The Howard J. Cohen collection , ranging in date from 1978 to 1990, consists of program listings, manuals, reference books, and directories. The bulk of the collection is made up of listings from companies that Cohen worked for during the 1970s and 1980s. Cohen retained these listings, which he either created himself or worked as a contributor on, to serve as a personal archive, with each set of listings generated as a project snapshot. About half of the program listings are from Oceanroutes, a company that provided weather routing services to vessels and predicted ocean wave statistics at fixed points, typically for offshore drilling. There is also a significant amount of material from Daisy Systems, which was an early electronic design automation company. There are small amounts of listings from ROLM and ENSCO. This portion of the collection contains three 8-inch floppy disks, which hold copies of source code and are stored with their related program listing. The rest of the collection consists of manuals and reference materials, most of which are related to IBM and Intel.
Biographical/Historical Note
Howard J. Cohen is a software engineer and developer who has worked as a consultant for over 30 years. He has experience in a broad spectrum of technologies, such as electronic design automation, ocean wave modeling, and bioinformatics and computational genomics. A native of New York, Cohen earned a BS in physics from the City College of New York in 1966 and a PhD in theoretical physics from Brandeis University in 1974. After graduating, Cohen worked for two years as a member of the technical staff at NASA Ames Research Center. He then worked as a program manager in research and development at Oceanroutes from 1976 to 1979, where he was a project leader for and designer of the company's East Coast, Australian, and "portable" spectral ocean wave models. In 1979, Cohen began working as a systems development consultant, first as a scientist at ENSCO from 1979 to 1982, where he was a lead investigator in applying passive underwater acoustic detection and estimation procedures to multi-sensor target localization. After leaving ENSCO, Cohen joined Daisy Systems as a senior software engineer and group leader, where he created tools for the automatic and interactive layout of semi-custom and full custom VLSI chips until 1987. After Daisy Systems, Cohen continued to work as a consultant and occasionally in full time engineering and management positions for Silicon Valley businesses, such as Nortel Networks, SurroMed, Incyte Genomics, BomDiver, Canon Information Systems, WeatherNews, PetroSoft, LSI Logic, Aspect Development, Abbott Laboratories, ETAK, Engineering DataXpress, Nikon Precision, Diasonics, Rolm, Mindcraft, Integrated CMOS Systems, Tibco Inc., and Zycad Corporation. Also, he has taught courses at several San Francisco Bay Area colleges, including San Jose State University, Foothill College, College of San Mateo, and City College of San Francisco. In addition to his software development consulting, Cohen has served as an expert witness since 1997, specializing in software intellectual property litigation.
The documentation relates to VIRGA (a generic non-relational database system used in the Daisy chip layout products. The data array (GateMaster) version was called GDB; the standard cell (CellMaster) version was called SAM. PDQ provided quick access to the VIRGA database files.
This folder contains several version of the documentation describing OceanRoutes models and model buidling software. This item is a rebuild of the original 1976 North Sea model using the Makemodel system. (The 1976 North Sea model was mostly created manuall). The code NS, which appears on most of the listings, stands for North Sea.
PLACE was a system for constructive initial placement of components on a gate array, a type of ASIC with fixed locations for components. It used an iterative force relaxation technique, with "forces" binding components that were connected via sets, and other "forces" to spread the components out over the chip's area.
The projects include a vessel simulator and plotting experiments (for example, plotting spectral ocean wave data in a 3D perspective at a given point and time).
PLACE (auto/interactive placement) was a system for constructive initial placement of components on a gate array, a type of ASIC with fixed locations for components. It used an iterative force relaxation technique, with "forces" binding components that were connected via sets, and other "forces" to spread the components out over the chip's area.
Part of GateMaster, PIM was a placement improvement system. It could start with any placement of components on the gate array (including a fast, stupid algorithm implemented here to seed the process), and do swaps of components to improve one of several possible metrics of routability. Later versions also included a simulated annealing algorithm that ran as a slave on a hardware accelerator of Daisy's design (the MegaLogician).
MPR (Mega Place and Route) is the interface between PIM and the simulated annealing code running in the MegaLogician hardware accelerator. PIM handles the interface and UI, downloads data and uploads results.
In addition to the program listing, this folder includes two 8-inch floppy disks that appear to hold source code backups.
EDS (Elite Database System) consists of a variety of subsystems for the ChipMaster product (also called Elite), and editor for full custom VLSI layout creation.
These listings are related to DRC/ERC (Design Rule Checker and Electrical Rules Checker) for ChipMaster, a full custom VLSI design tool. The system includes IPS (an electrical rules checker), IER (an error handling subsystem), NEX (a net extractor), TIO (a trapezoid manager), CAN (a connectivity analyzer), and DRC (a design rule checker).
The project involved listening to sound at two locations in the ocean and then doing cross-coherence calculations (compensating for relative differences in Doppler shift) to try to determine real sources in the ocean (ships, submarines) versus random noise. One issue was the use of appropriate statistics to call a "hit," and the M-out-of-N code was an exploration of this statistical analysis.
This item is Fortran testing code and code for having an IEC plasma display device emulate a Calcomp plotter. It is likely from the time when Howard Cohen consulted at ROLM.
