GPUs accelerate high-throughput applications, which require orders-of-magnitude higher memory bandwidth than traditional CPU-only systems. However, the capacity of such high-bandwidth memory tends to be relatively small. Buddy Compression is an architecture that makes novel use of compression to utilize a larger buddy-memory from the host or disaggregated memory, effectively increasing the memory capacity of the GPU.

Efficient manipulation of sparse matrices is critical to a wide range of HPC applications. Increasingly, GPUs are used to accelerate these sparse matrix operations. We study one common operation, Sparse Matrix Multi-Vector Multiplication (SpMM), and evaluate the impact of the sparsity, distribution of non-zero elements, and tiletraversal strategies on GPU implementations.

Training convolutional neural networks (CNNs) requires intense compute throughput and high memory bandwidth. Especially, convolution layers account for the majority of execution time of CNN training, and GPUs are commonly used to accelerate these layer workloads. GPU design optimization for efficient CNN training acceleration requires the accurate modeling of how their performance improves when computing and memory resources are increased.

Main memory (DRAM) consumes as much as half of the total system power in a computer today, due to the increasing demand for memory capacity and bandwidth. There is a growing need to understand and analyze DRAM power consumption, which can be used to research new DRAM architectures and systems that consume less power. A major obstacle against such research is the lack of detailed and accurate information on the power consumption behavior of modern DRAM devices.

PyTorchFI is a runtime perturbation tool for deep neural networks (DNNs), implemented for the popular PyTorch deep learning platform. PyTorchFI enables users to perform perturbations on weights or neurons of DNNs at runtime. It is designed with the programmer in mind, providing a simple and easy-to-use API, requiring as little as three lines of code for use.

We propose a novel approach for denoising Monte Carlo path traced images, which uses data from individual samples rather than relying on pixel aggregates. Samples are partitioned into layers, which are filtered separately, giving the network more freedom to handle outliers and complex visibility. Finally the layers are composited front-to-back using alpha blending. The system is trained end-to-end, with learned layer partitioning, filter kernels, and compositing.

Deep Learning (DL) models are becoming larger, because the increase in model size might offer significant accuracy gain. To enable the training of large deep networks, data parallelism and model parallelism are two well-known approaches for parallel training. However, data parallelism does not help reduce memory footprint per device. In this work, we introduce Large deep 3D ConvNets with Automated Model Parallelism (LAMP) and investigate the impact of both input's and deep 3D ConvNets' size on segmentation accuracy.

Grasping in cluttered environments is a fundamental but challenging robotic skill. It requires both reasoning about unseen object parts and potential collisions with the manipulator. Most existing data-driven approaches avoid this problem by limiting themselves to top-down planar grasps which is insufficient for many real-world scenarios and greatly limits possible grasps. We present a method that plans 6-DOF grasps for any desired object in a cluttered scene from partial point cloud observations.

End-to-end latency in remote-rendering systems can reduce user task performance. This notably includes aiming tasks on game streaming services, which are presently below the standards of competitive first-person desktop gaming. We evaluate the latency-induced penalty on task completion time in a controlled environment and show that it can be significantly mitigated by adopting and modifying image and simulation-warping techniques from virtual reality, eliminating up to 80\% of the penalty from 80 ms of added latency.