Concurrent computation and communication (C3) is a pervasive paradigm in ML and other domains, making its performance optimization crucial. In this paper, we carefully characterize C3 in ML on GPUs, which are most widely deployed for ML training and inference. We observe that while C3 leads to performance uplifts, the uplifts are far lower than ideal speedups (serial computation and communication versus maximum of computation or communication; all times from isolated executions). That is, C3 on average achieves only 21 % of ideal speedup. This is so, due to known challenges of compute and memory interference between concurrent GPU kernels (that is, sharing of GPU's compute units, caches and HBM). To attain better performance for C3, first, we evaluate dual strategies of schedule prioritization and careful resource partitioning of compute units on GPUs to push performance attained with C3 (on average 42 % of ideal speedup). We also provide heuristics that can guide a runtime while employing these strategies. To further enhance$\mathbf{C 3}$performance, we propose to mitigate C3 interference by offloading communication tasks to the GPU's DMA engines. To this end, we build concurrent communication collectives (ConCCL) proof-of-concepts that harness DMA engines for communication. We show how ConCCL considerably closes the gap between realized and ideal speedup for C3 (on average 72 % of ideal speedup is realized, up to$1.67 \times$speedup). Overall, our work makes a strong case for GPU DMA engine advancements to better support C3 on GPUs.

Modern accelerators like GPUs increasingly execute independent operations concurrently to improve the device’s compute utilization. However, effectively harnessing it on GPUs for important primitives such as general matrix multiplications (GEMMs) remains challenging. Although modern GPUs have significant hardware and software GEMM support, their kernel implementations and optimizations typically assume each kernel executes in isolation and can utilize all GPU resources. This approach is highly efficient when kernels execute in isolation, but causes significant resource contention and slowdowns when kernels execute concurrently. Moreover, current approaches often only statically expose and control parallelism within an application, without considering runtime information such as varying input size and concurrent applications—often exacerbating contention. These issues limit performance benefits from concurrently executing independent operations. Accordingly, we propose GOLDYLOC , which considers the global resources across all concurrent operations to identify performant GEMM kernels, which we call globally optimized (GO)-Kernels. GOLDYLOC also introduces a lightweight dynamic logic which considers the dynamic execution environment for available parallelism and input sizes to execute performant combinations of concurrent GEMMs on the GPU. Overall, GOLDYLOC improves the performance of concurrent GEMMs on a real GPU by up to 2× (18% geomean per workload) versus the default concurrency approach and provides up to 2.5× (43% geomean per workload) speedup over sequential execution.

Large Language Models increasingly rely on distributed techniques for their training and inference. These techniques require communication across devices which can reduce scaling efficiency as the number of devices increases. While some distributed techniques can overlap, and thus, hide this communication with independent computations, techniques such as Tensor Parallelism (TP) inherently serialize communication with model execution. One approach to hide this serialized communication is to interleave it with the producer operation (of the communicated data) in a fine-grained manner. However, this fine-grained interleaving of communication and computation in software can be difficult. Furthermore, as with any concurrent execution, it requires compute and memory resources to be shared between computation and communication, causing resource contention that reduces overlapping efficacy.

The ubiquity of deep neural networks (DNNs) continues to rise, making them a crucial application class for hardware optimizations. However, detailed profiling and characterization of DNN training remains difficult as these applications often run for hours to days on real hardware. Prior works have exploited the iterative nature of DNNs to profile a few training iterations to represent the entire training run. While such a strategy is sound for networks like convolutional neural networks (CNNs), where the nature of the computation is largely input independent, we observe in this work that this approach is sub-optimal for sequence-based neural networks (SQNNs) such as recurrent neural networks (RNNs). The amount and nature of computations in SQNNs can vary for each input, resulting in heterogeneity across iterations. Thus, arbitrarily selecting a few iterations is insufficient to accurately summarize the behavior of the entire training run. To tackle this challenge, we carefully study the factors that impact SQNN training iterations and identify input sequence length as the key determining factor for variations across iterations. We then use this observation to characterize all iterations of an SQNN training run (requiring no profiling or simulation of the application) and select representative iterations, which we term SeqPoints. We analyze two state-of-the-art SQNNs, DeepSpeech2 and Google's Neural Machine Translation (GNMT), and show that SeqPoints can represent their entire training runs accurately, resulting in geomean errors of only 0.11% and 0.53%, respectively, when projecting overall runtime and 0.13% and 1.50% when projecting speedups due to architectural changes. This high accuracy is achieved while reducing the time needed for profiling by 345x and 214x for the two networks compared to full training runs. As a result, SeqPoint can enable analysis of SQNN training runs in mere minutes instead of hours or days.

Machine learning (ML) has recently emerged as an important application driving future architecture design. Traditionally, architecture research has used detailed simulators to model and measure the impact of proposed changes. However, current open-source, publicly available simulators lack support for running a full ML stack like PyTorch. High-confidence, cycle-accurate simulations are crucial for architecture research and without them, it is difficult to rapidly prototype new ideas. In this paper, we describe changes we made to GPGPU-Sim, a popular, widely used GPU simulator, to run ML applications that use cuDNN and PyTorch, two widely used frameworks for running Deep Neural Networks (DNNs). This work has the potential to enable significant microarchitectural research into GPUs for DNNs. Our results show that the modified simulator, which has been made publicly available with this paper1Source code available at https://github.com/gpgpu-sim/gpgpu-sim_distribution (dev branch), provides execution time results within 18% of real hardware. We further use it to study other ML workloads and demonstrate how the simulator identifies opportunities for architectural optimization that prior tools are unable to provide.

We motivate DARTS, a new technique for workload sampling to drive architectural simulation, that carefully integrates hardware performance counter monitoring with fast, dynamic binary translation. This paper focuses on a key finding that motivates the DARTS methodology. We present our experience applying it to the generation of simulation points for the integer benchmarks of the SPEC CPU2006 suite. We were able to match existing approaches in terms of coverage of dynamic workload behaviours and simulated performance. DARTS achieved this with significant improvements in turn-around time, simulation effort, and storage requirements, with a slight increase in the number of simulation points.