Domain-Specific Multi-Level IR Rewriting for GPU

نویسندگان

چکیده

Most compilers have a single core intermediate representation (IR) (e.g., LLVM) sometimes complemented with vaguely defined IR-like data structures. This IR is commonly low-level and close to machine instructions. As result, optimizations relying on domain-specific information are either not possible or require complex analysis recover the missing information. In contrast, multi-level rewriting instantiates hierarchy of dialects (IRs), lowers programs level-by-level, performs code transformations at most suitable level. We demonstrate effectiveness this approach for weather climate domain. particular, we develop prototype compiler design stencil- GPU-specific based set newly introduced principles. find that two (500 lines code) realized top LLVM’s extensible MLIR infrastructure suffice outperform state-of-the-art solutions. essence, promises herald age specialized composed from domain- target-specific implemented shared infrastructure.

برای دانلود باید عضویت طلایی داشته باشید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

High-Level GPU Programming: Domain-Specific Optimization and Inference

When writing computer software one is often forced to balance the need for high runtime performance with high programmer productivity. By using a high-level language it is often possible to cut development times, but this typically comes at the cost of reduced run-time performance. Using a lower-level language, programs can be made very efficient but at the cost of increased development time. R...

متن کامل

Multi-level modelling via stochastic multi-level multiset rewriting

We present a simple stochastic rule-based approach to multilevel modelling for computational systems biology. Populations are modelled using multilevel multisets; these contain both species and agents, with the latter possibly containing further such multisets. Rules are pairs of such multisets, but now allowing variables to occur (as well as species and agents), together with an associated sto...

متن کامل

Domain-Specific IR for German, English and Russian Languages

In participating in this domain-specific track, our first objective is to propose and evaluate a light stemmer for the Russian language. Our second objective is to measure the relative merit of various search engines used for the German and to a lesser extent the English languages. To do so we evaluated the tf ·idf , Okapi, IR models derived from the Divergence from Randomness (DFR) paradigm, a...

متن کامل

A GPU-based Multi-level Algorithm for Boundary Value Problems

A novel and scalable geometric multi-level algorithm is presented for the numerical solution of elliptic partial differential equations, specially designed to run with high occupancy of streaming processors inside Graphics Processing Units (GPUs). The algorithm consists of iterative, superposed operations on a single grid, and it is composed of two simple full-grid routines: a restriction and a...

متن کامل

Multi-level Parallel Query Execution Framework for CPU and GPU

Recent developments have shown that classic database query execution techniques, such as the iterator model, are no longer optimal to leverage the features of modern hardware architectures. This is especially true for massive parallel architectures, such as many-core processors and GPUs. Here, the processing of single tuples in one step is not enough work to utilize the hardware resources and t...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: ACM Transactions on Architecture and Code Optimization

سال: 2021

ISSN: ['1544-3973', '1544-3566']

DOI: https://doi.org/10.1145/3469030