Efficient JIT compilation of MATLAB loops
Bibliographic record
Abstract
MATLAB R is a dynamic numerical scripting language widely used by scientists, engineers and students.It is praised because it allows fast prototyping, especially for numerical programs which manipulate matrices.However, numerical software can be computationally heavy, and MATLAB, as most interpreted languages, suffers from slow performance as compared to traditionally compiled languages such as FORTRAN or C++.One way to provide better performance for interpreted languages is through just-in-time compilation, where the program (or part of the program) is compiled at run-time.In this thesis, we introduce SJIT, a just-in-time compiler for MATLAB which focuses on providing good performance while keeping the compilation time extremely small.It is designed to integrate easily and transparently into an existing interpreter for MATLAB named Mc VM, and generates highly efficient assembly code intensive parts of the program, namely loops.In addition to its use for accelerating whole MATLAB programs, it is also suitable for accelerating the execution of fragments of MATLAB code inside an interactive environment such as a read-eval-print loop.In addition to the SJIT compiler, this thesis also contributes an efficient framework to develop static dataflow analyses, and a type inference analysis implemented within this framework.The SJIT compiler has been evaluated, both in terms of compilation time and execution time, on a collection of MATLAB benchmarks using traditional features such as matrices and structures.The results show that: (1) the achieved performance is several times faster than the original MATLAB implementation, and (2) that the compilation time is very reasonable, taking only a small fraction of the overall time.i I would like to thank all the persons who made these two years a nice experience, and thus helped me to make this thesis: My family in Canada for their dedication in making my life in Montreal nicer. My family in France for their long-distance support, from video calls to handwritten letters. The members of the Sable lab, for the lunchs in Chinatown. My supervisor, Laurie, for dealing
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.002 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".