The Speedy DDR2 controller is intended as an improvement on the Xilinx MIG controller for Virtex 5 FPGAs. Designed entirely from scratch on the ML505 development board, it achieves better performance at the same clock rate than the MIG controller while consuming comparable resources. The tight timing constraints imposed by high-speed DDR2 clash with the worst-case timing constraint style of FPGA design in a way that presents unique challenges. This paper discusses the primary design problems resulting from that paradox and contrasts approaches to their solutions. Performance then is compared between the Speedy DDR2 controller and the Xilinx MIG controller. The source code has been written to be more readable, maintainable, and modifiable than the MIG design and is freely downloadable from the Web.
This download installs the runtime and static checking tools for managed contracts along with Visual Studio integration. The tools enable runtime and static checking of managed contracts.
This is a software-development kit (SDK) for the creation of Windows Phone 7 (WP7) applications that leverage research services not yet available to the general public. The primary goal of this SDK is to support the efforts of Project Hawaii, a student-focused initiative for exploration of how cloud-based services can be used to enhance the WP7 experience.
This library provides highly optimized versions of primitive functions such as repmat, set intersection, and gammaln. It provides efficient random number generators and evaluation of common probability densities. It provides routines for counting floating-point operations (FLOPS), useful for benchmarking algorithms. There are also some useful utilities such as filename globbing and parsing of variable-length argument lists.
Z3 is an SMT solver that combines several theory solvers into a combined framework. It can be used to prove theorems and find counter-examples for non-theorems.
A parallel BLAST engine that runs on the Windows Azure cloud fabric, NCBI BLAST on Windows Azure can scale up to hundreds of compute nodes. This cloud-based implementation of the National Center for Biotechnology Information (NCBI) Basic Local Alignment Search Tool (BLAST) searches all available sequence databases for similarities between a protein or DNA query and known sequences. BLAST on Windows Azure enables researchers to take advantage of the scalability of the Windows Azure platform to execute BLAST jobs on demand in the cloud.
SCARF is a tool kit for recognizing speech with segmental conditional random fields. It supports a broad spectrum of research into segmental features and detection-based speech recognition, and it is intended to support core academic research in this area.
This download is provided for the purpose of the Speller Challenge. This is a development dataset based on the publicly available TREC queries (2008 Million Query Track). Queries are annotated by using the same guidelines and processes as in the creation of the Bing Test Dataset.
SLAyer is an automatic, separation-logic-based memory safety checker. It checks that its input C code doesn't deference dangling pointers, do double frees, nor leak memory.
Zentity 2.0 is a semantically enabled repository platform that provides a suite of building blocks, tools, and services to create and maintain an organization’s digital-library ecosystem. The research lifecycle generates a wide variety of outputs that are important to store and track as part of the ongoing research process. Zentity maximizes the value of this output by helping researchers access, analyze, and unlock the previously hidden structure and relationships among data elements.
This add-in for Word enables authors and editors to save Word files in the National Library of Medicine's Document Type Definitions format, which is used for publishing and archiving articles and books. In addition, the add-in enables more metadata to be captured and stored at the authoring stage and enables semantic information to be preserved through the publishing process, which is essential for enabling search and semantic analysis once the articles are archived within information repositories.
The "extensible MIPS" is a dynamically extensible processor that offers application-specific hardware in a general-purpose, multiuser system environment. It enables multiple secure extensions to load dynamically and extend the core instruction set of the microprocessor. Extended instructions can speed application programs dramatically just by patching their binaries. The eMIPS prototype is built out of a Xilinx FPGA using the XUP board. Source and software tests are included in this download.
Giano is a framework for the full-system simulation of arbitrary computer systems, with special emphasis on the hardware-software co-development of system software and Real-Time embedded applications. Giano includes both software models for CPU, I/O, busses and memories, and HDL simulators. Full source and a number of full-systems examples are included.
This download contains the ISO CD image for installing NetBSD 4.0.1 on an eMIPS processor system. The download works for the Xilinx XUP board and the Giano simulator.
