SCoPing SNAP

This post describes my (ultimately failed) attempts to run Tadashi (Vatai et al., 2024) on SNAP.

In the search for realistic apps I wrote scop_detector, a very simple program which would run PET’s automatic scop detection algorithm on any given .c file, and print out the found schedule trees along with a summary (number of SCoPs). This was obviously wrapped in a scops_in_dir script which ran it on all the .c files in a directory, which made “probing” projects for SCoPs trivial: clone/download the source code, run scops_in_dir and keep an eye out for a big/deep schedule tree. SNAP was one of the first apps which looked promising to extract SCoPs from!

scop_detector showed a nice big and deep schedule tree with multiple nested loop, which was promising, but it also printed an error which I saw for the first time saying “data dependent conditions not supported”. The error message was coming from code copied from PPCG, which was copied as part of the dead code elimination algorithm.

References

Tadashi: Enabling AI-Based Automated Code Generation With Guaranteed Correctness

Emil Vatai, Aleksandr Drozd, Ivan R. Ivanov, and 2 more authors

Oct 2024

Abs DOI

Frameworks and DSLs auto-generating code have traditionally relied on human experts developing them to have in place rigorous methods to assure the legality of the applied code transformations. Machine Learning (ML) is gaining wider adoption as a means to auto-generate code optimised for the hardware target. However, ML solutions, and in particular black-box DNNs, provide no such guarantees on legality. In this paper we propose a library, Tadashi, which leverages the polyhedral model to empower researchers seeking to curate datasets crucial for applying ML in code-generation. Tadashi provides the ability to reliably and practically check the legality of candidate transformations on polyhedral schedules applied on a baseline reference code. We provide a proof that our library guarantees the legality of generated transformations, and demonstrate its lightweight practical cost. Tadashi is available at https://github.com/vatai/tadashi/.

Enjoy Reading This Article?

Here are some more articles you might like to read next: