35 Years of Embedded Computing at Connect Tech

Back in the 1960s, the first integrated circuit was developed to reduce the size and weight of the Apollo Guidance Computer in the Apollo Command and Lunar Module, used to help collect real-time flight data. Fast forward five years and similar technology was used for the computer system used in the Minuteman I missile guidance system – recognized as the first mass-produced embedded system. This was the beginning of what would lead to giant leaps forward in compact computer and system capabilities.

In the 1970s and ‘80s, manufacturers such as IBM and Apple were fast developing the personal computer market.

Since the IBM PC was based upon open architecture, it spawned an entire industry of suppliers for expansion cards, peripheral devices, and software. One of these companies was Intel, whose big break came when IBM used its microcontroller to run real-time programs locally on IBM computers.

Connect Tech was founded as IBM PCs were skyrocketing to market domination and Intel’s x86 microcontroller was becoming widely adopted for new technology development. Networking of the day required computer stations to have multiple devices at a workstation; PCs needed to connect to a terminal in order to access the mainframe.

Our first commercially available product was the EMU78 – known as the Emulator.

The Emulator enabled PCs to directly access the mainframe, eliminating the need to house a terminal on your desk.

As the use cases for computers began expanding, the need to connect other devices and inject variable data for processing required more sophisticated peripherals to be developed. We extended our product line by designing Multi-Port Serial Adapters, a range of products designed to collect signals from sensors and other data acquisition devices, or connect the PC to modems and other serial devices.

The use of real-time computing outside of personal computers began expanding into other market verticals as technology requiring simplified functionality in comparison to computers became mainstream. In 1987, Ampro designed the first embedded computer board for the x86 microcontroller, a design that prompted what would later become the basis for the PC/104 standard. Embedded technology and its peripherals became a standard part of every product design, with CPUs and expansion boards becoming industry standards and readily available off the shelf.

With industry quickly adopting the open PC/104 standard, we designed multiple products and peripherals to help designers build flexible solutions within this standard. With multiple boards available for PC/104 designs, our product line expanded into other technologies such as Ethernet-to-serial.

In the early 2000s, we expanded into FPGA technology, which allowed users to customize programming for specific tasks. FPGAs played an important role within the history of embedded computing with their ability to create system partitions between hardware and software, allowing software development and functional capabilities to soar.

In the late 2000s, the industry standard Qseven for embedded hardware design began to gain momentum. While designing products for this standard, we developed the first board that allowed users to integrate PCI/104-Express technology within Qseven designs.

Nearing the 2000s, the first embedded systems based on Linux began appearing. Today, Linux is one of the most prevalent operating systems used for embedded computing design.

Moore’s Law is a commonly accepted theory where the number of transistors in an integrated circuit doubles approximately every two years. This is the driving theory that helps put a timeline to the price reduction of componentry, capacity for storage and memory, and advancements of data compression and quality for electronics.

Throughout the 2000s and ‘10s, software became more sophisticated, hardware components became smaller, and the battle for function vs size waged war. The emergence of smart phones gave everyone a computer in their pocket, and the concept of connected devices with real-time information became a reality.

The use of Computer on Modules (COM) allowed the function of computer processing to be done within tiny form factors. While these systems provide the memory and processing power for applications, they lack the ability to directly ingest information needed for applications.

Once COM technology hit the market, our focus then shifted to include designing and manufacturing carrier boards that would house the COM and break the bus out to standard peripheral connections.

In 2012, a team from University of Toronto won the ImageNet computer image recognition competition by designing a neural network and training it to recognize images by itself. The program beat out software written by several computer vision experts. The AI and deep learning breakthrough came when researchers discovered how deep neural networks could run compute-intensive, high-performance programs by utilizing GPUs during training.

We designed our first PC 104 GPU adapter in 2011, and in the years since have developed a wide array of GPU-centric embedded hardware systems. NVIDIA introduced the Jetson Platform for Autonomous Machines in 2014, and our GPU expertise led to NVIDIA selecting us among the first global hardware partners in 2016. With a full team of in-house Jetson Experts, Connect Tech is now an Elite NVIDIA Partner, and the largest embedded hardware provider in the Jetson ecosystem.

As the Internet of Things movement took shape, the expectation of real-time responsiveness showed constraints in physical networking or cloud computing. The need to process data and react in real-time required devices to run programs locally, commonly referred to as Edge Computing. The emergence of System on Modules – small systems that contain a GPU, CPU, and memory – allows for a staggering amount of information to be ingested and processed simultaneously.

In current times, in addition to an expansive catalog of board and adapter technology, we’ve developed a full range of Integrated Systems – giving customers the ability to utilize off-the-shelf hardware solutions that leverage our 35 years of computing experience. Combining expertise designing GPGPU and embedded Ethernet switch technology has led to us developing hardware solutions for the world’s most cutting-edge AI and deep learning applications. With such staggering developments achieved over the last 35 years, one can only imagine what’s next in the world of embedded computing.