Asher's Avatar

Asher

@ashermancinelli

compilers and coffee and lifting Compiler engineer at nvidia These views are exclusively my own https://www.ashermancinelli.com/about.html

37
Followers
79
Following
26
Posts
15.11.2024
Joined
Posts Following

Latest posts by Asher @ashermancinelli

The Flang documentation and source code is up there too! Glad you liked the episode!

21.09.2025 16:39 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
High-Performance Compilers for Parallel Computing This work covers everything necessary to build a compet…

Thanks! Michael Wolfe’s racecar book might be a fun read: www.goodreads.com/book/show/33...

21.09.2025 16:38 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Compilers are like fish: the closer you look, the harder they are to define

27.07.2025 13:25 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
The Unreasonable Effectiveness of Progressive Lowering - Asher's Blog

www.ashermancinelli.com/csblog/2025-... The Unreasonable Effectiveness of Progressive Lowering - Asher's Blog

27.07.2025 04:36 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Thanks a bunch!

25.07.2025 20:56 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Thanks for the read! I know next to nothing about rust so I’d love to revise it

25.07.2025 16:27 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

There are some warts but a huge development in compiler tech imo

25.07.2025 16:23 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

You might think of it as 500 seperate IRs or progressively lowering one mega ir, but principle is the same. Nicer than the typical micropass arch imo. Things can move up or down in the pipeline as needed which isnt always the case w micropasses

25.07.2025 16:22 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Lol ic ic. One of the nice features of mlir is its heterogeneity for this reason- all your IRs still use the same substrate+infrastructure and you can progressively lower from higher to lower level semantics

25.07.2025 16:20 πŸ‘ 1 πŸ” 0 πŸ’¬ 2 πŸ“Œ 0

Interesting. Whoms your intended user? What kinda code are they writing?

25.07.2025 16:16 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Oh gotcha! I’d love to chat more about it haha

25.07.2025 16:14 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
My Ideal Array Language - Notes

That was sorta the premise of this blog post for anyone interested www.ashermancinelli.com/csblog/2025-...

25.07.2025 16:13 πŸ‘ 2 πŸ” 0 πŸ’¬ 2 πŸ“Œ 0

Excited to see what y’all figure out! I think it’s really cool to have unbufferized values at the ssa stage - most compilers choose between the two but when you have both, it’s like youve forced the src lang to be pure functional, so its easier to optimize

25.07.2025 16:12 πŸ‘ 0 πŸ” 0 πŸ’¬ 2 πŸ“Œ 0

So the hir isnt ssa?

25.07.2025 15:26 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

The llvm flang compiler has some cool unbufferized array SSA types at the highest level IR, wonder if theres space/use for that in HIR

25.07.2025 14:58 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

What semantics are you representing in the higher level ir that you couldnt/didnt at the lower ones?

25.07.2025 14:24 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Why UB is Good (When It Comes to Strict Aliasing) - Notes

Post on why I think UB is useful in some cases like strict-aliasing, but to the extent it's observable by the user it's largely a disaster

www.ashermancinelli.com/csblog/2025-...

#compilers #programminglanguages

24.07.2025 02:07 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Let's talk about poison. - Notes

Some thoughts on poisons in LLVM www.ashermancinelli.com/csblog/2025-...

23.07.2025 20:31 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

What’s julia’s memory model for arrays? It’s got an llvm based jit, right?

23.07.2025 12:46 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

I’m all for julia but there’s some super neat stuff in fortran! Also, the llvm fortran compiler is super fancy- maybe the first general purpose mlir based compiler (as in, not ai related)

23.07.2025 12:45 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

What was your fortran xp like?

23.07.2025 01:09 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Yoooo

23.07.2025 01:09 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Hey welcome!

25.11.2024 15:18 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Hey thank you for this!

25.11.2024 15:13 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Please make a starter pack if you do find moreπŸ™

21.11.2024 02:17 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Post image

Buny

15.11.2024 15:56 πŸ‘ 4 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0