Modern DRAM is based on a brilliant design from IBM.
But, we're still paying for a latency penalty that's existed since the 60s!
In this video, I'm introducing my research project (Tailslayer) that immensely reduces p99.99 latency on traditional RAM!
This was the dream with more bfx tooling moving to Rust. We could start pulling the core pieces of many tools as libraries instead of individual tools.
This is not the road I expected to take to get there, but it still a road there!
One step closer to bfx tools turning into function calls instead of CLI invocations.
As a general Mojo / Modular update - Everything very much looks like it's still tracking for Mojo 1.0 this year, followed by the compiler open sourcing, also this year.
1) There's a new edition of Programming Massively Parallel Processors
2) Modular has the Mojo version of it already up: github.com/modular/modu...
3) The author of PMPP was one of Chris Lattner's PhD committee advisors. Wild how things interconnect.
Summarized from:
x.com/clattner_llv...
github.com/sstadick/moj...
A command line argument parser in the vein of structopt, for Mojo.
Helicase: Vectorized parsing and bitpacking of genomic sequences www.biorxiv.org/content/10.64898/2026.03...
This would be a fantastic addition.
The scariest part about this is that it makes me wonder if header files were onto something...
Sufficiently detailed specs are just code.
haskellforall.com/2026/03/a-su...
I think the cartoon pretty much covers it.
I finally wrote up a paper for perbase!
The process with @joss-openjournals.bsky.social was really great!
github.com/sstadick/per...
This is very cool. Wasmify all the things
…. The name is available.
It could be the ‘yes’ of the 💩emoji. The final evolution of enshitification.
* pooled
This is what happens when you Bluesky and try to watch kids at the same time.
Agh, and I reread that like three times before hitting send 🤦
I don’t remember the exact reason, but if I had to guess it was just to make it easy to fit into the pooped writer.
It also looks like that bgzip crate improved a lot about 3 years ago, so it might have been just missing pieces when we needed them 4 years ago.
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.6... and github github.com/pachterlab/k...
Figure 1 shows they key result
Some thoughts on AI based on working on this project: Some thoughts on AI based on working on this project: liorpachter.wordpress.com/2026/02/19/t...
We are excited to announce that we can successfully use Rust's async/await on the GPU. This has never been done before.
www.vectorware.com/blog/async-a...
Supporting Rust's async/await (and futures) enables existing Rust code to work on the GPU and makes GPU programming more ergonomic.
I’ve written a post about my recent experiences (successes) with AI coding models; the experiences that caused me to re-evaluate my initial judgements, the surprise I had at what can be accomplished, & some fears I have about these tools. Discussion welcome! combine-lab.github.io/blog/2026/02...
setting on github to set commit message of a squashed PR to PR title + description
TIL this setting on GitHub. kinda mad about the years of git commits could have contained my beautiful PR descriptions
ziglang.org/devlog/2026/...
> it seems we are indeed reaching the Promised Land, where Zig code can have Io implementations effortlessly swapped out
hat: a cargo-like wrapper around pixi for managing mojo projects.
github.com/sstadick/hat
It literally just shells out to pixi and is meant to be very transparent! 100% compatible with pixi.
And to counter the slightly anti-ai stance in the essay, here's pretty pro-ai take that seems like a really nice balanced usage model:
mitchellh.com/writing/my-a...
This essay keeps feeling more salient the more I work with various AI agents.
That's not to say that agents aren't massive enablers in some cases. There's just tradeoffs.
So anyway:
BiRank & QuadRank: single-cache-miss rank queries that are double the throughput of other Rust crates and fully saturate the memory bandwidth.
Side effect: QuadFm is smaller and 2-4x faster than the next-best FM-index.
github.com/RagnarGrootK...
raw.githubusercontent.com/RagnarGrootK...
I've just released #rust-htslib 1.0. After a long time with a pretty stable API usage of rust-htslib in production, it feels like the right time to finally move to 1.0. Most important change is probably a switch to thread-safe pointers in BAM record handling. github.com/rust-bio/rus...
