memkit
Deterministic memory allocation for systems requiring predictable performance.
memkit provides intent-driven CPU memory allocation with explicit lifetimes and zero-overhead abstractions. It is designed for high-performance systems where latency spikes are unacceptable and memory behavior must be auditable.
⚡ The memkit Advantage
| Feature | System Monitor | memkit | Impact |
|---|---|---|---|
| Allocation Cost | O(n) / Varies | O(1) | Zero latency spikes |
| fragmentation | High over time | Zero (Reset) | Stable long-term uptime |
| Cache Locality | Random | Contiguous | fast iteration & SIMD |
| Thread Safety | Locks / Contention | Lock-Free TLS | Linear scaling |
| Cleanup | Per-object Drop | Bulk Reset | Instant frame cleanup |
🛡️ Sentinel Mode (New in Beta)
Enable the sentinel feature to activate debug-hardened memory verification:
- Poisoning: Fills allocated (
0xAA) and freed (0xDD) memory to catch use-after-free bugs. - Trace Logging: Reports allocation hazards to stderr.
- Leak Detection: Verifies zero leaks on shutdown.
[]
= { = "0.1.1-beta.1", = ["sentinel"] }
🚀 Quick Start
use ;
// Standard setup
let alloc = new;
loop
📦 Ecosystem
| Crate | Purpose | Status |
|---|---|---|
| memkit | Core allocators & primitives | Beta |
| memkit-gpu | Type-safe GPU memory management | Beta |
| memkit-co | CPU↔GPU transfer coordination | Beta |
| memkit-async | Async-aware backpressure & pools | Beta |
| cargo-memlense | Static analysis linter | Beta |
🤝 License
Licensed under the Mozilla Public License 2.0 (MPL-2.0). See LICENSE.md for details.