SaaSHub helps you find the best software and product alternatives Learn more →
Top 23 Rust High Performance Projects
-
Project mention: Tauri's updater downloads your whole app every time. Making it download only the diff took more than bsdiff | dev.to | 2026-09-27
tauri-plugin-updater-delta is that equivalent: it sends only what changed between two releases. Delta updates for Tauri have been an open feature request since December 2024, and the suggested fix there was "run bsdiff on the binaries".
-
AppSignal
AppSignal knows why the f*#k it crashed. Stop vibe-debugging. Every exception, every backtrace, grouped so you see patterns, not noise.
-
hyperswitch
Open source, composable payments platform | PCI compliant | SaaS and Self-host options | Enables connectivity to multiple payment, payout, fraud, vault and tokenization providers | Uplifts authorization with intelligent routing and revenue recovery | Reduce payment processing costs with cost observability | Reduces payment ops with reconciliation
-
Project mention: We Cut Log Costs by 35% Using Vector 0.30 and Loki 3.0: Lessons from a 3-Month Tuning | dev.to | 2026-05-04
We evaluated three alternatives: ClickHouse for log storage, Fluent Bit for log collection, and the Vector (https://github.com/vectordotdev/vector) + Loki (https://github.com/grafana/loki) stack. ClickHouse had great query performance but required manual index management, which would add operational overhead. Fluent Bit was lightweight but lacked the transform capabilities we needed to mask PII and drop low-value logs. Vector and Loki stood out: Vector is a Rust-based agent with 1/10th the memory footprint of Filebeat, and Loki is designed for cost-efficient log storage with a query model that aligns with how our team actually debugs (using labels, not full-text search).
-
-
hora
🚀 efficient approximate nearest neighbor search algorithm collections library written in Rust 🦀 .
-
Project mention: Show HN: ExANS – Lossless KV cache compression at 622 GB/s on H100 | news.ycombinator.com | 2026-08-05
- 622 GB/s median GPU decode
Decompression is ~10× faster than a 400 Gb/s NIC bandwidth delivering data losslessly without quality change.
We've are open sourcing this as: ExANS which will be available through our vLLM and SGLang connectors on OpenLake v0.8 version. No changes are required in the inference engine.
I'm curious how others are handling KV transfer today. Are you using KV compression or is bandwidth not a bottleneck yet?
Thanks!
GitHub: https://github.com/openlake-project/openlake
-
-
SaaSHub
SaaSHub - Software Alternatives and Reviews. SaaSHub helps you find the best software and product alternatives
-
-
-
EmbedAnything
Highly Performant, Modular, Memory Safe and Production-ready Inference, Ingestion and Indexing built in Rust 🦀
-
-
-
-
Project mention: Rsloop – an event loop for asyncio written in Rust | news.ycombinator.com | 2026-04-16
-
Project mention: MuseAir hash: new fastest high-quality portable hashing algorithm, v1 released | news.ycombinator.com | 2026-07-10
-
-
-
-
-
wse
WSE - Rust-powered WebSocket engine for Python. Up to 5M del/s fan-out, native cluster binary protocol, zero-GIL JWT, E2E encryption
Project mention: Show HN: WSE Rust-powered WebSocket engine for Python 2.1M del/SEC fan-out | news.ycombinator.com | 2026-02-24https://github.com/silvermpx/wse
-
-
blockframe-rs
Self-healing erasure-coded storage engine with transparent filesystem mounting. Reed-Solomon fault tolerance in a single binary
Removed mmap conditions and thresholds to enable default mmap for file read in tier 2 commits. Previously mmap would be selected if a file was greater than 10mb, use_mmap would be checked within the segment loop constantly, even though use_mmap was a constant. Presence of conditional mmap meant doing variable sized buffer fills which would eventually lead to short reads and corruption of segments due to filling EOF segment with dummy data as the buffer would be statically sized for reusability.
Added padding to generate_parity_segmented. The Reed-Solomon erasure coding is ran on SIMD instructions, meaning the data which is being processed on the specific lanes must be 64 bytes. Prior if 65 bytes were passed, the 1 left over bit wouldn't be filled into the SIMD vector which created undefined behaviour. Added next multiple of 64 padding for exception chunks.
Changed commit boundaries for tier 1 from 10mb to 25mb. Changed sha256 function name to blake3_hash_bytes.
Feel free to check it out, clone or download the release for the platform of your choice. And if you appreciated what I built, please consider giving it a star and sharing, that would mean a lot. Thank you, have a cracking day :)
https://github.com/crushr3sist/blockframe-rs/releases/tag/v1...
-
flashq
Blazingly fast job queue server. Process millions of jobs/sec with sub-millisecond latency. Built with Rust.
-
SaaSHub
SaaSHub - Software Alternatives and Reviews. SaaSHub helps you find the best software and product alternatives
NOTE:
The open source projects on this list are ordered by number of github stars.
The number of mentions indicates repo mentiontions in the last 12 Months or
since we started tracking (Dec 2020).
Rust High Performance discussion
Rust High Performance related posts
-
Tauri's updater downloads your whole app every time. Making it download only the diff took more than bsdiff
-
Show HN: Linux server management over SSH – written in Rust and Tauri
-
Show HN: ExANS – Lossless KV cache compression at 622 GB/s on H100
-
MuseAir hash: new fastest high-quality portable hashing algorithm, v1 released
-
DBX: An Open-Source, 15 MB Database Client for 17+ Databases (Built with Tauri)
-
5 desktop app pain points that pushed me to build my own Go framework
-
Rsloop – an event loop for asyncio written in Rust
-
A note from our sponsor - SaaSHub
www.saashub.com | 6 Oct 2026
Index
What are some of the best open-source High Performance projects in Rust? This list will help you:
| # | Project | Stars |
|---|---|---|
| 1 | tauri | 111,514 |
| 2 | hyperswitch | 45,271 |
| 3 | vector | 22,667 |
| 4 | sled | 9,096 |
| 5 | hora | 2,656 |
| 6 | openlake | 2,652 |
| 7 | may | 2,437 |
| 8 | uzu | 1,820 |
| 9 | tensorbase | 1,460 |
| 10 | EmbedAnything | 1,310 |
| 11 | hyperion | 1,083 |
| 12 | aquatic | 622 |
| 13 | fast-float-rust | 292 |
| 14 | rsloop | 203 |
| 15 | museair | 94 |
| 16 | spinoza | 91 |
| 17 | feox-server | 82 |
| 18 | furnace | 64 |
| 19 | UNIT3D-Announce | 63 |
| 20 | wse | 50 |
| 21 | slas | 36 |
| 22 | blockframe-rs | 25 |
| 23 | flashq | 24 |
Sponsored
AppSignal knows why the f*#k it crashed.
Stop vibe-debugging. Every exception, every backtrace, grouped so you see patterns, not noise.
www.appsignal.com