Abstract: An interface circuit (10) for coupling a parallel data device (12) to a serial data channel (14, 16) over which Manchester-type codes are transmitted. In the interface circuit, an efficient and reliable Manchester decoder (22), comprising a flip-flop (50), an exclusive-or gate (52), and at least one delay line (58A or 58B) separates the data and clocking signals. The serial data signals are clocked into a serial register (30) under control of the external clocking signals from the channel. A carrier detector (24) enables the serial register only when valid information signals are present. A parallel data register (40) receives in parallel the data from the serial data register. To get in phase the external clocking signals with the internal clock source, an internal clock synchronizing circuit (34, 42) recycles the internal clock source upon the occurrence of a synchronizing character that is transmitted over the serial data channel.

Abstract: A multiprocessor data processing system, the processors (30) and input/output devices (32) of which share a common control unit (CCU 10) that includes a write-through cache memory (20), a memory management circuit (22) and an address translation circuit (24). The data processing system further includes random access memory (28) and a secondary storage facility (40, 42, 68, 70). The processors (30) and the input/output devices (32) use the memory management circuit (22), the address translation circuit (24) and the cache memory (20) in an ordered pipelined sequence. When a read command "misses" the cache memory (20), the CCU accesses the memory modules (28) for allocating its cache memory (20) and for returning read data to the processors (30) or input/output devices (32).

Abstract: A cached multiprocessor system operates in an ordered pipeline timing sequence in which the time slot for use of the cache is made long enough to permit only one cache access. Further, the time slot for data transfers to and from the processors succeeds the time slot for accessing the cache. The sequence is optimized for transactions that require only one cache access, e.g., read operations that hit the cache. Transactions that require two cache accesses must complete the second cache access during a later available pipeline sequence. A processor indexed random access memory specifies when any given processor has a write operation outstanding for a location in the cache. This prevents the processor from reading the location before the write operation is completed.

Abstract: A memory arrangement for a digital data processing system that includes a high-speed associative memory unit and a random access back-up unit. The associative memory unit contains a multiple location address memory and a multiple location data memory wherein there is a correspondence between each address location and a data location. Each time a central processor initiates a reading operation, it issues an address to define a data location. If the associative memory unit contains that address at a location in its address memory, it performs a reading memory cycle and transfers data from the corresponding location in the data memory directly to the central processor. If the data is not available in the associative memory unit during a reading operation, or if the central processor is transferring data to the address location during a writing operation, the associative memory unit causes the back-up unit to perform a corresponding memory cycle.