mar 1999 18:54:11 -0500 | Subject | Re: MOSIX and kernel mods. |
| |
Richard Gooch on Fri 5/03 19:05 +1100: > > > I can't think of any CS invention that is used in the real world > > > that has cost 4 orders of magnitude in performance. > > > > paging to/from disk > > No, that's not a fair comparison. Paging/swapping is the only way to > extend virtual memory space without buying more RAM. Is there an > alternative method of increasing VM size without paging, and is it any > faster? Also, typical programmes don't suffer much performance loss > with paging.
This only goes to show how your "4 orders of magnitude" is ridiculous. Saying that you suffer a loss of speed equal to 4 orders of magnitude is like saying that because one's disk subsystem is 4 orders of magnitude slower, this means that the whole environment runs that slow. Sure it does...*when you are doing disk I/O and waiting on the data*. If you can divide your data up and process in paralell, and have infrequent lock communications over the network, then the 4 orders of magnitude loss you suffer while doing those communications will not slow you down anywheres near that much when looking at the overall picture.
Now if the whole system assumes that it's all running locally because the programmer has been lulled into equating the abstraction layers with reality, then your communications will take place inefficiently and those 4 orders of magnitude will start rearing their ugly head. Nobody is arguing that. One would always reduce inter-machine communications as much as possible.
-- Scott
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |