Just took the new engine in Python-Blosc2 4.0 for a spin, and the performance results are quite awesome. Processing a 400 MB array:

🔹 NumPy baseline: 146 ms
🔹 Blosc2 (on NumPy arrays): 73.1 ms (2x faster)
🔹 Blosc2 (on native Blosc2 arrays): 15.1 ms (**9.6x faster!**) 🤯

The best part? It fully supports NumPy's array and ufunc interfaces. High performance with zero friction! 🏎️💨

More info: https://ironarray.io/blog/miniexpr-powered-blosc2

#Python #DataScience #NumPy #HighPerformanceComputing #Blosc2 #OpenSource

We’ve rebuilt the engine. Python-Blosc2 4.0 is live! 🚀

Key novelty: the `miniexpr` engine. It moves evaluation from RAM-heavy chunks to private-cache blocks (KBs). This keeps data hot, beating the "Memory Wall".

🔹 World-class compute & compression in one pass
🔹 Outperforms NumPy in specific workloads
🔹 4.5x speedup in real-world Cat2Cloud operations

Read the technical breakdown:
👉 https://ironarray.io/blog/miniexpr-powered-blosc2

#Python #HPC #Performance #Blosc2 #DataScience

🚀 Our #PyDataGlobal 2025 tutorial recording on modern #Blosc2 & #Caterva2 features is out!
We show how compression is more than just a space saver, boosting performance for large in-memory & out-of-memory arrays via auto-chunking & parallelism.

We also cover: 🌐 Serving Blosc2/#HDF5 data online with Caterva2 ☁️ Computing directly in the cloud (no downloads needed!)

Watch here: 👉 https://www.youtube.com/watch?v=tUvSI3EpTBQ&list=PLGVZCDnMOq0qmerwB1eITnr5AfYRGm0DF&index=80

#Python #DataScience #BigData #HPC #DataHandling

SSDs, but specially RAM prices are skyrocketing: https://arstechnica.com/gadgets/2025/11/spiking-memory-prices-mean-that-it-is-once-again-a-horrible-time-to-build-a-pc

These are excellent times for using compression to reduce your storage pressure.

https://www.blosc.org/posts/roofline-analysis-blosc2/

#Blosc2 #Compression #HPC

Interact with your vasts remote datasets right in your phone! 📱

I've built a demo Jupyter notebook that connects to a Cat2Cloud server from an Android phone and slices into an 8 TB dataset, downloading a 1 MB chunk in under 100 milliseconds. ⚡

The 8 TB dataset is from the Gaia DR3 catalogue. As it turns out, there are ~1000 stars in a cube of 100 light-years in our vicinity; the space is mostly empty.🔭 🌌

Try this out by visiting: https://cat2.cloud/demo/roots/@public/large/slice-gaia-3d.ipynb

#Cat2Cloud #Blosc2 #BigData

Our @EuroSciPy 2025 tutorial on modern Blosc2 features is now online! 🚀

For the first time, we present our holistic view on how compression can revolutionize data handling, sharing, and computing.

Learn how Blosc2 boosts performance for large datasets, how to serve data online with Caterva2, and compute directly in the cloud.

Watch now! 👉 https://www.youtube.com/watch?v=BdpTtzX2cuk

#Blosc2 #EuroSciPy #Python #DataScience #BigData #HPC #OpenSource

💊IronPill 2💊
In the second of our series of short videos ("ironPills") showcasing ironArray's work, we see how Blosc2 can be used to power heavy-duty linear algebra (100GB!) workflows
⚡1.5-2x faster than PyTorch + h5py!
🧱 automated chunking optimised for your machine's cache hierarchy
🐍 simple one-line syntax 𝚋𝚕𝚘𝚜𝚌𝟸.𝚖𝚊𝚝𝚖𝚞𝚕(𝙰, 𝙱, 𝚞𝚛𝚕𝚙𝚊𝚝𝚑='𝚘𝚞𝚝.𝚋𝟸𝚗𝚍')

See blog here: https://ironarray.io/blog/la-blosc

#Blosc2
#Data
#SignalProcessing
#LinearAlgebra

💊 IronPill 1💊
In the first of a series of short videos ("ironPills") showcasing ironArray's work, we see how Blosc2 can be used to calculate Fourier approximations:
⚡5x faster than NumPy
👣 fraction of the memory footprint
🐍 pythonic one-line syntax 𝚜𝚞𝚖(𝚊 * 𝚌𝚘𝚜(𝚝) + 𝚋 * 𝚜𝚒𝚗(𝚝), 𝚊𝚡𝚒𝚜=𝟷)

See full notebook here: https://github.com/Blosc/python-blosc2/blob/main/examples/ndarray/ironpill1.ipynb

(inspired by this blog post: https://towardsdatascience.com/numexpr-the-faster-than-numpy-library-that-no-ones-heard-of/)

#Blosc2
#Data
#SignalProcessing

🗣️ Announcing Python-Blosc2 3.8.0 🚀

A step closer to compliance with the array-api standard: data-apis.org/array-api!
This is an effort across all array-based libraries so that your code works (e.g. for both blosc2 and NumPy) by simply changing the import statement below!

Highlights:
✅ C-Blosc2 updated to latest 2.21.2
✅ Incorporate isnan, isfinite, isinf
✅ Better indexing coverage
✅ linspace and arange functions more numerically stable
✅ Improved array-api compliance

#arrayapi #Blosc2

Struggling to get performant code from LLMs? 🤔 They can't do the empirical, target-specific optimization needed for modern CPUs.

We can help! 💪 We've spent countless hours profiling and micro-benchmarking Blosc2 for you. Use our C/Python implementations as building blocks for your high-performance apps.

Ready to dive in? 🚀
🔹 EuroSciPy Talk: https://www.blosc.org/docs/2025-EuroSciPy-Blosc2.pdf
🔹 Tutorial: https://github.com/Blosc/EuroSciPy2025-CCC-Tutorial

#HPC #Python #Blosc2 #Optimization