The Working Overloaded Linux Kernel (WOLK) project provides stable and development kernels for either server or testing purposes. These kernels provide a server kernel and also a service for developers and end users who can't be up-to-date with the latest kernels/patches but want to test new kernel features. Patches may be added upon request.
chpox provides transparent checkpointing and restarting of processes on Linux clusters. It was originally designed for recovering tasks that have a long execution time (i.e. numerical simulations) in case of system crashes, power failures, etc. It may work with openMosix, is SMP safe, does not require program recompiling/relinking, and supports virtual memory, regular open files, pipes, Unix domain sockets, current directory, and child processes.
openMosix is a a set of extensions to the standard Linux kernel allowing you to build a cluster of out of off-the-shelf PC hardware. openMosix scales perfectly up to thousands of nodes. You do not need to modify your applications to benefit from your cluster (unlike PVM, MPI, Linda, etc.). Processes in openMosix migrate transparently between nodes and the cluster will always auto-balance.
joeq is a Java 2 (JDK 1.3 and 1.4) compatible virtual machine. It is unique in that it is entirely implemented in Java, leading to greater reliability, portability, maintainability, and efficiency. It is also language-independent, so code from any supported language can be seamlessly compiled, linked, and executed dynamically.
MakeNG is a system of makefiles that, in conjuncture with a patch to GNU Make, provides a concise, extensible, logically organized and easily modified system of makefiles to build any source tree or combination of components you can envision. MakeNG sports a fully documented extensible API on top of the classic GNU Make syntax (which is great for backwards compatibility).
DSP implements several low-level digital signal processing (DSP) primitives accelerated, when available, by the Intel/AMD SIMD instruction sets MMX/SSE/SSE2. Portable C versions are provided for compatibility on non-IA32 machines. Routines are provided to compute 16-bit integer dot products (FIR filtering, correlation); sum-of-squares (signal energy measurement), and peak sample detection (for automatic gain control, etc.).
SIMD Viterbi Decoder provides library functions to decode certain popular error correction codes. This version supports two codes: a rate 1/2, constraint length 7 (r=1/2, k=7) code, and an r=1/2, k=9 code. Four implementations of each decoder are provided. One is in portable C and should run in any GNU C environment. The other three use the IA32 SIMD (single instruction, multiple data) instruction sets: MMX, SSE, and SSE2. The SSE version of the k=7 decoder executes at ~9 megabits/sec on a 1GHz Pentium-III.