Firefly: a multiprocessor workstation

Abstract

Firefly is a shared-memory multiprocessor workstation that contains from one to seven MicroVAX 78032 processors, each with a floating point unit and a sixteen kilobyte cache. The caches are coherent, so that all processors see a consistent view of main memory. A system may contain from four to sixteen megabytes of storage. Input-output is done via a standard DEC QBus. Input-output devices are an Ethernet controller, fixed disks, and a monochrome 1024 x 768 display with keyboard and mouse. Optional hardware includes a high resolution color display and a controller for high capacity disks. Figure 1 is a system block diagram.The Firefly runs a software system that emulates the Ultrix system call interface. It also supports medium- and coarse-grained multiprocessing through multiple threads of control in a single address space. Communications are implemented uniformly through the use of remote procedure calls.This paper describes the goals, architecture, implementation and performance analysis of the Firefly. It then presents some measurements of hardware performance, and discusses the degree to which SRC has been successful in producing software to take advantage of multiprocessing.

Scheduling support for concurrency and parallelism in the Mach operating system

“…The local run queue is examined first to give absolute preference to bound threads over unbound threads. This avoids bottlenecks by maximizing throughput of the unparallelized Unix compatibility code 8 , and improves the performance of processor allocation by causing it to preempt other operations. If the local queue is empty, then the global run queue is examined.…”

“…7 Most of this code has been parallelized, see [5] for details. 8 See [6] for an extensive study of this area…”

“…This can impact performance if the primitives are not a good match to the programming language or model. The Topaz system uses this approach, but specialization is not an issue in that environment because most programming is done in Modula-2+ [8].…”

Architectural mechanisms for explicit communication in shared memory multiprocessors

“…RCupd: This memory system uses RC memory model, a simple write-update protocol not unlike the one used in Firefly [37] for state transitions, and no explicit communication primitives. From the point of view of the processor, writes are handled exactly similarly as in RCinv.…”

Multi-model parallel programming in psyche

“…Other systems in this class include Amoeba [21], Chorus [2], Topaz [27], and V [11]. To facilitate parallelism within applications, these systems allow more than one kernel-supported process to run in one address space.…”