#lapack
Compiled scipy from source today, on macOS, with Apple clang and Apple BLAS/LAPACK

Please clap
March 10, 2025 at 10:53 PM
yeah, like, I'm not sure how to read that screenshot other than "gpus are alchemy but blas and lapack get a pass"
September 5, 2026 at 12:35 AM
LAPACK too, and a lot of legacy scientific code is in Fortran.
March 2, 2024 at 1:25 AM
Se usan las bibliotecas que se escribieron en fortran hace 60 años, LAPACK es una de ellas, si mal no recuerdo. Están tan cagadas a palos y escritas pensando en optimizar recursos que son seguras y eficientes.
September 21, 2026 at 9:00 PM
Stop doing LAPACK

Vectors were never supposed to be arrays

Norms DON’T exist, do what thou wilt

Years of QR iterations and no one knows what a Jordan normal form is

they’ve played us for absolute fools
May 13, 2025 at 1:21 AM
Haven't blogged in a little while, here's a new one: "R Version 4.5.0 is Out!"

www.njtierney.com/post/2025/04...

- `install.packages()` is faster
- new datasets! `penguins`, `gait`
- BLAS/LAPACK updates
- My thoughts on reading NEWS files and updating R packages

#rstats #rse #rpkgs #cran
R Version 4.5.0 is Out! | Credibly Curious
www.njtierney.com
April 14, 2025 at 8:26 AM
Come for an explanation about floating point errors and LAPACK, and stay for the footnotes
May 18, 2025 at 9:14 PM
I do love the era of capitalized software (ELIZA, LAPACK, PGPLOT, etc)
January 24, 2026 at 4:40 AM
LAPACK working notes are such a guilty pleasure
May 17, 2025 at 7:42 AM
Why do we need libraries like FlexiBLAS?

Michał Górny's latest blog post explains how multiplexing libraries solve the API/ABI compatibility puzzle in BLAS/LAPACK packaging, making life easier for users, developers, and distributions alike.

Learn how it all fits together: buff.ly/wo861po
BLAS/LAPACK packaging
An overview of differences between BLAS / LAPACK implementations and their packaging.
labs.quansight.org
November 13, 2025 at 4:13 PM
the reason Fortran is so fast and still gets used is partly that it, and its libraries (LAPACK, BLAS in particular) have been continually optimized and updated for the last... 69 years. nice.
January 3, 2026 at 10:59 PM
This is why I only write software using the FORTRAN77 distribution included with VAX/VMS 4.0. It kinda sucks that I had to write my own implementation of LAPACK from scratch, but anything for security, hey?
July 19, 2025 at 8:02 PM
Cool, the lapack-sys crate was ‘modernised’ to use a bindgen build script with just this github.com/blas-lapack-...

Relatedly, just learnt you use a [patch.crates.io] section in Cargo.toml to override a *transitive* dependency
May 6, 2025 at 10:39 AM
I'm laying in bed and while I *should* be going to sleep I'm here reading the BLAS and LAPACK documentation, bweh
November 8, 2025 at 1:41 AM
Good mornyan,

This might have been a result of me reading the BLAS and LAPACK docs before going to bed, but my dreams consisted mostly of matrix multiplication,

so that's fun.
November 8, 2025 at 11:21 AM
I wonder if the #lapack that comes with #AOCL is being picked up by flexiblas in #rstats.
The things I have to do for the love of electronic health records analytics #bigdata #blas
March 6, 2026 at 5:19 AM
What's up?
gfortran
LAPACK
BLAS
Ongoing Homebrew conflicts
Stores are closed and I am running low on booze.
December 25, 2024 at 11:47 PM
Exciting work on BLAS/LAPACK packaging!📦

As part of bringing better BLAS/LAPACK support to Meson, Michał Górny learned a lot about how these essential libraries differ across implementations and distributions, and he is sharing it all with the community.

📖 Dive into the details: buff.ly/JAxtRop
BLAS/LAPACK packaging
An overview of differences between BLAS / LAPACK implementations and their packaging.
labs.quansight.org
October 23, 2025 at 5:24 PM
The LaPack family also. We love Mami.
October 22, 2024 at 2:51 AM
That book is a classic. But also MVnorm shouldn’t be doing an eigen-anything. It should be using the lapack cholesky routine. (Which will also not be reproducible)
May 13, 2025 at 9:56 AM
Oddly, it is possible that you might end up with that pattern if MASS::mvrnorm() is the culprit, because the underlying problem occurs when LAPACK flips the sign of the final eigenvector in the matrix decomposition due to floating point nonsense. So it can manifest as a qualitative difference
January 29, 2026 at 9:49 PM
It is almost 6AM and the ADHD meds kicked in and decided "You should only think about writing a BLAS/LAPACK/ATALS implementation only, and not about what you actually need to do"

But It's the weekend so I guess that's fine,
November 8, 2025 at 1:46 PM
incidentally this is why the common linear algebra ABI (blas/lapack) still uses a bunch of absolutely inscrutable 5 letter function names
October 27, 2025 at 10:22 PM
Would be very curious to know if this does end up fixing the issue. We ran into the MASS::mvrnorm() problem at work last year and it took me forever to trace it back to LAPACK floating point indeterminacy. Ever since writing that blog post I've wondered how often the issue arises in the wild
February 1, 2026 at 8:18 PM
LAPACK/BLAS have a neat vector normalisation trick: find the max axis and divide each component by that before rsqrt and scale. Avoids overflow and gradual underflow due to squaring in the rsqrt. If the max axis is tiny then all axes are tiny so the components can be considered roughly proportional.
October 9, 2025 at 9:58 AM