Rubens Lacouture

Erik Hellsten, Artur Souza, Johannes Lenfers et al. · stanford

We introduce the Bayesian Compiler Optimization framework (BaCO), a general purpose autotuner for modern compilers targeting CPUs, GPUs, and FPGAs. BaCO provides the flexibility needed to handle the requirements of modern autotuning tasks. Particularly, it deals with permutation, ordered, and continuous parameter types along with both known and unknown parameter constraints. To reason about these parameter types and efficiently deliver high-quality code, BaCO uses Bayesian optimiza tion algorithms specialized towards the autotuning domain. We demonstrate BaCO's effectiveness on three modern compiler systems: TACO, RISE & ELEVATE, and HPVM2FPGA for CPUs, GPUs, and FPGAs respectively. For these domains, BaCO outperforms current state-of-the-art autotuners by delivering on average 1.36x-1.56x faster code with a tiny search budget, and BaCO is able to reach expert-level performance 2.9x-3.9x faster.

Rubens Lacouture, Nathan Zhang, Ritvik Sharma et al.

As deep learning models scale, sparse computation and specialized dataflow hardware have emerged as powerful solutions to address efficiency. We propose FuseFlow, a compiler that converts sparse machine learning models written in PyTorch to fused sparse dataflow graphs for reconfigurable dataflow architectures (RDAs). FuseFlow is the first compiler to support general cross-expression fusion of sparse operations. In addition to fusion across kernels (expressions), FuseFlow also supports optimizations like parallelization, dataflow ordering, and sparsity blocking. It targets a cycle-accurate dataflow simulator for microarchitectural analysis of fusion strategies. We use FuseFlow for design-space exploration across four real-world machine learning applications with sparsity, showing that full fusion (entire cross-expression fusion across all computation in an end-to-end model) is not always optimal for sparse models-fusion granularity depends on the model itself. FuseFlow also provides a heuristic to identify and prune suboptimal configurations. Using Fuseflow, we achieve performance improvements, including a ~2.7x speedup over an unfused baseline for GPT-3 with BigBird block-sparse attention.

Atharva Chougule, Alexander J Root, Rubens Lacouture et al.

Sparse tensor algebra is challenging to efficiently parallelize due to the irregular, data-dependent, and potentially skewed structure of sparse computation. We propose the first partitioning algorithm that provably load balances the computation of any sparse tensor algebra expression across parallel execution units. Our algorithm generalizes parallel merging algorithms to any number of operands, and to multi-dimensional, hierarchical sparse data structures. We implement our algorithm within an existing sparse tensor algebra compilation framework to automatically generate parallel sparse tensor algebra kernels that target multi-core CPUs and GPUs. We show that our generated code is competitive with hand-implemented parallelization strategies used by vendor libraries like Intel MKL and NVIDIA cuSPARSE (geo-means of $0.73$--$3.4\times$) and \textsc{Taco} (geo-means of $1.0$--$2.4\times$), and significantly outperforms general-purpose strategies for sparse tensor expressions where specialized algorithms have not been developed (geo-means of $2.0$--$6.4\times$).

Jungwoo Kim, Rubens Lacouture, Genghan Zhang et al.

Mixture-of-Experts (MoE) has become a dominant architecture for scaling large language models (LLMs). However, the execution characteristics of MoE inference are changing rapidly and increasingly mismatch the assumptions underlying existing Processing-in-Memory (PIM) systems. Prior PIM systems for LLMs rely on static rules to offload memory-bound operations to PIM, without accounting for the combined effects of load imbalance and inter-GPU communication. Meanwhile, modern MoE models activate fewer experts out of increasingly many, creating a bimodal expert distribution: a small set of experts receives many tokens, while a long tail of experts receives only one or a few. We identify a trend in modern MoE models toward increasingly bimodal token-to-expert distributions, quantify the resulting disparity in arithmetic intensity across experts, and show that this disparity dramatically reduces the efficiency of state-of-the-art PIM systems for LLMs. To address this problem, we propose a scheduler for serving MoE models on multi-GPU systems with attached HBM-PIM stacks. Our scheduler partitions expert execution between GPU and PIM based on runtime token-to-expert distributions, while jointly considering interconnect overhead, memory bandwidth, GPU throughput, and PIM throughput. Moreover, we propose Sieve, a runtime framework that employs the scheduler to coordinate execution across GPUs and their attached HBM-PIM stacks. Sieve overlaps GPU computation, PIM computation, and intra- and inter-device communication while preserving cross-device dependencies induced by expert parallelism. Sieve is evaluated on our cycle-accurate simulator based on Ramulator 2.0. Compared to state-of-the-art PIM systems for MoE, Sieve improves both throughput and interactivity by 1.3x, 1.3x, and 1.6x on Qwen3.5-397B-A17B, GPT-OSS-120B, and Qwen3-30B-A3B, respectively.