Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!sun-barr!rutgers!cbmvax!daveh From: daveh@cbmvax.commodore.com (Dave Haynie) Newsgroups: comp.sys.amiga.hardware Subject: Re: Is the A3000 Really Worth Buying? An honest question. Message-ID: <21895@cbmvax.commodore.com> Date: 24 May 91 17:15:14 GMT References: <377X24w162w@alchemy.tcnet.ithaca.ny.us> <1991May23.144823.16337@casbah.acns.nwu.edu> Reply-To: daveh@cbmvax.commodore.com (Dave Haynie) Organization: Commodore, West Chester, PA Lines: 54 In article <1991May23.144823.16337@casbah.acns.nwu.edu> tsarver@andersen.uucp (Tom Sarver) writes: >In article <377X24w162w@alchemy.tcnet.ithaca.ny.us> judge@alchemy.tcnet.ithaca.ny.us (rory toma) writes: >>> So, assuming that one has a couple of months more to wait for 2.0 and >>> ECS chips to become available for the A2000, can you really tell me >>> where the the major win lies in getting an A3000? >>One thing the 2000 doesn't have, is a 32-bit cpu slot. >In addition to above, I'm willing to pay for the 32-bit bus all around. >The only 16-bit bus in an A3000 connects the custom chips to stuff (chip >memory, I/O, etc.). Conversely, the only 32-bit bus on an A2000 is >between the 68030 and its 32-bit RAM. >Tell me that don't make a big difference. It does, especially for UNIX. If your just run a simple CPU benchmark, the A3000 without DRAM burst on and an A2500/30 should be the same. However, the CPU is the easy part. UNIX tends to rely pretty heavily on hard disk bandwidth. The A3000's disk DMA bandwidth is around 5x that of the A2500/30. Which means the CPU still gets to do lots of work, even when you're taking page faults. Also, UNIX's X windows is all CPU driven. The A3000's CPU has twice the bandwidth to Chip RAM that the A2500/30 does. So there are definite differences. >I'm also willing to buy into a platform (A3000) which was built to run Unix. >I'm not saying an A2000 CAN'T run Unix, but you might have a bit more >difficult time doing it. Strangely enough, the A2500 came into existence based on a desire to run UNIX, even though it managed to essentially get replaced a couple of times before UNIX was done. When the A500 was finished, Bob Welland, the guy behind the A500/A2000 Gary chip and the Gary/Fat Agnus architecture, went looking for a new project. He and George Robbins (the other 1/2 of the A500 systems team) had previously been working on the C900, a Z8000-based UNIX system that was cancelled when C= bought Amiga. Anyway, Bob wanted UNIX on the Amiga. So he started to play around with 68020s and custom MMU designs. He actually came up with an MMU and cache chip set that looked strangely like a 68040 when all put together. However, by then the 68851 was available, so he started building a 68020 proto board based on that MMU. About that time, I got done with A2000 stuff and came along to help out with the 68020 project, the A2620. A few months later, Bob left and I ultimately finished up the A2620, and later whipped up the A2630. So, while not as ideal for UNIX as an A3000, these A2500 systems were designed with UNIX in mind. The Amiga will handle UNIX better than most personal computers because it was designed with multitasking in mind, so things are DMA and interrupt driven rather than PIO and polled. -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "That's me in the corner, that's me in the spotlight" -R.E.M.