Sandeep Dasgupta

Work @Google PhD @UIUC ML for Compilers & Compiler for ML, Formal Methods, Reverse Engineering Love Dancing :)

[Mehdi Amini @JokerEph]

Tomorrow, I present at the #MLIR Open Meeting a deep-dive on nitty-gritty impl. details behind MLIR Core data structures: Operation and Attribute.
Then I'll introduce some core changes with the new concept of "Properties".
No dialect or codegen, but "under the hood" kind of talk!

[Mehdi Amini @JokerEph]

Excited about developing the #OpenXLA project!
We'll grow the #MLIR ecosystem extensively in the coming years :)

[Francis Visoiu Mistrih @francisvismis]

Want to see some cool compiler optimizations that my team has been working on? Check out Ahmed’s talk on improving size and performance! https://t.co/FLkbIsizqk

[Nikolaj Bjorner @BjornerNikolaj]

Are you curious about SMT solvers? Want to learn about Z3, a state-of-the-art SMT solver, and run Z3 programs *in your browser*? We are excited to host the first online Z3 learning workshop with research interns @rlisahuang and @Ayana_Monr on August 24. https://t.co/bJMuyHp2DZ

[Sandeep Dasgupta @sdasgup3]

@AartBik @ChapmanU That’s awesome 👏 🎉

[Sandeep Dasgupta @sdasgup3]

@alastair_d_reid Best wishes !

[Aart Bik @AartBik]

Sad to read this. Granted, I was a Commodore 64 guy myself, but the contributions of Sir Clive Sinclair on making home computing possible are tremendous!

[Grigore Rosu @RosuGrigore]

Exciting times for RV @rv_inc: just opened our office in Singapore and Hao (Spencer) Xiao https://t.co/UO4JtzuBSt has joined RV to build and lead our Singapore team. Hao has a PhD in computer science and is an expert in formal methods, static and runtime analysis.

Welcome Hao!

[Chris Lattner @clattner_llvm]

Congratulations to Aho and Ullman, who propelled the development of the compiler field and touched many of us with their work and writing. The dragon book cover art also inspired the next generation of dragons in the LLVM community! 🐉🐉

[Sandeep Dasgupta @sdasgup3]

@krismicinski @alastair_d_reid @johnregehr @peter_sewell_ @samth @secretasianman @vadve @RosuGrigore And feel free to engage me if you wanted to explore.

[Sandeep Dasgupta @sdasgup3]

@krismicinski @alastair_d_reid @johnregehr @peter_sewell_ @samth @secretasianman Thanks for the details!
You may have a look at
https://t.co/2VmqsVl3ZJ. This semantics is for ATT style X86-86 assembly. @vadve @RosuGrigore
From https://t.co/NxU0fLvYxI

[Vikram Adve @vadve]

Hard to believe that it has been 20 years since our first discussions about a persistent, language-indep, machine-indep, SSA-based IR suitable for both static and JIT compilation. Design ideas rooted in compiler research like SSA, separation of concerns, LTO... #CelebrateLLVM

[Sandeep Dasgupta @sdasgup3]

@notypes IMHO: an opportunity to use such parallel machines could be is to do post IR auto-tuning for generating optimized backend-code (with the search space as the huge number of pass sequences or different RA strategies)

[Sandeep Dasgupta @sdasgup3]

A nice article by @alastair_d_reid: (https://t.co/j3nEc1h7ff)
"Aim to increase the uptake and usability of formal methods by developing libraries and tools that build on activities that programmers are already comfortable and effective with .."

[Chandler Carruth @chandlerc1024]

Going back through Singer's 2006 PhD dissertation on SSI, and wow is it just amazing...

If you're interested in compiler IRs, especially the control flow graph family, and doubly those using virtual registers, it is totally worth reading. Chapter 7 is an amazing survey...

[Peter Goodman @peter_a_goodman]

@alastair_d_reid @rustlang We created a similar tool called DeepState: https://t.co/8Kd9EuVhUQ. DeepState is a C++ property test framework modelled around Google Test that connects to fuzzers and symbolic executors.

[Alastair Reid @alastair_d_reid]

Verifying vs testing? Why not both?
New blog post about our work on Rust verification based on reimplementing the proptest library for formal verification tools so you can use the same test harness with both proptest and with “propverify”.
@rustlang #KLEE
https://t.co/rAzAvW6MCI

[TensorFlow @TensorFlow]

Introducing TF-Coder 🖥

A program synthesis tool that helps you write tensor manipulations in TensorFlow. Simply provide an input/output example of the desired behavior, and leave the rest to TF-Coder!

Learn more → https://t.co/p6e6dlpyQO
Try it out → https://t.co/tYtXQJnidJ

[Vikram Adve @vadve]

Super proud to be leading the new National AI Institute for Ag Innovation with a broad and deep team of AI and Ag researchers! It's going to be an exciting 5 years! #AIFARMS