Overview
What this challenge is about.
Implement Vyukov's bounded MPMC queue in C++20 using atomic operations with explicit memory orderings (relaxed where safe, acquire/release on the enqueue/dequeue handshake). Use cache-line padding to avoid false sharing on producer/consumer sequence counters. Validate correctness: TSan-clean over 100 million ops, and a CDSChecker (or Loom-style) exhaustive interleaving test on a smaller queue size. Benchmark on a 2-socket Intel Ice Lake-SP (24 cores per socket, isolcpus + tuned cpufreq governor): single-producer-single-consumer, 4P-1C, 1P-4C, 8P-8C; report mean + p99 latency per enqueue/dequeue and overall throughput vs. moodycamel and vs. boost::lockfree::queue. Deliver the queue source + tests, the memory-ordering proof sketch, the benchmark dataset, and a 7-page write-up for the trading-tech lead.
The Brief
What you'll do, and what you'll demonstrate.
Implement a lock-free Vyukov MPMC queue with provable correctness and beat moodycamel by >30 percent on the firm's trading-tech benchmark scenarios.
Earning criteria — what you'll demonstrate
- Implement a bounded lock-free MPMC queue with correct memory orderings
- Use TSan + interleaving-exploration tools to validate concurrency
- Pad data structures to eliminate false sharing on hot fields
- Benchmark with statistical rigor against industry-standard alternatives
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Career mappings coming soon.