geoff@desint.UUCP (Geoff Kuenning) (08/29/86)
Some of you may have noticed that I am writing the "UNIX and Real-Time"
chapter of "The UNIX Papers". As part of shooting off my mouth, I find
that I need to be able to talk about the System V virtual memory
system and compare it to the BSD implementation. Unfortunately, I don't
know much about either, and don't have access to the sources.
I would appreciate it very much if somebody knowledgeable could e-mail
me answers to the following questions:
(1) In general terms, how does System V virtual memory perform
compared to BSD?
(2) As I understand it, in the BSD implementation all processes page
against all other processes. This is as opposed to the more
modern method used in VMS, where each process pages only against
itself. Am I right about BSD? And which method does System V
use?
(3) BSD has a swapper which will page out an entire process under
certain conditions (presumably when it is hurting the CPU too badly).
This swapper does not seem to communicate much with the paging daemon.
Is this also true of System V?
(4) For both systems, what sort of pathological situations (notably
thrashing) do they tend to get into?
Thanks very much for the help.
--
Geoff Kuenning
{hplabs,ihnp4}!trwrb!desint!geoffdaveb@rtech.UUCP (Dave Brower) (08/30/86)
In article <251@desint.UUCP> geoff@desint.UUCP (Geoff Kuenning) writes: >Some of you may have noticed that I am writing the "UNIX and Real-Time" >chapter of "The UNIX Papers". As part of shooting off my mouth, I find >that I need to be able to talk about the System V virtual memory >system and compare it to the BSD implementation. Unfortunately, I don't >know much about either, and don't have access to the sources. At the risk of being flippant, how can you then be qulified to write about these subjects? It sure makes me question the credibility of the finished work. I think I'd be much happier if, say, Doug Gwyn, John Quarterman, or Henry Spencer wrote such a comparison, instead of someone who can't read the code. Is the the same "UNIX papers" by the same publisher who someone else accused of ridicuously low author royalty rates? Might this be adversely affecting the quality of submissions? -dB -- {amdahl, sun, mtxinu, cbosgd}!rtech!daveb
geoff@desint.UUCP (Geoff Kuenning) (09/03/86)
In article <429@rtech.UUCP> daveb@rtech.UUCP (Dave Brower) writes: > In article <251@desint.UUCP> geoff@desint.UUCP (Geoff Kuenning) writes:> >>Some of you may have noticed that I am writing the "UNIX and Real-Time" >>chapter of "The UNIX Papers". As part of shooting off my mouth, I find >>that I need to be able to talk about the System V virtual memory >>system and compare it to the BSD implementation. Unfortunately, I don't >>know much about either, and don't have access to the sources. > > At the risk of being flippant, how can you then be qulified to write > about these subjects? It sure makes me question the credibility of > the finished work. I'd suggest you learn to read before passing judgement. As it happens, I am not writing about either of these subjects; it's just that I need to mention them in passing. It also happens that I *do* know quite a bit about these subjects, but there are people on the net who know still more. In the interests of accuracy, I made a request for more information from those people. I guess it was predictable that the smaller-minded folks on the net wouldn't be able to resist sneering. > I think I'd be much happier if, say, Doug Gwyn, John Quarterman, or > Henry Spencer wrote such a comparison, instead of someone who can't read > the code. If the paper were titled "A Detailed Technical Comparison of the BSD, VMS, and System V Paging Systems," I too would prefer Doug, John, or Henry. However, as you quoted (but did not read, apparently), the title of the paper is "UNIX and Real-Time". It happens that the behavior of the paging subsystem can be critical to the performance of a real-time application (if you don't understand why, I suggest you buy the book when it comes out and do a little learning for a change). Since I don't have complete information about the behavior of these pagers, I made a polite request for help. I phrased it deprecatingly, as is my wont; this is not the same as saying I am ignorant. The final material on this subject will likely comprise a single paragraph, and will be accurate, precisely because I took the time and effort to check out my facts. As to the accusation that I "can't read the code", I'd like to know exactly where you got your evidence. People who have worked with me must be rolling in the aisles at that one. In fact, I can read and have read both the V.0 and the 4.2 kernels, among many others (and not just limited to UNIX). However, as I stated in my original posting, I do not currently have legal access to any UNIX code that would allow me to check the accuracy of my memory. I find it sourly amusing that a request for help from those who *do* have access has generated a totally unsupported (and totally unsupportable) accusation of rampant ignorance. > Is the the same "UNIX papers" by the same publisher who someone else > accused of ridicuously low author royalty rates? Might this be > adversely affecting the quality of submissions? There is a difference between an accusation and a conviction. This is not my first book contract, and I do not feel "ripped off" about the royalties, despite the fact that my first contract gave me 10% of all sales, and this one only gives me a one-time flat per-page fee. In the first place, I am a businessman and I understand the extra costs the Waite group must incur in a multi-author project. In the second place, anybody who expects to get rich writing a single technical book is a ripe candidate for buying the Brooklyn Bridge. And in the third place, anybody who thinks that the Waite Group's royalties are too low is perfectly welcome to send his book proposal to Wiley, Prentice, Van Nostrand, or any of a host of other publishers. Now can we stop this childishness and get back to the technical issues this group is supposed to be for? -- Geoff Kuenning {hplabs,ihnp4}!trwrb!desint!geoff
tim@ism780c.UUCP (Tim Smith) (09/06/86)
In article <429@rtech.UUCP> daveb@rtech.UUCP (Dave Brower) writes: >In article <251@desint.UUCP> geoff@desint.UUCP (Geoff Kuenning) writes: >>Some of you may have noticed that I am writing the "UNIX and Real-Time" >>chapter of "The UNIX Papers". As part of shooting off my mouth, I find >>that I need to be able to talk about the System V virtual memory >>system and compare it to the BSD implementation. Unfortunately, I don't >>know much about either, and don't have access to the sources. > >At the risk of being flippant, how can you then be qulified to write >about these subjects? It sure makes me question the credibility of >the finished work. > Instead of being flippant, why don't you learn to read what you quote? Geoff is writing about "UNIX and Real-Time", which he knows a lot about. He states that quite clearly. Also note that he says he does not have access to the sources. How do you expect him to go and read them? ( at least Geoff *can* read! ) I know that he has in the past had access to many different Unix sources, and has read them. He probably just wants to refresh his memory. -- I admit it! I don't believe in Mary Worth! I Lied! HaHaHaHaHaHa!!!!!! Tim Smith USENET: sdcrdcf!ism780c!tim Compuserve: 72257,3706 Delphi or GEnie: mnementh