A customizable image viewer for use with the SenseCam or the Vicon Revue, providing facilities for reviewing a sequence of SenseCam images, as well as for tagging and exporting them.
Pivot is an experimental application for exploring large data sets with smooth visual interactions. The application originally was released by Microsoft Live Labs in October 2009, and it is being re-released by Microsoft Research to enable the research community to continue to use it for experiments. If you have Internet Explorer 9 installed, disable GPU rendering in Internet Explorer to enable Pivot to work correctly. The Pivot collection home page points to content no longer available, but Pivot still can be used for viewing user-created local or web collections. This standalone version of Pivot is unsupported and might stop functioning properly in the future.
Infer.NET is a .NET framework for machine learning. It provides state-of-the-art message-passing algorithms and statistical routines for performing Bayesian inference. It has applications in a wide variety of domains, including information retrieval, bioinformatics, epidemiology, vision, and many others.
Diff-IE is a prototype Internet Explorer add-on that highlights the changes to a page since the last time you visited it and enables you to view and compare previously cached versions of the page.
This package contains the F# and F7 source files to aid in the verification of a distributed key-management system. This new component implements a data-protection API for groups of clients. To enable long-term data protection, it supports cryptographic agility so cryptography algorithms and policies can evolve for protecting fresh data while preserving access to old data. To verify the security of our design and production code, written in C#, we write a reference implementation in F#. Formally, we verify our F# code against a logical cryptographic model using F7, a refinement type checker coupled with a model checker. Experimentally, we test that the corresponding C# and F# code fragments are interchangeable.
The Microsoft Research Software Radio (Sora) is a fully programmable software radio platform based on the commodity multicore CPU in a host PC. With Sora, researchers and engineers quickly can prototype new, high-speed wireless physical and media-access-control layers with a minimum of effort. This package includes the Sora SDK, the 802.11b sample, and the 802.11a sample.
Holmes is an automated debugging and failure-analysis tool that uses statistical analysis of code-coverage data to identify the most likely cause of test failures. Holmes uses a custom data-collection tool to amass detailed path coverage during testing. The tool analyzes this coverage data to find paths that strongly correlate with failure. Holmes integrates seamlessly with all test workflows in Visual Studio. Developers can use Holmes to select and analyze a set of test runs, to navigate to the most likely causes of test failures in source code, to set breakpoints and to re-run specific failing tests.
This data consists of about 120K sentences collected during the summer of 2010. Workers on Mechanical Turk were paid to watch a short video snippet and then summarize the action in a single sentence. The result is a set of roughly parallel descriptions of more than 2,000 video snippets. Because the workers were urged to complete the task in the language of their choice, both paraphrase and bilingual alternations are captured in the data. We expect this data to be useful for training and testing translation and paraphrase algorithms. A paper describing how the data was created and used is in progress.
Sho is a playground for data: a set of libraries and an interactive environment for rapid prototyping, data analysis, and visualization. It consists of libraries for math and visualization that can be accessed from any .NET language and an interactive console and other modules built in IronPython for data manipulation, visualization, and analysis of arbitrary .NET objects and data structures.
This tool finds data races in multithreaded code. It uses a combination of code and data break points to find data races effectively with little or no runtime overhead.
The primary function of this add-in is to add a couple of buttons to the Outlook ribbon to prevent people from replying to all the recipients of your message or forwarding it. The add-in uses a facility built into Outlook and Exchange that is more lightweight than information-rights management but is not exposed in the existing UI. The add-in also includes a check for common email errors, such as omitting attachments or subject lines.
This Visual Studio Add-in and bare-metal environment enables the development of bare-metal applications for the Intel Single-chip Cloud Computer (SCC) research platform.
Pex (Program EXploration) is a white-box test generation tool. Given a hand-written parameterized unit test, Pex analyzes the code to determine relevant test inputs fully automatically. The result is a traditional unit test suite with high code coverage.
This scalable language-model tool is used to build language models from large amounts of data. It supports modified absolute discounting and Kneser-Ney smoothing. The tool has been used successfully to build a seven-gram language model on 40 billion words within eight hours.
Loading…
Reverse order
