Path: utzoo!attcan!uunet!zaphod.mps.ohio-state.edu!rpi!dali.cs.montana.edu!milton!ogicse!oregon!jmeissen From: jmeissen@oregon.oacis.org ( Staff OACIS) Newsgroups: comp.sys.amiga Subject: Re: Where should X concentrate its CPU cycles? Message-ID: <47@oregon.oacis.org> Date: 25 Oct 90 17:07:14 GMT References: <2382@trlluna.trl.oz> <6877@sugar.hackercorp.com> <1990Oct24.221622.3575@zorch.SF-Bay.ORG> Organization: Oregon Advanced Computing Institute (OACIS) Lines: 37 In article <1990Oct24.221622.3575@zorch.SF-Bay.ORG> xanthian@zorch.SF-Bay.ORG (Kent Paul Dolan) writes: ->peter@sugar.hackercorp.com (Peter da Silva) writes: ->> every program is responsible for scheduling) or X (where the ->> application is responsible for maintaining display integrity) it seems ->> to me a waste of time to duplicate functions needlessly. -> ->To the contrary, this is exactly in line with the move from mainframes ->to microcomputers. Remember the horrid bottleneck when 20 students all ->logged onto a VAX 11/750? -> ->The common (server) device is going to be the bottleneck, so the ->appropriate thing to do is to offload as many cpu cycles as possible to ->the client machine. Anything else would lead to _dreadful_ performance. -> ->You pay for this with huge executables on the client, but that's ->perfectly fine when memory is cheap and cpu cycles (especially shared ->ones), as well as network bandwidth on a common ethernet are dear. But that's EXACTLY the problem with X. Instead of having the device which actually performs the display function (the server) handle clip regions and window refreshing, the application is responsible for these things,\ which puts the burden back on the central system running multitudinous X applications for users at distributed locations. PLUS it has to push all this data back and forth over the local net, adding to the net load. Don't confuse window manager with server. In most cases, the X user has a box configured as a server, tied in to a central system. This distributes the SERVER's function, while centralizing CLIENT activity. Obviously you want to off-load as much as possible onto the server! -- John Meissen .............................. Oregon Advanced Computing Institute jmeissen@oacis.org (Internet) | "That's the remarkable thing about life; ..!sequent!oacis!jmeissen (UUCP) | things are never so bad that they can't jmeissen (BIX) | get worse." - Calvin & Hobbes