dlb5404@tamuts.tamu.edu (Daryl Biberdorf) (02/17/91)
In article <8651@gollum.twg.com> david@twg.com (David S. Herron) writes: > >The RAM & disk controllers make as much sense for an A3000 as for >a NeXTstation. The A3000 can have ~18Megs of memory on motherboard, >and has an SCSI-II interface on motherboard. 18 megs is smaller, >yes, but is far more than sufficient for both AmigaDOS and Unix >use. 32 Megs would be overkill, even (I expect) on a NeXT. *Any* machine that uses virtual memory can benefit from having more RAM. More RAM means fewer disk accesses which means better performance. While I think 18 MB is a reasonable limit for motherboard expansion on the 3000, one should *never* think that 18 MB is the most anyone will ever need or use. At my last co-op assignment, the quickest 'fix' for bad system performance (on a UNIX box) was 'install another 8 MB of RAM'. This was usually faster and cheaper than re-writing whatever application was inefficient. >when its clicked upon kinda disappears. The worst part of using >DP is that the programmers probably will have to program in >Postscript. Yes it's possible to program in Postscript, but its >"harder" because it's RPN and we're all taught to do our math in >INFIX. Not if you use a *real* calculator (one made by Hewlett-Packard)! 8-) (Disclaimer: I fought RPN for 4 years before (just recently) I got sick of doing matrix manipulations by hand or on my (slow) TI-59 and bought a brand new HP-42S this last Tuesday... And *I* *like* it!) >My decision to get an A3000 with Unix (ASAP) is because > >-- I want to run & work with X. NeXT's dont' use X as their native > windowing system so all those gee-whizzy interface building > things are useless to me. While agree with your point about NeXT not running X (although X is kind of icky in the amount of effort it takes to get results), NeXT's Interface Builder is *cool*. How much of your time do you really spend on getting the interface built? Too much. -Daryl Biberdorf, dlb5404@tamuts.tamu.edu OR @rigel.tamu.edu (note new uid) Texas A&M University
peter@sugar.hackercorp.com (Peter da Silva) (02/17/91)
In article <12268@helios.TAMU.EDU> dlb5404@tamuts.tamu.edu (Daryl Biberdorf) writes: > At my last co-op assignment, the quickest 'fix' for > bad system performance (on a UNIX box) was 'install another 8 MB of > RAM'. This was usually faster and cheaper than re-writing whatever > application was inefficient. Well, except that fixing the application would have improved the performance of every system that used it simultaneously. Even with a relatively low installed base of X systems compared to PCs I think it'd be worth it to fix the GUIs. I had a nightmare last night in which I was prototyping an application using X, a fancy UI toolkit, a prototyping tool, something like Oracle, Amoeba, and so on. It took 666 Megs of VM to run, and all it did was pop up 4 windows and let you select displays in one by clicking hot-points on another. And the customer was *happy*. They wanted to buy the prototype! (666 Megs of VM. Musta been a side-effect of Telepathic Control Protocol) -- Peter da Silva. `-_-' <peter@sugar.hackercorp.com>.