Related papers: A Case Study in CUDA Kernel Fusion: Implementing F…

Residual block is a very common component in recent state-of-the art CNNs such as EfficientNet or EfficientDet. Shortcut data accounts for nearly 40% of feature-maps access in ResNet152 [8]. Most of the previous DNN compilers, accelerators…

Custom hardware accelerators for Deep Neural Networks are increasingly popular: in fact, the flexibility and performance offered by FPGAs are well-suited to the computational effort and low latency constraints required by many image…

Attention is a core operation in numerous machine learning and artificial intelligence models. This work focuses on the acceleration of attention kernel using FlashAttention algorithm, in vector processors, particularly those based on the…

LiDAR super-resolution addresses the challenge of achieving high-quality 3D perception from cost-effective, low-resolution sensors. While recent transformer-based approaches like TULIP show promise, they remain limited to spatial-domain…

Efficient inference on GPUs using large language models remains challenging due to memory bandwidth limitations, particularly during data transfers between High Bandwidth Memory (HBM) and SRAM in attention computations. Approximate…

To cope with the increasing demand and computational intensity of deep neural networks (DNNs), industry and academia have turned to accelerator technologies. In particular, FPGAs have been shown to provide a good balance between performance…

Deploying Transformer models on edge devices is limited by latency and energy budgets. While INT8 quantization effectively accelerates the primary matrix multiplications, it exposes the softmax as the dominant bottleneck. This stage incurs…

An essential component of modern recurrent sequence models is the forget gate. While Transformers do not have an explicit recurrent form, we show that a forget gate can be naturally incorporated into Transformers by down-weighting the…

In the fusion community, the use of high performance computing (HPC) has been mostly dominated by heavy-duty plasma simulations, such as those based on particle-in-cell and gyrokinetic codes. However, there has been a growing interest in…

We propose AttentionMixer, a unified deep learning framework for multimodal detection of brain edema that combines structural head CT (HCT) with routine clinical metadata. While HCT provides rich spatial information, clinical variables such…

In the domain of point cloud registration, the coarse-to-fine feature matching paradigm has received substantial attention owing to its impressive performance. This paradigm involves a two-step process: first, the extraction of multi-level…

Multimodal remote sensing data, including spectral and lidar or photogrammetry, is crucial for achieving satisfactory land-use / land-cover classification results in urban scenes. So far, most studies have been conducted in a 2D context.…

Token pruning methods for Vision Transformers (ViTs) promise quadratic reductions in attention FLOPs by dropping uninformative patches. Yet standard variable-length attention APIs -- including FlashAttention-2's varlen and PyTorch's…

Point cloud processing is a challenging task due to its sparsity and irregularity. Prior works introduce delicate designs on either local feature aggregator or global geometric architecture, but few combine both advantages. We propose…

Point-based Neural Networks (PNNs) have become a key approach for point cloud processing. However, a core operation in these models, Farthest Point Sampling (FPS), often introduces significant inference latency, especially for large-scale…

The evolution of Vision Transformers has led to their widespread adaptation to different domains. Despite large-scale success, there remain significant challenges including their reliance on extensive computational and memory resources for…

This paper proposes a neural network architecture for tackling the query-by-example user-defined keyword spotting task. A multi-head attention module is added on top of a multi-layered GRU for effective feature extraction, and a normalized…

The attention layer, a core component of Transformer-based LLMs, brings out inefficiencies in current GPU systems due to its low operational intensity and the substantial memory requirements of KV caches. We propose a High-bandwidth…

We introduce the CUDA Tensor Transpose (cuTT) library that implements high-performance tensor transposes for NVIDIA GPUs with Kepler and above architectures. cuTT achieves high performance by (a) utilizing two GPU-optimized transpose…

We describe an implementation of compressible inviscid fluid solvers with block-structured adaptive mesh refinement on Graphics Processing Units using NVIDIA's CUDA. We show that a class of high resolution shock capturing schemes can be…