parallel-block-stm

Parallel speculative transaction execution engine implementing the Block-STM algorithm (Gelashvili et al., PPoPP 2023) in C++.

CMU 15-418/618 Final Project, Spring 2026

Team: Hua-Yo Wu (huayow), Pi-Jung Chang (pijungc)

Project Page

---

Overview

Given a block of N ordered transactions, the engine executes them speculatively in parallel across threads, detects read/write conflicts, aborts and re-executes conflicting transactions, and guarantees the final state matches sequential execution in the given order.

Two pieces do most of the work:

• MVMemory - a multi-version store. Each key keeps a version chain (lock-free sorted linked list, CAS insert, exponential backoff under contention). Readers find the highest write from a tx with idx less than theirs.
• Scheduler - a collaborative scheduler with two shared atomic counters (execution_idx, validation_idx). Any thread can grab any task; aborted tx's become ESTIMATE markers so readers suspend instead of reading stale data.

Transaction status

Each (txn_idx, incarnation) transitions through:

• READY_TO_EXECUTE(i) -> EXECUTING(i) via try_incarnate()
• EXECUTING(i) -> EXECUTED(i) via finish_execution()
• EXECUTING(i) -> ABORTING(i) via add_dependency() when a read hits an ESTIMATE
• EXECUTED(i) -> ABORTING(i) via try_validation_abort() (first caller wins)
• ABORTING(i) -> READY_TO_EXECUTE(i+1) via set_ready_status()

Building

mkdir build && cd build
cmake ..
make -j$(nproc)

Dependencies

• C++20 compiler (g++ 10+ / clang 13+ / AppleClang)
• CMake 3.20+

Usage

./build/test/test_blockstm
./build/bench/bench_scaling --threads 8 --block-size 10000 --accounts 1000

References

• Gelashvili, Spiegelman, Xiang, et al. Block-STM: Scaling Blockchain Execution by Turning Ordering Curse to a Performance Blessing. PPoPP, 2023.
• Shavit, Touitou. Software Transactional Memory. PODC, 1995.