Jane C. Blake,
Managing Editor

In 1992, DIGITAL announced the fastest 64-bit RISC microprocessor, the Alpha, with a clock rate of 200 MHz. Today’s Alpha processor remains the leader in performance; the newest generation operates at 600 MHz, and the next generation will operate at greater than 1,000 MHz—gigahertz speed. With the industry’s most powerful processor in hand, DIGITAL’s engineers are working to apply Alpha in different areas of computing and effect optimal solutions to computing problems. Samples of that work are presented in this issue and include programming performance tools, the OpenVMS operating system for very large memory (VLM) applications, graphics adapters for workstations, and the DART network adapter for high-end systems.

Spike is a profile-directed performance tool for optimizing Alpha executables running on the Windows NT operating system. Designed specifically to improve the performance of large, call-intensive programs, such as commercial databases, CAD programs, compilers, and productivity tools, Spike has been shown to speed program execution by as much as 33 percent. Robert Cohn, Dave Goodwin, and Geoff Lowney describe Spike’s two components. The Optimizer modifies code layout to improve instruction cache behavior and performs hot-cold optimization to reduce the number of instructions executed on frequent paths through the program. The Optimization Environment collects, manages, and applies profile information transparently for the programmer.

An experimental Atom-based performance tool presented by Susanne Balle and Simon Steely provides programmers with an understanding of the access pattern behavior of their technical applications. The tool generates histograms for each memory reference in a program, thus allowing the programmer to spot bottlenecks. The authors step through an instructive case study in the use of the tool with Fortran programs, showing how different compiler switches affect the execution of a program algorithm.

The OpenVMS Alpha operating system version 7.1 extends its support for VLM applications. The design work discussed by Karen Noel and Nitin Karkhanis focused on increasing flexibility for VLM applications and on adding system management capabilities. Areas reviewed are the shared memory objects designed to improve application scaling on the system, shared page tables to reduce application start-up/shut-down times, and the physical memory reservation system to allow efficient application use of system components, namely the translation buffer.

DIGITAL’s PowerStorm series of graphics adapters for mid-range workstations provides exceptional performance on the DIGITAL UNIX and the Windows NT operating systems. Benj Lipchak, Tom Frisinger, Karen Bircsak, Keith Comeford, and Mike Rosenblum have written an informative tutorial about the PowerStorm adapter design that was shaped in large part by the existing competitive environment. Their discussion covers selected benchmarks and real-world performance experiences, the advantages and disadvantages in choosing a direct-rendering or an indirect-rendering scheme, and the ways in which the engineering team exploited the Alpha microprocessor’s exceptional floating-point speed.

DART is a 622-megabit-per-second network adapter that connects gigabit-class networks to gigabit-class I/O buses. It is designed to increase network throughput and decrease system overhead. Bob Walsh explains that the DART project, started in the late 1980s, anticipated the need to address fundamental memory bandwidth bottleneck issues from a system-level perspective. The main approach taken in the DART adapter is data copy avoidance, without requiring changes to system call semantics.

The upcoming Journal will be a special issue that features papers on programming languages and tools. Topics include C and Fortran parallelizing compilers, the C++ template facility, alias analysis algorithms, debuggers, and performance tools for software running on the Windows NT, UNIX, and OpenVMS operating systems.