Technical Report about Tiramisu: a Three-Layered Abstraction for Hiding Hardware Complexity from DSL Compilers

TL;DR

Tiramisu introduces a three-layer IR for DSL compilers that simplifies optimization and enables efficient, multi-architecture code generation, significantly reducing complexity while maintaining high performance.

Contribution

It presents Tiramisu, a novel three-level IR that separates algorithm, execution, and data storage, facilitating multi-architecture code generation from a common middle-end.

Findings

Enabled new algorithm expressions like recurrent filters.

Achieved complex loop transformations such as wavefront parallelization.

Generated code with up to 16X speedup over existing implementations.

Abstract

High-performance DSL developers work hard to take advantage of modern hardware. The DSL compilers have to build their own complex middle-ends before they can target a common back-end such as LLVM, which only handles single instruction streams with SIMD instructions. We introduce Tiramisu, a common middle-end that can generate efficient code for modern processors and accelerators such as multicores, GPUs, FPGAs and distributed clusters. Tiramisu introduces a novel three-level IR that separates the algorithm, how that algorithm is executed, and where intermediate data are stored. This separation simplifies optimization and makes targeting multiple hardware architectures from the same algorithm easier. As a result, DSL compilers can be made considerably less complex with no loss of performance while immediately targeting multiple hardware or hardware combinations such as distributed nodes with both CPUs and GPUs. We evaluated Tiramisu by creating a new middle-end for the Halide and Julia compilers. We show that Tiramisu extends Halide and Julia with many new capabilities including the ability to: express new algorithms (such as recurrent filters and non-rectangular iteration spaces), perform new complex loop nest transformations (such as wavefront parallelization, loop shifting and loop fusion) and generate efficient code for more architectures (such as combinations of distributed clusters, multicores, GPUs and FPGAs). Finally, we demonstrate that Tiramisu can generate very efficient code that matches the highly optimized Intel MKL gemm (generalized matrix multiplication) implementation, we also show speedups reaching 4X in Halide and 16X in Julia due to optimizations enabled by Tiramisu.