Isurwars/Correlation / .agents/skills/concurrency
concurrency
VerifiedEnforce thread-safety rules for OpenMP and TBB loops, std::scoped_lock synchronization, atomic state management, and cache alignment in parallel structural calculators. Use when writing, debugging, or profiling multi-threaded algorithms.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 2.6 KB |
SKILL.md
View on GitHub---
name: concurrency
description: Enforce thread-safety rules for OpenMP and TBB loops, std::scoped_lock synchronization, atomic state management, and cache alignment in parallel structural calculators. Use when writing, debugging, or profiling multi-threaded algorithms.
---
# Concurrency & Parallel Patterns
This skill governs multi-threaded execution, OpenMP/TBB safety rules, and lock-free memory locality across Correlation calculators (`RDF`, `S(Q)`, `PAD`, `VACF`, `MSD`).
## 1. OpenMP Loop Safety Rules
### Rule 1: Thread-Private Iterators
Loop variables inside `#pragma omp parallel for` constructs must be thread-private. Never share loop indices across threads.
```cpp
// GOOD: OpenMP thread-private loop bounds
#pragma omp parallel for schedule(dynamic)
for (int i = 0; i < num_atoms; ++i) {
for (int j = i + 1; j < num_atoms; ++j) {
// Distance calculation
}
}
```
### Rule 2: Cache-Line Alignment (`alignas(64)`)
Eliminate false sharing by padding thread-local accumulators to 64-byte cache lines:
```cpp
struct alignas(64) ThreadLocalHistogram {
std::vector<uint64_t> bins;
};
```
---
## 2. Synchronization Primitives
### Rule 1: Safe Mutex Locking (`std::scoped_lock`)
Always use `std::scoped_lock` for locking one or more mutexes (C++17 RAII deadlock-free locking). Avoid raw `.lock()` / `.unlock()`.
```cpp
#include <mutex>
void update_shared_state(SharedData& data1, SharedData& data2) {
std::scoped_lock lock(data1.mutex, data2.mutex);
// Deadlock-safe multi-mutex mutation
}
```
### Rule 2: Thread-Safe Async Slint Dispatch
Background worker threads MUST dispatch UI progress updates using `slint::invoke_from_event_loop()`, and threads must be managed via RAII (`std::jthread`):
```cpp
#include <slint/slint.h>
#include <thread>
std::jthread calculate_async(slint::ComponentHandle<AppWindow> window) {
return std::jthread([window](std::stop_token stop_token) {
// Parallel computation loop
#pragma omp parallel for
for (int i = 0; i < total; ++i) {
if (stop_token.stop_requested()) break;
/* ... */
}
// Async UI update
slint::invoke_from_event_loop([window]() {
window->set_progress(1.0f);
});
});
}
```
---
## 3. Concurrency Anti-Patterns
- ❌ Modifying shared `std::vector` inside OpenMP loops without thread-local buffers or mutexes (data race).
- ❌ Using `std::endl` or raw `std::cout` inside parallel loops (interleaved IO streams).
- ❌ Omitting `alignas(64)` on thread-local reduction arrays (false sharing cache degradation).
- ❌ Using `std::thread::detach()` with unmanaged raw pointers or uncoordinated lifetimes.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |