garret@uw-june (Garret F. Swart) (11/29/83)
I haven't read anything about the Micro Vax I just announced by Digital. They gave their second product demo here last week so I thought I would throw in some information about it onto the net. The Micro Vax I is a LSI implementation of the Micro Vax architecture, a subset of the full Vax architecture. The major missing instructions are: Decimal arithmetic, edit instruction, odd floating point formats, PDP-11 emulation mode. The design is split into three parts, a micro programmed CPU with an LSI data path chip (containing a barrel shifter), a micro programmed Memory Manager with a cache, look aside buffer etc. for a full implementation of the VAX virtual memory, and a Q-bus controller. It fits on two quarter size boards and fits in the LSI-11 case and interfaces to all its periferals. It is currently running VMS and a limping version of Unix 4.1BSD. They are currently working on tuning the Unix version. As was expected they have unbundled the software from the machine and allow one to buy what one wants as needed. The entire software set for this machine is expected to sell for $4000. By the way, the machine with 2Megs of main memory and a 25Meg hard disk and controllers for 5 terminals is expected to sell for $10000, sans software. Ethernet boards and everything else made for the LSI-11 will work on the machine. They are currently working on a bitmap terminal that will plug right onto the Q-Bus as part of their "Vax Station" concept. They are not giving out bench mark data, but they suggest that the machine is as fast as a 730 for most instructions, and twice as fast for floating point. The design team contains some of the brightest people at digital who are working near Seattle at DEC-West. The design seems to be as risk free as is possible. All the control logic is in SSI, and MSI and in 17 PLA's, with only the most stable part of the design, the data path, in LSI. The data path chip was, by the way, designed correctly the first time with help from Carver Mead. The bottle neck on the machine is the 22 bit Q-bus which was used so that the Micro Vax I could be sold as a complete system as soon as possible. DEC West is still full of bright ideas and I don't think we have long to wait for an improved Micro Vax II. Garret Swart University of Washington (I have no relationship with DEC)
garret@uw-june (Garret F. Swart) (11/29/83)
From garret@uw-june (Garret F. Swart) Wed Dec 31 16:00:00 1969 Relay-Version: version B 2.10 5/3/83; site uw-june Posting-Version: version B 2.10 5/3/83; site uw-june Path: uw-june!garret From: garret@uw-june (Garret F. Swart) Newsgroups: net.micro Subject: Micro Vax I Message-ID: <775@uw-june> Date: Tue, 29-Nov-83 11:17:30 PST Article-I.D.: uw-june.775 Posted: Tue Nov 29 11:17:30 1983 Date-Received: Tue, 29-Nov-83 11:17:30 PST Organization: U. Washington, Computer Sci Lines: 45 I haven't read anything about the Micro Vax I just announced by Digital. They gave their second product demo here last week so I thought I would throw in some information about it onto the net. The Micro Vax I is a LSI implementation of the Micro Vax architecture, a subset of the full Vax architecture. The major missing instructions are: Decimal arithmetic, edit instruction, odd floating point formats, PDP-11 emulation mode. The design is split into three parts, a micro programmed CPU with an LSI data path chip (containing a barrel shifter), a micro programmed Memory Manager with a cache, look aside buffer etc. for a full implementation of the VAX virtual memory, and a Q-bus controller. It fits on two quarter size boards and fits in the LSI-11 case and interfaces to all its peripherals. It is currently running VMS and a limping version of Unix 4.1BSD. They are currently working on tuning the Unix version. As was expected they have unbundled the software from the machine and allow one to buy what one wants as needed. The entire software set for this machine is expected to sell for $4000. By the way, the machine with 2Megs of main memory and a 25Meg hard disk and controllers for 5 terminals is expected to sell for $10000, sans software. Ethernet boards and everything else made for the LSI-11 will work on the machine. They are currently working on a bitmap terminal that will plug right onto the Q-Bus as part of their "Vax Station" concept. They are not giving out bench mark data, but they suggest that the machine is as fast as a 730 for most instructions, and twice as fast for floating point. The design team contains some of the brightest people at digital who are working near Seattle at DEC-West. The design seems to be as risk free as is possible. All the control logic is in SSI, and MSI and in 17 PLA's, with only the most stable part of the design, the data path, in LSI. The data path chip was, by the way, designed correctly the first time with help from Carver Mead. The bottle neck on the machine is the 22 bit Q-bus which was used so that the Micro Vax I could be sold as a complete system as soon as possible. DEC West is still full of bright ideas and I don't think we have long to wait for an improved Micro Vax II. Garret Swart University of Washington (I have no relationship with DEC)
minow@decvax.UUCP (Martin Minow) (12/01/83)
Garret Swart's interesting article on the Micro Vax contained one statement that might be confusing to readers unfamiliar with the LSI-11 Q-bus -- that it is a "22-bit bus" The 22 bits refers to the number of address bits that the bus supports. The bus may be a bottleneck because address and data are carried on the same signal lines, rather than on separate lines as on the Unibus. (This is from memory -- so don't blame Dec if I'm wrong.) The Q-bus and Q-bus periferals are much less expensive to manufacture than the Unibus, a savings that is reflected in the price of the product. Martin Minow decvax!minow