skippydb 0.2.0

A high-performance verifiable key-value store with SHA256 Merkle trees and optional CUDA GPU acceleration, designed for blockchain state storage.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56

# skippydb

A high-performance verifiable key-value store optimized for blockchain state storage, featuring SHA256 Merkle trees with optional CUDA GPU acceleration.

## Installation

```bash
cargo add skippydb
# With GPU acceleration:
cargo add skippydb --features cuda
```

## Quick Start

```rust
use skippydb::config::Config;
use skippydb::AdsCore;

let config = Config::from_dir("/path/to/data");
AdsCore::init_dir(&config);
```

## Feature Flags

| Feature | Description | Default |
|---------|-------------|---------|
| `default` | Includes jemalloc allocator | Yes |
| `cuda` | CUDA GPU-accelerated SHA256 batch hashing | No |
| `directio` | Linux io_uring direct I/O for reads | No |
| `hpfile_all_in_mem` | Keep heap files entirely in memory | No |
| `tee_cipher` | AES-GCM encryption for data at rest | No |
| `slow_hashing` | Slow hashing mode for testing | No |

## GPU Acceleration

With `--features cuda`, KyumDB batches Merkle tree SHA256 operations to CUDA cores:
- 3-5x throughput improvement at batch sizes >10K
- Automatic kernel selection (CPU/AoS/SoA) via `auto_batch_node_hash`
- Async pipelining, fused active bits, GPU-resident upper tree
- Multi-GPU support via `MultiGpuHasher`

See the [GPU Integration Guide](https://github.com/bajpainaman/skippydb/blob/main/docs/gpu-integration-guide.md) for usage patterns.

## Documentation

- [API Docs (docs.rs)](https://docs.rs/skippydb)
- [GPU Acceleration Details](https://github.com/bajpainaman/skippydb/blob/main/docs/gpu-acceleration.md)
- [GPU Integration Guide](https://github.com/bajpainaman/skippydb/blob/main/docs/gpu-integration-guide.md)

## License

Licensed under either of:
- Apache License, Version 2.0
- MIT License

at your option.