TitleAdvanced Scientific Computer (ASC) printed circuit board artwork
Description"LAYER REVISION A" is printed on the bottom. This object was in a box with a label that reads "682919/20 ASC, Box 2 of 2, Layers 8-10 + Voltage layers". Another label on the box reads "682919/20 ASC, Box 1 of 2, Layers 1-7".
The Advanced Scientific Computer, or ASC, was a supercomputer architecture designed by Texas Instruments (TI) between 1966 and 1973. Key to the ASC's design was a single, high-speed shared memory. The memory system had 8 banks and 8 ports, each connected to an ASC subsystem. Access was through an eight way full crossbar switch. The switch allowed all 8 ports to simultaneous access memory. Each access produces an "octet" or eight 32 bit words.
The ASC consisted of two primary subsystems: the Peripheral Processor (PP) and the Central Processor (CP). The PP executed the operating system. The CP executed the user application programs.
The PP had 8 virtual processors. It had one set of execution hardware and 8 sets of registers. The registers were connected to the execution hardware in a rotating fashion similar to Seymour Cray's ground breaking CDC 6600.
The CP was a vector, pipelined processor. There were 48 thirty-two bit registers: 16 arithmetic, 16 base address and 16 vector registers. The CP had three kinds of subsystems: an Instruction Processor unit (IP), Memory Buffer (MB) units and Vector units. A MB and a Vector unit were paired to create a vector pipeline. The IP fetched an octet, broke it into single instructions and dispatched them. The MB performed the memory accesses for each instruction. Each vector pipeline unit had an Arithmetic unit and a vector processor. A scalar instruction simply went through the Arithmetic unit as in a traditional computer. A vector instruction went through the Arithmetic unit multiple times with addresses and operands modified by the vector processor.
The ASC could be configured with 1, 2 or 4 vector pipelines. Most were 2 pipe machines. The first prototype had one pipe. Only one 4 pipe system was built. Each vector pipeline had its own memory buffer unit. Each memory buffer unit had its own port to the shared memory. Since the Arithmetic unit was pipelined, it could produce a result every clock cycle. This arrangement could perform a massive amount of calculation. For example, the ASC Fortran compiler could break a "Do-Loop" into 4 vector instructions. In a 4 pipe machine, each vector instruction was sent to its own pipe and they would execute in parallel. The net result was 4 calculations from the "Do-Loop" every clock cycle.
The TI Advanced Scientific Computer was a supercomputer that combined a Peripheral Processor and a Central Processor along with software that could automatically produce vectorized code from unmodified source code.
ManufacturerTexas Instruments Incorporated (TI)
|Other number||682919/20||On label on box that contained object|
|Other number||ASSY 682920||Printed on bottom|
|Other number||AVC7||Printed in lower right corner|
|Other number||LAYER 1 PCMB||Printed on bottom|
|Other number||PWB 682919-/B/||Printed on bottom|