Today, we are celebrating 10 years of open source!
Thank you to all the ArrayFire developers and users over the years. It's been an incredible ride. And we're working on our best release yet, coming soon :-)
QF Foundation's first paper explores GPU acceleration for primitive operations in Substrate-based blockchains, marking the completion of phase 1 for team 2's roadmap.
Key findings:
-Implemented GPU acceleration using WebAssembly and arrayfire-rust
-Achieved 2-3x speedup for basic arithmetic on large datasets (1M float operations)
-GPU: 0.38-0.50ms vs CPU: 1.15-1.85ms average execution time
This research lays groundwork for GPU-accelerated smart contract execution and introduces potential for dynamic CPU/GPU operation balancing. It aligns with our broader goal of revolutionizing blockchain performance through advanced parallel processing.
Next steps include exploring more complex GPU-accelerated operations. Our work with PolkaVM promises even greater efficiency gains beyond WASM.
Want to explore the code or dive deeper?
Read the full study: qf.foundation/Running_primit…
Run the code yourself: github.com/QuantumFusion-Fou…
See what was implemented: github.com/paritytech/polkad…
With over 500 papers on Google Scholar using ArrayFire, here are some of the amazing things you have built because GPU programming is a little easier with a higher-level array API,
arrayfire.com/blog/category/…
An exciting step towards open standards and ecosystems in GPU computing.
Download ArrayFire v3.9 today, which includes a new SYCL (oneAPI) backend, arrayfire.com/download
We are thrilled to announce the launch of the UXL Foundation, a cross-industry group committed to delivering an open standard accelerator programming model that simplifies development of performant, cross-platform applications.
Announcement: hubs.la/Q022FpF30#uxl#oneapi
**New ArrayFire v3.9.0 Release**
We are excited to announce a major release to ArrayFire, including a new oneAPI backend for greater flexibility and commitment to open GPU and XPU computing standards.
Binaries and source code are available for download.
arrayfire.com/blog/arrayfire…
*New Blog Post*
"Reconstruction of 3D Phase Contrast Atomic Electron Tomography using ArrayFire"
by researchers from UC Berkeley and Lawrence Berkeley National Laboratory.
Enjoy!
arrayfire.com/blog/reconstru…
*New Blog Post*
Parallelization of FOAGDD (First-order Anisotropic Gaussian Direction Derivative) Point of Interest Extraction
by Gustavo Stahl from São Paulo State University in Brazil.
Enjoy!
arrayfire.com/blog/paralleli…
New blog post today from our series on quantum computing:
"How does a Quantum Computer work?"
These are related to the ArrayFire Quantum Simulator.
Enjoy!
arrayfire.com/blog/how-does-…
Check out Shumai by Facebook Research, built with ArrayFire and Flashlight.
It is a fast differentiable tensor library in JavaScript and TypeScript. Benchmarks show this ArrayFire-based project beating TensorFlow by 2-8X for many different cases.
Enjoy!
github.com/facebookresearch/…
Announcement: ArrayFire v3.8.3 Release is now available.
This patch release contains support for CUDA 12 as well as many performance improvements and bug fixes.
arrayfire.com/blog/arrayfire…
New Blog Post: "Simulating Soliton Excitations in Open Systems"
In mathematics and physics, a soliton, or solitary wave, is a self-reinforcing wave that maintains its shape while it propagates at a constant velocity. A cool ArrayFire project by @UPorto.
arrayfire.com/blog/simulatin…
New Blog Post: "ArrayFire Quantum Simulator"
This summer we built an open-source quantum simulator using ArrayFire that outperforms other simulators. It's easy-to-use and comes with a command-line visualizer.
If you take a look, your state will change!
arrayfire.com/blog/arrayfire…
"Intel's oneAPI work has been extremely promising along with their open-source projects at large over the years."
Totally agree! Thanks for the @phoronix shout-out, @michaellarabel!