Over the years, many frameworks and optimization techniques have been proposed to accelerate graph neural networks (GNNs). In contrast to the optimizations explored in these systems, we observe that different matrix re-associations of GNN computations lead to novel input-sensitive performance behavior. We leverage this observation to propose GRANII, a system that exposes different compositions of sparse and dense matrix primitives based on different matrix re-associations of GNN computations and selects the best among them based on input attributes. GRANII executes in two stages: (1) an offline compilation stage that enumerates all valid re-associations leading to different sparse-dense matrix compositions and uses input-oblivious pruning techniques to prune away clearly unprofitable candidates, and (2) an online runtime system that explores the remaining candidates and uses lightweight cost models to select the best re-association based on the input graph and the embedding sizes. On a wide range of configurations, GRANII achieves a geo-mean speedup of 1.56× for inference and 1.4× for training across multiple GNN models and systems. We also show GRANII’s technique functions on diverse implementations and with techniques such as sampling.

Sparse tensor programs are essential in deep learning and graph analytics, driving the need for optimized processing. To meet this demand, specialized hardware accelerators are being developed. Optimizing these programs for accelerators is challenging for two reasons: program performance is highly sensitive to variations in sparse inputs, and early-stage accelerators rely on expensive simulators. Therefore, ML-based cost models used for optimizing such programs on general-purpose hardware are often ineffective for early-stage accelerators, as they require large datasets for proper training. To this end, we introduce COGNATE, a novel framework that leverages inexpensive data samples from general-purpose hardware (e.g., CPUs) to train cost models, followed by few-shot fine-tuning on emerging hardware. COGNATE exploits the homogeneity of input features across hardware platforms while effectively mitigating heterogeneity, enabling cost model training with just 5% of the data samples needed by accelerator-specific models to achieve comparable performance. We conduct extensive experiments to demonstrate that COGNATE outperforms existing techniques, achieving average speedups of 1.47× (up to 5.46×) for SpMM and 1.39× (up to 4.22×) for SDDMM.

Multiple frameworks and optimizations have been proposed for accelerating Graph Neural Network (GNN) workloads over the years, achieving sizable runtime performance improvements. However, we notice that existing systems usually explore optimizing either at the intra-operator level or at the inter-operator level, missing synergies that exist due to their compositions. Further, most existing works focus primarily on optimizing the forward computation of GNNs, often overlooking opportunities for training-specific optimizations. To exploit these missed optimization opportunities, we introduce GALA, a domain-specific language (DSL) and a compiler that allows composing optimizations at different levels. The GALA DSL exposes intra-operator transformations as scheduling commands, while we introduce novel inter-operator transformations as part of the compiler. The composition of these transformations is made possible through the introduction of two novel intermediate representations (IR) in the GALA compiler that tracks and composes transformations at both the intra- and inter-operator levels. Further, the IRs maintain a global view of the GNN program, including its training process. This allows us to introduce training-specific transformations to aggressively optimize GNN training. Our evaluations show that GALA achieves a geo-mean speedup of 2.55× for inference and 2.52× for training across multiple systems, graphs, and GNN models. We also show that GALA performs well across different graph sizes and GNN model configurations, as well as allows users to explore different methods of performing similar optimizations leading to different tradeoff spaces.

The widespread use of Sparse Matrix Dense Matrix Multiplication (SpMM) and Sampled Dense Matrix Dense Matrix Multiplication (SDDMM) kernels makes them candidates for hardware acceleration. However, accelerator design for these kernels faces two main challenges: (1) the overhead of moving data between CPU and accelerator (often including an address space conversion from the CPU's virtual addresses) and (2) marginal flexibility to leverage the fact that different sparse input matrices benefit from different variations of the SpMM and SDDMM algorithms.

Data pre-processing is a fundamental component in any data-driven application. With the increasing complexity of data processing operations and volume of data, Cylon, a distributed dataframe system, is developed to facilitate data processing both as a standalone application and as a library, especially for Python applications. While Cylon shows promising performance results, we experienced difficulties trying to integrate with frameworks incompatible with the traditional Message Passing Interface (MPI). While MPI implementations encompass scalable and efficient c ommunication routines, their process launching mechanisms work well with mainstream HPC systems but are incompatible with some environments that adopt their own resource management systems. In this work, we alleviated this issue by directly integrating the Unified Communication X (UCX) framework, which supports a variety of classic HPC and non-HPC process-bootstrapping mechanisms as our communication framework. While we experimented with our methodology on Cylon, the same technique can be used to bring MPI communication to other applications that do not employ MPI’s built-in process management approach.

In the current era of Big Data, data engineering has transformed into an essential field of study across many branches of science. Advancements in Artificial Intelligence (AI) have broadened the scope of data engineering and opened up new applications in both enterprise and research communities. Aggregations (also termed reduce in functional programming) are an integral functionality in these applications. They are traditionally aimed at generating meaningful information on large data-sets, and today, they are being used for engineering more effective features for complex AI models. Aggregations are usually carried out on top of data abstractions such as tables/ arrays and are combined with other operations such as grouping of values. There are frameworks that excel in the said domains individually. But, we believe that there is an essential requirement for a data analytics tool that can universally integrate with existing frameworks, and thereby increase the productivity and efficiency of the entire data analytics pipeline. Cylon endeavors to fulfill this void. In this paper, we present Cylon's fast and scalable aggregation operations implemented on top of a distributed in-memory table structure that universally integrates with existing frameworks.