Veda
High-performance Rust parallel runtime with adaptive work-stealing.
A parallel execution runtime in Rust featuring adaptive work-stealing scheduling, lock-free queues, and NUMA-aware thread affinity for compute-intensive pipelines.
Standard thread pools bottleneck on task synchronization and cross-core cache invalidation when running millions of micro-tasks across heterogeneous CPU topologies.
Systems programmers writing high-throughput simulation, rendering, and data transformation engines.
Custom Chase-Lev work-stealing deques combined with NUMA node affinity pinning to keep thread communication within local memory channels.
Verified System Capabilities
Architectural Decisions & Rationales
Lock-free deque implementation with atomic primitives
Rationale: Eliminates mutex lock contention in critical dispatch loops.
64-byte cache line padding on worker counters
Rationale: Prevents false sharing across multi-core CPU architectures.
Technical Trade-offs & Honest Limitations
- Optimized for compute-bound workloads; heavy I/O tasks should use Tokio or asynchronous runtimes instead.
- NUMA thread pinning requires appropriate OS permissions and topology discovery support in the host kernel.
Public repository in TIVerse organization (`github.com/TIVerse/veda-rs`) with Criterion benchmarks.
Related Research & Architectural Writing
Adaptive Work-Stealing Scheduling & NUMA Affinity in Rust Task Runtimes
Parallel task scheduling across modern multi-socket hardware encounters severe throughput degradation from cross-NUMA interconnect contention and false cache-line sharing. We present the design of Veda, a high-performance Rust parallel execution runtime implementing an adaptive work-stealing algorithm with NUMA node affinity pinning. Empirical benchmarks on 32-core and 64-core topologies demonstrate up to a 2.4x reduction in task dispatch latency compared to naive work-stealing.
High-Performance Concurrency in Rust: Lessons from Building Veda and Polaris
Practical lessons learned building low-latency parallel runtimes, lock-free work-stealing deques, and declarative procedural macro engines in production Rust.