ags@pucc-i (Seaman) (06/18/84)
> ... you probably don't remember the story a few years back in a trade > journal explaining how someplace had figured out a clever way to get around > the 16-MEGAbyte address-space limitation per process that exists on a 370. > That's right - someone not only blew out the top end of 16 Mb in one > process; it hurt enough that they "solved" the problem. > [Please don't bother to write to tell me that the 370 is most of the > problem. You can divide the memory size by 2 or 4 and it's still absurd.] Is there something wrong with running tasks in a virtual space larger than 16Mb? We run many such jobs on our CDC CYBER 205 every day. -- Dave Seaman "My hovercraft is full of eels." ..!pur-ee!pucc-i:ags
rcd@opus.UUCP (Dick Dunn) (06/22/84)
It's a little amazing to me that we're playing this game of "everybody has a VAX or 68000; why worry about 11's?" Does anyone realize that DEC has turned out a high-end 11-on-a-chip (or two or three) within the last year? That's point #1 - it ain't dead yet, by a long shot. Point #2: > So what ? Given any amount of memory, there will always be a > programmer out there to fill it up with an application that > used to run on half as much memory before, without any significant > improvements over anything. (Oh sure ! space vs. speed trade-off - > only if you knew what needs to run faster !!!) This was well-put, and it goes for ANY address space. If you don't believe it, you probably don't remember the story a few years back in a trade journal explaining how someplace had figured out a clever way to get around the 16-MEGAbyte address-space limitation per process that exists on a 370. That's right - someone not only blew out the top end of 16 Mb in one process; it hurt enough that they "solved" the problem. [Please don't bother to write to tell me that the 370 is most of the problem. You can divide the memory size by 2 or 4 and it's still absurd.] -- Dick Dunn {hao,ucbvax,allegra}!nbires!rcd (303)444-5710 x3086 ...Cerebus for dictator!
davecl@mako.UUCP (Dave Clemans) (06/24/84)
A lot of this discussion is getting off the point. What is important is that the "official" USENET protocol not require such things as megabytes of paged virtual memory, and that implementations of that protocol exist to demonstrate that fact. If other implementations exist that do use some of the special features of the local environment (say to experiment with different/better user interfaces) well and good. There's two sort of "cliche" points that can illustrate this point. I don't care what editor you use as long as the files it writes can be used by my editor (and vice-versa). I don't care what mail interface, transmission and receiving systems are used by you as long as they don't do something that interfers with mine (and vice-versa). dgc
thomas@utah-gr.UUCP (Spencer W. Thomas) (06/24/84)
(This isn't really the right place for this, but it started here.) I can easily see blowing the top off of a 16Mbyte address space. Some friends of mine are running programs that need at least 24 Mbytes to run (almost all data), and may be over that by now, I haven't talked to them recently. This isn't to say that that number might be reduced if they HAD to, but it's not worth the effort for them. =Spencer