The Allen School provides and supports a wide variety of state-of-the-art computing facilities for research and instructional use, housed in two buildings; the Paul G. Allen Center, commissioned in 2003, and the Bill and Melinda Gates Center for Computer Science & Engineering, commissioned in January of 2019. Each of these remarkable buildings has more than 80,000 sq feet of net assignable space. Both complexes provide significant laboratory space to support new and ongoing research initiatives and interdisciplinary efforts. Both buildings were designed to facilitate interaction—both purposeful and serendipitous. The design provides for plentiful breakout areas, reservable meeting spaces big and small and seminar rooms to nurture a sense of community and to foster collaborations among students, faculty, staff and visitors.
The Computer Science Laboratory Group (CSE IT) coordinates the acquisition, maintenance, and operation of the computing equipment and network services. General-purpose research computing is provided by over 800 Windows and Unix-based workstations located in laboratories, machine rooms and offices. The back-end infrastructure is comprised of general-purpose compute, file, web, mail and print servers, operating as a well-integrated Linux and Windows environment. In addition, this group supports around 250 physical project funded server systems and clusters hosted in two 1,200 sq ft access-controlled datacenters and 4 smaller research data rooms, all of which are climate-controlled and monitored centrally. The School networking utilizes 1 and 10 gigabit Ethernet connections to servers and desktop machines, and a dual-band wireless network provides 802.11b/g/n connectivity between and throughout each building and in surrounding exterior areas.
Research in computer systems (including architecture, networking, operating systems, and distributed systems) involves a wide and constantly updated variety of hardware, software, and networks. Current hardware includes high-performance Intel multicore platforms, a 200-node Intel cluster with several tens of terabytes of networked storage, a networking testbed cluster, and PC workstations. Our facilities include Linux, FreeBSD, and Windows systems, and our clusters enjoy 1 and 10 gigabit switched Ethernet connectivity. In addition, the Systems lab provides a common workspace for operating systems, networking, and architecture students, and features Windows workstations, a video projector, and floor-to-ceiling whiteboards.
Research in VLSI, digital hardware, and embedded systems is supported by a set of PC workstations and multiprocessor compute servers. A large collection of both commercial and university computer-aided design tools form the core of the design environment providing capabilities for the design of CMOS VLSI chips, FPGA and microprocessor-based systems, and printed-circuit boards. A variety of specialized equipment for the prototyping, debugging, and testing of microelectronic systems is also available and is housed within the Hardware and
Embedded Systems Research Laboratory. These resources are utilized by research projects involved in the design of configurable computing architectures, devices to support ubiquitous and invisible computing, embedded systems, neurally-inspired computing and learning devices. Additional equipment and facilities are available in the Embedded Capstone Laboratory, which is used for undergraduate courses including Advanced Logic Design and Embedded System Design.
Research in graphics, image processing, and user interfaces, centered in the Graphics and Imaging Laboratories, utilizes a set of high-end graphics workstations, two multiple-node compute clusters with GPU processing, and a variety of special-purpose devices, including Microsoft Kinect and Leap Motion controllers, Oculus Rift VR hardware, digital cameras (still and video), a desktop Cyberware 3D laser scanner, video projectors for shape capture, and rotational and translational motion control platforms. Most of the lighting and imaging hardware resides in a special-purpose scanning and imaging laboratory, which is ideal for experiments that require controlled illumination. The main lab spaces contain an array of workstations and an audio/video hardware suite with non-linear digital video editing capabilities. The workstations in the main labs are also used as development stations for experimental teaching software in graphics and vision.
Research in robotics is carried out across five laboratories: Robotics and State Estimation, Sensor Systems, Neural Systems, Movement Control and Human-Centered Robotics. A diverse range of robotic platforms are available in these laboratories, including state-of-the-are research platforms such as PR2 (Hobbs and Rosie), Baxter, Nao, TurtleBot, Darwin-OP, and Fujitsu HOAP-3; as well as custom platforms built in-house, such as Gambit, Marvin, and Android. All robots utilize wireless networking to communicate with each other and the lab PCs running Linux. The Gates Center (CSE2) building includes a new 3,000 square foot robotics lab to allow for expansion and seamless collaboration in the robotics groups.
Research in artificial intelligence, data mining, machine learning, natural language processing, and satisfiability/experimental algorithms is supported by many high-end workstations and server class compute resources for training models. Many of these workstation systems support 3-4 GPU cards, with servers that will support up to 8 cards. Projects also leverage cloud resources and a centralized University HPC cluster for larger scaled experiments. Baxter, Nao, TurtleBot, Darwin-OP, and Fujitsu HOAP-3; as well as custom platforms built in-house, such as Gambit, Marvin, and Android. All robots utilize wireless networking to communicate with each other and the lab PCs running Linux. The Gates Center (CSE2) building includes a new 3,000 square foot robotics lab to allow for expansion and seamless collaboration in the robotics groups.
Research in data management is supported by a combination of laptops, desktops, and a machine-cluster all running a suite of software systems. The current hardware configuration for the cluster includes 20 high-performance, Intel mulitcore servers with several tens of terabytes of storage and hundreds of gigabytes of RAM in total. The machines are configured with either Windows or Linux and run several state-of-the art database management systems including SQL Server, Spark, Hadoop, and Myria. In addition, the Database lab provides a common workspace for students, and features Windows and Linux workstations, a video projector, and floor-to-ceiling whiteboards.
Other research groups utilize equipment located in a set of new research laboratories that utilize various high performance servers or compute clusters. Additional information can be found in the web pages for individual research projects, at http://www.cs.washington.edu/research/.
Instructional computing is provided through laboratories and back-end services operated within the school. These include six general use laboratories with 150 PC workstations running Windows and CentOS Linux, and seating capacity for more than 275 students simultaneously. Additional back-end resources are provided by Intel PC-based compute, web, database, and file servers in an integrated Linux/Windows infrastructure. The school also operates five special-purpose laboratories containing approximately 100 PC workstations. To support digital system design courses, the Embedded Systems Project Laboratory contains 18 PC workstations for design entry and simulation along with Tektronix logic analyzers, digital oscilloscopes and other test equipment. The Animation Lab includes more than two dozen workstations, digital video production equipment, and a 3D printer, and is used for teaching interdisciplinary courses in computer animation. Three Capstone labs can host up to 60 PC workstations, and are routinely used to teach capstone courses in accessibility in technology, game design, operating systems, robotics, computer security, virtual reality and other courses requiring specialized equipment and/or dedicated access. These labs different systems and software at different times, depending on each courses needs.