Great to see Pete again, we were classmates on the Computer Science course at Manchester University. DEC played a big part in our learning at Manchester, and in my early career until mid 1980's. The DEC PDP11, usually running under RSX11M or RT11 operating systems, enabled new possibilities in industry for multiple reasons including cost vs. performance, its relatively modest physical size and power requirements, open architectures for which 3rd parties could develop specialist hardware and software, and above all, the ease and speed of integration with prototype and rapid development equipment for real-time applications. It did have contemporary competitors in the 16-bit minicomputer environment, but for some years DEC were dominant in the market owing to its unique mix of these features. To give an example - in the late 1970's there was a realisation that small computers might then be powerful enough to deal with signal processing for HF radar systems (over-the-horizon) using emerging techniques such as digital beam steering and phased array receivers. The processing requirements for these systems were demanding owing to the physics of radio propagation at HF frequencies. Finding a target is more of a challenges that at higher radar frequencies, but the over-the-horizon advantage achieves greater range and early warning capability. At Marconi Radar Systems Limitedin the 1970s, and working with scientists at the Marconi Research Laboratories just outside Chelmsford, we undertook a project or the RAF to look into what might be possible. After some research into the state-of-the-art in signal processing hardware in the USA we identified that an attached array processor, the AP120B produced by Floating Point Systems in Beaverton, Oregon, could be readily hosted by a PDP11/34 to create a suitable experimental system. With quite modest resources, and in a relatively short time, we were able to interface the PDP11 to the AP120B and establish a suitable software development and testing environment for the signal processing algorithms. We also interfaced the PDP11 to pair (phase and quad) of very high performance A-to-D converters used to generate inputs to the system, and a plan position indicator (PPI) where outputs could be displayed in real time. We also created a simple graphical output to record our results using DEC's 120-column dot matrix printer. The results of this work attracted wide interest, for example the Chief Scientist of GEC , Sir Eric Eastwood, visited our PortaKabin, located in a rabbit field, to see the output on the PPI and observe, in real time, how this changed with input conditions. So the PDP11 enabled experiments in signal processing techniques and helped to create a test environment for subsystems that might be used in a full-scale radar, and the exploration of system engineering trade-offs between them. Concurrent with this, Marconi and other companies were applying PDP11's to many real-time applications, or investigating the possibilities. This included other situations where the combination of a PDP11 host and attached AP120 processor provided a very cost effective and accessible solution, the architecture of the AP120B being highly optimised for signal processing calulations such as the FFT butterfly. The peak processing power of this combination under optimum conditions was 12 megaflops per second, using equipment in a single 19" rack and running off a 13A socket. Only a few years before equivalent computing capability would have required a vastly larger qualtity of equipment. This trajectory of rapidly increasing computing power and reducing costs has certainly continued throughout my (ongong) career, and at the age of almost 70 I look back with incredulity at the progress achieved. However, do sometimes think wistfully to my Manchester days when the subject of Computer Science was still just about small enough that an individual could develop basic learning and knowledge about most aspects of the subject!
Great to see Pete again, we were classmates on the Computer Science course at Manchester University. DEC played a big part in our learning at Manchester, and in my early career until mid 1980's. The DEC PDP11, usually running under RSX11M or RT11 operating systems, enabled new possibilities in industry for multiple reasons including cost vs. performance, its relatively modest physical size and power requirements, open architectures for which 3rd parties could develop specialist hardware and software, and above all, the ease and speed of integration with prototype and rapid development equipment for real-time applications.
It did have contemporary competitors in the 16-bit minicomputer environment, but for some years DEC were dominant in the market owing to its unique mix of these features.
To give an example - in the late 1970's there was a realisation that small computers might then be powerful enough to deal with signal processing for HF radar systems (over-the-horizon) using emerging techniques such as digital beam steering and phased array receivers. The processing requirements for these systems were demanding owing to the physics of radio propagation at HF frequencies. Finding a target is more of a challenges that at higher radar frequencies, but the over-the-horizon advantage achieves greater range and early warning capability.
At Marconi Radar Systems Limitedin the 1970s, and working with scientists at the Marconi Research Laboratories just outside Chelmsford, we undertook a project or the RAF to look into what might be possible. After some research into the state-of-the-art in signal processing hardware in the USA we identified that an attached array processor, the AP120B produced by Floating Point Systems in Beaverton, Oregon, could be readily hosted by a PDP11/34 to create a suitable experimental system. With quite modest resources, and in a relatively short time, we were able to interface the PDP11 to the AP120B and establish a suitable software development and testing environment for the signal processing algorithms. We also interfaced the PDP11 to pair (phase and quad) of very high performance A-to-D converters used to generate inputs to the system, and a plan position indicator (PPI) where outputs could be displayed in real time. We also created a simple graphical output to record our results using DEC's 120-column dot matrix printer. The results of this work attracted wide interest, for example the Chief Scientist of GEC , Sir Eric Eastwood, visited our PortaKabin, located in a rabbit field, to see the output on the PPI and observe, in real time, how this changed with input conditions. So the PDP11 enabled experiments in signal processing techniques and helped to create a test environment for subsystems that might be used in a full-scale radar, and the exploration of system engineering trade-offs between them.
Concurrent with this, Marconi and other companies were applying PDP11's to many real-time applications, or investigating the possibilities. This included other situations where the combination of a PDP11 host and attached AP120 processor provided a very cost effective and accessible solution, the architecture of the AP120B being highly optimised for signal processing calulations such as the FFT butterfly. The peak processing power of this combination under optimum conditions was 12 megaflops per second, using equipment in a single 19" rack and running off a 13A socket.
Only a few years before equivalent computing capability would have required a vastly larger qualtity of equipment. This trajectory of rapidly increasing computing power and reducing costs has certainly continued throughout my (ongong) career, and at the age of almost 70 I look back with incredulity at the progress achieved. However, do sometimes think wistfully to my Manchester days when the subject of Computer Science was still just about small enough that an individual could develop basic learning and knowledge about most aspects of the subject!
Thanks for you feedback and memories Philip.