Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!cbmvax!daveh From: daveh@cbmvax.commodore.com (Dave Haynie) Newsgroups: comp.sys.amiga Subject: Re: Question about IBM and Multitasking: What is Micro channel? Message-ID: <14256@cbmvax.commodore.com> Date: 6 Sep 90 19:55:37 GMT References: <1132@orange9.qtp.ufl.edu> Reply-To: daveh@cbmvax (Dave Haynie) Organization: Commodore, West Chester, PA Lines: 45 In article joseph@valnet.UUCP (Joseph P. Hillenburg) writes: >Well...MicroChannel is 3 years old now. Basically, IBM (Blah!) tries to make >a bus that beats the AT style. It does, but there are better one, like >NuBus, EISA, Zorro II, and or course Zorro III (Doesn't Zorro III have >better thoroughput than the others?). In some situations, Zorro III may have a better throughput than some of the other buses. Specifically, the CPU to Zorro III interface in an A3000 runs faster than the CPU to NuBus connection in any Mac. There are two reasons for this: [1] Being an asynchronous bus, there's no need for the A3000's main CPU to "sync-up" with the Zorro III bus, and the "termination lag" at the end of a Zorro III cycle (the time between the Zorro III cycle being over and the 68030 cycle actually ending) is reasonably small. NuBus is a synchronous bus, and for a CPU to talk to it, that CPU's cycle must be synchronized to the 10MHz NuBus clock. [2] Zorro III's answer to "burst" mode, officially called "Multiple Transfer Cycles", can result from a translation of the 68030's Cache Burst mode. On any bus that supports the concept of a "burst" mode, burst transfers are faster than single transfers. NuBus supports a "burst" concept, technically called a block transfer. But you can't translate 68030 Cache Burst into NuBus burst. As for EISA and MCA, it's hard to say. Both are synchronous, EISA using an 8MHz clock, MCA using a 20MHz clock (though they use the clocks differently). Both EISA and MCA use unmultiplexed address/data buses, as opposed to the fully multiplexed A-D NuBus and partially multiplexed A-D of Zorro III. This doesn't appear as a major factor in the basic specifications for any of these buses; all four have maximum theoretical bus speeds between 20 and 40 MB/s, at least in their current implementations. MCA, at least, has several variations. The PS/2 version is limited to 20MB/s. The PS6000 implementation sports a burst mode that lets it theoretically run at 40MB/s, and IBM has announced specifications, at least, to multiplex Address and Data in a 64 bit burst more, for 80MB/s, and next to change the drivers and double the bus clock frequency, for 160 MB/s. But those two modes don't exist yet. And the actual bus speeds are often far below the theoretical ones. All four 32 bit buses are very likely adequate for anything you need to plug into a 32 bit desktop computer. And speed isn't the only issue: complexity, cost, available board space and power, patent licensing, market, etc. all enter into the formula when you're designing a board for a machine. > -Joseph Hillenburg (Sultan of Asm) -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy Get that coffee outta my face, put a Margarita in its place!