On 4/22/2016 8:54 PM, Jecel Assumpcao Jr. wrote:
Eric Smith mentioned:
[2901 A, B, and C, CMOS versions]
[2903 and 29203]
[Intel 3001 and 3002]
[MMI 5701/6701]
[Motorola MC10800]
I'd add the Texas Instruments SN74S481, SN54LS481 and SN74LS481 TTL 4
bit slices. The Schottky version had a 90ns clock cycle and the low
power versions 120ns. These were 48 pins chips and didn't have an
internal register bank like the 2901. The idea was that you implemented
a memory to memory architecture like the TMS9900.
The TI SBP0400A and SBP0401A were I2L 4 bit slices. The 400 had an
internal pipeline register while the 401 was designed for external
pipelining. ALU operations took 240ns at 200mW. It had an 8 register
bank besides the working register.
All I know about these is what I read in "The Bipolar Micromputer
Components Data Book", December 1977 edition. I have no idea if these
chips were actually shipped or if they were used in any product.
-- Jecel
I think the problem was memory at the time. What was the use having
a FAST bitslice machine, but real memory at the time was SMALL and SLOw.
Then if you could get a nice machine, you could never get out of the
School or Lab home with you.
The Lilith computer comes to mind here.
Ben.
PS:
http://www.projectoberon.com/ A computer design A few generations
after Lilith. Here a FPGA risc cpu and fast static memory.