Sometimes I talk to machines, but more often machines talk to me. Compilers, LLVM, Mutation Testing. Ruby runtime at https://nitter.cf/t.co/ZC5I7W5ifI
Berlin, Germany
Joined April 2011
- Tweets5.5K
- Following448
- Followers1.5K
- Likes3.9K
Pinned Tweet
Collected a few links for anyone interested in learning compilers in general, and LLVM in particular
lowlevelbits.org/how-to-lear…
What are your favorite resources? Please them!
If you find the Linalg dialect confusing, then try looking at it from a different perspective:
A beginner’s guide to MLIR’s linalg dialect: how to read linalg.generic’s indexing maps and how to lower it to simple loops by hand.
open.substack.com/pub/lowlev…
@[email protected] 🇺🇦 retweeted
Here's the fourth part of my blog series on Compiler and LLVM internals. Co-authored with @0xP0ch1ta
This blog provides a basic introduction to Machine IR, followed by an in-depth explanation of how to write and run an LLVM Machine Function Pass.
Key concepts covered:
1. Machine IR
2. Machine Basic Blocks
3. MachineModuleInfo
4. Target, TargetRegistry and TargetMachine.
5. MIR Parser
sh4dy.com/2026/10/04/learnin…
@[email protected] 🇺🇦 retweeted
I'm looking for my next role, in Montreal or remote! My background spans compilers, systems, and AI research.
I'd welcome introductions, leads, or a repost to help spread the word ❤️
More details about my experience and what I'm looking for: pointersgonewild.com/2026-09…
Decided it's time to give a shot to vibe coding, so finally built a thing I've been missing for many years.
Currently, only a handful of repos are analyzed, but already found a few cool papers.
papergrep.dev
@[email protected] 🇺🇦 retweeted
New blog post: Machine Scheduler in LLVM - Part II
myhsu.xyz/llvm-machine-sched…
@[email protected] 🇺🇦 retweeted
Machine Scheduler in LLVM (Part I)
.
It's a bit weird I keep writing about scheduling model in LLVM but never really talk about the scheduler itself. So here I am to (finally) close this loop
myhsu.xyz/llvm-machine-sched…
Recently I had a chance to play with MLIR's Dataflow Analysis Framework.
I really love how it is implemented and encourage anyone working with MLIR to get familiar with it.
The missing guide to Dataflow Analysis in MLIR open.substack.com/pub/lowlev…
Wrote a technical note on how to choose the right LLVM/MLIR distribution to build a compiler or a compiler-adjacent tool
lowlevelbits.com/p/different…
@[email protected] 🇺🇦 retweeted
Are there Senior/Staff+ Engineers on my timeline who want to work on the intersection of Databases, Caches, Infra, K8s/Operators and Scaling Algorithms (think time series, predictive scaling etc)?
If so, please drop me a DM. Our small but mighty team is growing!
@[email protected] 🇺🇦 retweeted
maskray.me/blog/2025-04-06-l… LLVM integrated assembler: Improving MCExpr and MCValue – at Los Altos, CA
Wrote a short technical note on building LLVM plugin using Bazel
lowlevelbits.com/p/building-…
Started working on another blog post, I don't know if/when I'm going to finish it, but at least I crafted this nice image
@[email protected] 🇺🇦 retweeted
I am looking to recruit a PhD student (fully funded at UK home tuition rate) to work on automated testing and verification of machine learning compilers and runtimes! Deadline: 30th April. Please spread the word! Details here: doc.ic.ac.uk/~afd/PhD-Advert…
Just booked my trip to fosdem.org/2025/, anyone coming to Brussels?
Recently, landed my first non-trivial PyTorch PR which adds NVTX annotations for CUDA profiling during training, makes it easier to analyze profiles in a visual and automated way.
Should be available in the next version (2.6) 🥳
@[email protected] 🇺🇦 retweeted
If you dont have time to go through a 1000 page book about compilers but you are curious about them you might follow this instead
lowlevelbits.org/how-to-lear…
Recently we open sourced our minimalistic AOT Ruby compiler, the goal was to check if the idea behind the compilation makes sense from the performance perspective. It seems it does!
So I wrote a short article about the implementation details blog.llvm.org/posts/2024-12-…
While I implemented this transformation pipeline, seeing it from the end2end perspective is very satisfying
@[email protected] 🇺🇦 retweeted
My LLVM Dev Meeting 2024 talk on RISC-V vector and performance tuning in compiler. Cheers!
youtu.be/h2tDWoTWUnw?si=eBic…
@[email protected] 🇺🇦 retweeted
A mathematician enters a bar. The rest of the joke is left as an exercise to the interested reader.