Related papers: Beyond a Single Queue: Multi-Level-Multi-Queue as …

In high-performance computing, hotspot GPU kernels are primary bottlenecks, and expert manual tuning is costly and hard to port. Large language model methods often assume kernels can be compiled and executed cheaply, which fails in large…

Sequence parallelism (SP) serves as a prevalent strategy to handle long sequences that exceed the memory limit of a single device. However, for linear sequence modeling methods like linear attention, existing SP approaches do not take…

With the growing number of data-intensive workloads, GPU, which is the state-of-the-art single-instruction-multiple-thread (SIMT) processor, is hindered by the memory bandwidth wall. To alleviate this bottleneck, previously proposed…

The concurrency literature presents a number of approaches for building non-blocking, FIFO, multiple-producer and multiple-consumer (MPMC) queues. However, only a fraction of them have high performance. In addition, many queue designs, such…

Recent advancements in Large Language Models (LLMs) have led to increasingly diverse requests, accompanied with varying resource (compute and memory) demands to serve them. However, this in turn degrades the cost-efficiency of LLM serving…

More and more large data collections are gathered worldwide in various IT systems. Many of them possess the networked nature and need to be processed and analysed as graph structures. Due to their size they require very often usage of…

GPUs are now used for a wide range of problems within HPC. However, making efficient use of the computational power available with multiple GPUs is challenging. The main challenges in achieving good performance are memory layout, affecting…

Linear Programs (LPs) appear in a large number of applications and offloading them to the GPU is viable to gain performance. Existing work on offloading and solving an LP on GPU suggests that performance is gained from large sized LPs…

Dynamically scheduled high-level synthesis (HLS) enables the use of load-store queues (LSQs) which can disambiguate data hazards at circuit runtime, increasing throughput in codes with unpredictable memory accesses. However, the increased…

Large Language Models (LLMs) are increasingly being deployed on edge devices for long-context settings, creating a growing need for fast and efficient long-context inference. In these scenarios, the Key-Value (KV) cache is the primary…

NVIDIA Multi-Process Service (MPS) enables fine-grained GPU sharing by allowing multiple processes to execute concurrently on the same GPU, making it an important mechanism for improving GPU utilization. However, MPS has weak fault…

Nowadays, the data to be processed by database systems has grown so large that any conventional, centralized technique is inadequate. At the same time, general purpose computation on GPU (GPGPU) recently has successfully drawn attention…

Efficient parallelism is necessary for achieving low-latency, high-throughput inference with large language models (LLMs). Tensor parallelism (TP) is the state-of-the-art method for reducing LLM response latency, however GPU communications…

Recurrent neural networks (RNNs) have shown outstanding performance on processing sequence data. However, they suffer from long training time, which demands parallel implementations of the training procedure. Parallelization of the training…

As large language models (LLMs) continue to scale, their workloads increasingly rely on distributed execution across multiple GPUs. However, the conventional bulk synchronous parallel~(BSP) model used in such settings introduces significant…

Sequential quadratic programming (SQP) is widely used in solving nonlinear optimization problem, with advantages of warm-starting solutions, as well as finding high-accurate solution and converging quadratically using second-order…

As inference workloads for large language models (LLMs) scale to meet growing user demand, pipeline parallelism (PP) has become a widely adopted strategy for multi-GPU deployment, particularly in cross-node setups, to improve key-value (KV)…

One of the main bottlenecks when designing a network processing system is very often its memory subsystem. This is mainly due to the state-of-the-art network links operating at very high speeds and to the fact that in order to support…

Existing large language model (LLM) serving systems fall into two categories: 1) a unified system where prefill phase and decode phase are co-located on the same GPU, sharing the unified computational resource and storage, and 2) a…

Matrix Factorization (MF) on large scale data takes substantial time on a Central Processing Unit (CPU). While Graphical Processing Unit (GPU)s could expedite the computation of MF, the available memory on a GPU is finite. Leveraging GPUs…