voin 🫟

[SIMD @simdcompute]

93 million $SIMD tokens locked again until January 2027.

We're here to build: GPU-native simulation infrastructure and real utility for the SIMD ecosystem - starting soon.

Live Roadmap: https://t.co/HCL8ADBvrG

SIMD on top!

[voin 🫟 @voindotfun]

@simdcompute SIMD cooking

[SIMD @simdcompute]

Brae is heading to the supercomputer 🚀

We’ve secured access to LuxProvide’s MeluXina platform, including NVIDIA A100s GPU infrastructure.

We’ll use this pilot to accelerate BRAE’s development, benchmark it in an HPC environment, and advance our goal of building large, multiscale simulations.

Thank you to @luxprovide for the support!

[voin 🫟 @voindotfun]

@Alex_Oloyede2 @grok is this true?

[voin 🫟 @voindotfun]

@simdcompute SIMD 🧩

[SIMD @simdcompute]

SIMD is getting an experimental GPU network via brae.

Lend an idle GPU. It runs real OpenFOAM cases.
More details soon

https://t.co/0kz6XIgzYM

[voin 🫟 @voindotfun]

@simdcompute SIMD 🧩

[SIMD @simdcompute]

So much learning.

https://t.co/RTDDs4y6Af

[SIMD @simdcompute]

A full F-16 at 32° angle of attack, deep stall, resolved on one GPU.
25M cells. SA-IDDES. 50 m/s freestream, 1.5 s of physical time, five convective passes over the airframe, solved in 4.4 hours on a single H100.

For comparison, SPUMA would be ~4.5× slower on the same case/hardware (~19.8 hours)

Solved with brae.
1/ orbit view

[mango @mangoizgwd]

Cooking

[SIMD @simdcompute]

Brae, our GPU-native OpenFOAM solver, on one H100. Same case, matched <1% on the fields.

brae is faster by:
- 2.5x vs OpenFOAM, 24 cores
- 3.9x vs SPUMA
- 26x vs OpenFOAM + AMGX
- 30x vs OpenFOAM + PETSc

The whole solve stays on the GPU.
https://t.co/DDzXCeUQAE

[mango @mangoizgwd]

New independent results.
SIMD leading the way 🦀
solana:EMeugag3yfyvKqNKknGDWAudNALafZjbv9ByzCE8pump

[SIMD @simdcompute]

Independent benchmark (not ours): a user ran brae on a single RTX 5090.

- 4.89M cells: 21s vs OpenFOAM 768s CPU / 564s AMGX / 643s PETSc. ~30x.

- It beats even GPU-accelerated OpenFOAM, because AMGX/PETSc only offload the solver while brae runs it all on-GPU.

[bayang @Bayang_BM]

We had banchmark brae on Blackwell hardware (GB10) without HBM. Now, we are running the same banchmark on hardware with High Memory bandwidth. We expect to have speed-up.

[voin 🫟 @voindotfun]

@mangoizgwd Lfg

[bayang @Bayang_BM]

No 1,000-core cluster.
One GPU cna do the same job and cheaper. It is entirely written in CUDA and validated against OpenFOAM.

[SIMD @simdcompute]

Introducing brae: GPU-native computational fluid dynamics. 🌊

The entire simulation runs on one GPU, no matrix shuffled back to the CPU every iteration.

~5× faster than GPU-accelerated OpenFOAM, on the same GPU. Validated to <1%.

🌐 https://t.co/hNi7lL9gm5

[Aglio olio e peperoncino @aglioglioliolio]

So you're telling me there's a very solid distributed compute + critical physics simulation project for a trillion dollar deep tech industry (aerospace, cryogenics, heavy-engineering) sitting at a $284k mc just because CT is midcurving it, with:

- A fully doxxed and ridiculously overqualified team :

Aditya (PhD, Quantum Chem) also founded https://t.co/esO0BbhTe9 (incubated by Berlin Quantum Pioneer) that got tapped for the Alchemist Accelerator (backed by a16z, Sequoia, Accel) which previously scaled billion dollar companies like Rigetti Quantum. He also has links to the European defense tech ecosystem.

Sam (PhD, Computer Science): Deep research pedigree (Max Planck, Institute for Systems Neuroscience). Currently Co-founder/CTO of Polyploy, pushing AI protein.

Bayang: Built DeepFile (secure, local AI for high-performance doc indexing/privacy).

Ankit: Currently at Lake Fusion Technologies (founded by ex-Airbus engineers). Before that, he was at Dassault Systèmes—yeah, the company making fighter jets and stuff.

- A fully functionnal infra with constant upgrades targeting a massive multi-billion dollar market and disrupting giants like ANSYS, Siemens, and Dassault by bringing competitive Computational Fluid Dynamics (CFD) at a fraction of the cost. For context: A single ANSYS license can easily cost $100k+ a year.

- Already onboarding partners like Onnes Cryogenics (an Amazon and Starburst x Israel Aerospace Industries-backed startup).

- Working towards an aggressive GTM strategy.

- Won hackathons for fun and was present at prestigious academic venues, attending the Chemnitz Finite Element Symposium—a premier international conference dedicated to the mathematical analysis and application of the Finite Element Method (FEM).

and NONE of those $SQUIRE $WARD $CPX $ZERO and $SURPLUS maxis are aware of it?

$SIMD
@simdcompute

[RTS @RTradeAcademy]

$SIMD | @simdcompute

I am very bullish on $SIMD

SIMD is basically Cursor for engineering simulations. Prompt → CFD / thermal sim → GPU compute → production-ready outputs.

They’re abstracting away one of the hardest workflows in engineering with AI agents.
Aerospace, defense, energy, robotics massive markets still using outdated tooling.

If AI changed coding, SIMD could do the same for physics & simulation.

$SIMD EMeugag3yfyvKqNKknGDWAudNALafZjbv9ByzCE8pump