Path: utzoo!attcan!uunet!microsoft!jimad From: jimad@microsoft.UUCP (Jim ADCOCK) Newsgroups: comp.object Subject: Re: C++ and garbage collection Message-ID: <57867@microsoft.UUCP> Date: 1 Oct 90 20:57:59 GMT References: <2030@aber-cs.UUCP> <57695@microsoft.UUCP> Reply-To: jimad@microsoft.UUCP (Jim ADCOCK) Organization: Microsoft Corp., Redmond WA Lines: 21 In article tom@ssd.csd.harris.com (Tom Horsley) writes: |To say that "nothing forces you to use near and far pointers in the same |program" is absolute nonsense. Performance is not `nothing'. When dealing |with insane and brain damaged architectures produced by the twisted |minds of Intel engineers it is often *absolutely necessary* to mix far |and near pointers in order to get a program that even comes close to |having an acceptable amount of overhead. Saying, "OK, well I'll just use |far pointers for everything" is a sure-fire way to make a 30 second |task take 2 hours. In partial defense of those twisted minds: 386sx, 386, and 486 machines are perfectly capable of programming in 0:32 flat model mode, and without any 10X performance hit. The real problem is those few millions of customers out there who keep buying 286 machines. 386, 486 also support 16:32 mode, which I believe will prove interesting for persistent objects. Where are 0:32 architectures going to go when 32 bit pointers prove too limiting? Will they go to 0:48 pointers, and lose backwards compatibility, or will they also move to 16:32 pointers? Or is someone going to start making a commercially successful OO CPU ? [[The real problem with Intel is they never include enough GD registers!]]