commonware (@commonwarexyz) / X

commonware

483 posts

commonware

@commonwarexyz

Joined July 2024

Pinned
commonware
@commonwarexyz
Nov 7, 2025
patrickogrady.xyz
@_patrickogrady
Nov 7, 2025
Welcome, @tempo Today, we're thrilled to share that Tempo is building with the @commonwarexyz Library, supporting our growth as a new core contributor, and accelerating our research and engineering with a strategic investment. commonware.xyz/blogs/welcome-…
28K
commonware reposted
Jeff Schroeder
@SEJeff
6h
We really love working with @_patrickogrady and the whole commonware team. Absolute chads, all of them 🫶
asymmetric research
@asymmetric_re
7h
Fuzzing @commonwarexyz isn't a pile of one-off harnesses. It's engineered: deterministic runtime, controlled randomness, reviewable fuzz targets, reproducible crashes, promoted regressions. The result: 60+ bugs found across its primitives, 90 targets running 24/7. Full post ↓
808
commonware
@commonwarexyz
5h
asymmetric research
@asymmetric_re
7h
Fuzzing @commonwarexyz isn't a pile of one-off harnesses. It's engineered: deterministic runtime, controlled randomness, reviewable fuzz targets, reproducible crashes, promoted regressions. The result: 60+ bugs found across its primitives, 90 targets running 24/7. Full post ↓
913
commonware reposted
patrickogrady.xyz
@_patrickogrady
5h
To support targeted manipulation of @commonwarexyz's deterministic runtime, we built FuzzRng with @asymmetric_re TL;DR tweak task ordering by flipping input bits
asymmetric research
@asymmetric_re
7h
Fuzzing @commonwarexyz isn't a pile of one-off harnesses. It's engineered: deterministic runtime, controlled randomness, reviewable fuzz targets, reproducible crashes, promoted regressions. The result: 60+ bugs found across its primitives, 90 targets running 24/7. Full post ↓
1.3K
commonware reposted
asymmetric research
@asymmetric_re
7h
Replying to @asymmetric_re
Under the Hood: Engineering Commonware Fuzzing
From blog.asymmetric.re
389
commonware reposted
asymmetric research
@asymmetric_re
7h
Fuzzing @commonwarexyz isn't a pile of one-off harnesses. It's engineered: deterministic runtime, controlled randomness, reviewable fuzz targets, reproducible crashes, promoted regressions. The result: 60+ bugs found across its primitives, 90 targets running 24/7. Full post ↓
Made with AI
4.1K
commonware reposted
Max Resnick
@MaxResnick
Jun 17
The latest generation of LLMs + @commonwarexyz primatives have greatly accelerated the pace at which we can go from design to testing!
patrickogrady.xyz
@_patrickogrady
Jun 17
Since the start, its been a dream of ours to accelerate systems research with the @commonwarexyz Library. Stoked to see @GScaffino, @jneu_net, and @MaxResnick find a use for 10+ primitives in their new Gatling prototype (~200ms e2e tx latency 🤯):
3.3K
commonware reposted
Joachim Neu
@jneu_net
Jun 17
The @commonwarexyz code base has been a great resource (and accelerant) for this project indeed!
patrickogrady.xyz
@_patrickogrady
Jun 17
Since the start, its been a dream of ours to accelerate systems research with the @commonwarexyz Library. Stoked to see @GScaffino, @jneu_net, and @MaxResnick find a use for 10+ primitives in their new Gatling prototype (~200ms e2e tx latency 🤯):
1.6K
commonware reposted
Giulia Scaffino
@GScaffino
Jun 17
Replying to @GScaffino
10/ Want the full story? 📄 Read the paper: arxiv.org/abs/2606.18220 💻 Check out our Gatling prototype at: github.com/scaffino/Gatli…
arxiv.org
Gatling: Rapid-Fire Consensus from Parallel Composition
Consensus protocols form the core of blockchains and other replicated state machines, ensuring that all correct nodes process the same totally ordered log of input transactions. In fault-free...
5.7K
commonware reposted
Giulia Scaffino
@GScaffino
Jun 17
Replying to @GScaffino
9/ The catch is head-of-line blocking: if one instance stalls (crashed leader), it holds up the merged log. So there's an optimal K, which we derive in closed form. More reliable validator set → more barrels you can run.
1.1K
commonware reposted
Giulia Scaffino
@GScaffino
Jun 17
Replying to @GScaffino
8/ Key point: extreme protocol-level optimization is not needed for ultra-low latency. Parallel composition of a simple, slow component protocol beats highly-tuned single-instance designs. Gatling stays compatible with pipelined / fast-path tricks layered inside the components.
1.2K
commonware reposted
Giulia Scaffino
@GScaffino
Jun 17
Replying to @GScaffino
7/ We built Gatling from the production Simplex deployment by @commonwarexyz @_patrickogrady with a deliberately unoptimized 500ms per-instance block time, on a global cluster. Result: 214ms end-to-end transaction latency and 10ms inter-proposal time. Rotating leaders every 10ms.
1.3K
commonware reposted
Giulia Scaffino
@GScaffino
Jun 17
Replying to @GScaffino
6/
3K