Related papers: Lit Silicon: A Case Where Thermal Imbalance Couple…
Hybrid thinking enables LLMs to switch between reasoning and direct answering, offering a balance between efficiency and reasoning capability. Yet our experiments reveal that current hybrid thinking LLMs only achieve partial mode…
Modern large-scale computing systems distribute jobs into multiple smaller tasks which execute in parallel to accelerate job completion rates and reduce energy consumption. However, a common performance problem in such systems is dealing…
The continuous increase in computational power of GPUs, essential for advancements in areas like artificial intelligence and data processing, is driving the adoption of liquid cooling in data centers. Skived copper cold plates featuring…
As Exascale computing becomes a reality, the energy needs of compute nodes in cloud data centers will continue to grow. A common approach to reducing this energy demand is to limit the power consumption of hardware components when workloads…
Hot-spots are considered among the unavoidable consequences of the high integration density of 3D-ICs. Eliminating hot spots requires employing efficient cooling techniques. Using embedded channels, liquid cooling systems can be designed to…
This paper highlights first steps towards enabling graphics processing unit (GPU) acceleration of the task-parallel smoothed particle hydrodynamics (SPH) solver SWIFT. Novel combinations of algorithms are presented, enabling SWIFT to…
This paper investigates the multi-GPU performance of a 3D buoyancy driven cavity solver using MPI and OpenACC directives on different platforms. The paper shows that decomposing the total problem in different dimensions affects the strong…
NVIDIA's Multi-Instance GPU (MIG) technology enables partitioning GPU computing power and memory into separate hardware instances, providing complete isolation including compute resources, caches, and memory. However, prior work identifies…
It is known that 3D-ICs suffer from hot spot temperatures that can reach thousands of degrees, if they are not cooled to reasonable operating temperatures. The problem of hot spots is not limited to the high temperatures of the IC; thermal…
Continuous learning (CL) has emerged as one of the most popular deep learning paradigms deployed in modern cloud GPUs. Specifically, CL has the capability to continuously update the model parameters (through model retraining) and use the…
Training transformer models requires substantial GPU compute and memory resources. In homogeneous clusters, distributed strategies allocate resources evenly, but this approach is inefficient for heterogeneous clusters, where GPUs differ in…
The rapid advancement of Large Language Models (LLMs) necessitates a deep understanding of their fundamental performance limits. This paper investigates the limits of LLM inference, focusing on hardware-imposed bottlenecks in…
Regulatory efforts to govern large language model (LLM) development have predominantly focused on restricting access to high-performance computational resources. This study evaluates the efficacy of such measures by examining whether LLM…
Thermal behavior has become a first-order constraint in advanced 2.5D/3D integrated circuits (ICs) and heterogeneous packages. As power densities rise and multiple active dies are vertically integrated, heat removal paths become…
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…
In large-scale distributed LLM training, communication between devices becomes the key performance bottleneck. Chiplet technology can integrate multiple dies into a package to scale-up node performance with higher bandwidth. Meanwhile,…
Thermal simulation plays a fundamental role in the thermal design of integrated circuits, especially 3D ICs. Current simulators require significant runtime for high-resolution simulation, and dismiss the complex nonlinear thermal effects,…
Latent Dirichlet Allocation(LDA) is a popular topic model. Given the fact that the input corpus of LDA algorithms consists of millions to billions of tokens, the LDA training process is very time-consuming, which may prevent the usage of…
In distributed machine learning (DML), the training data is distributed across multiple worker nodes to perform the underlying training in parallel. One major problem affecting the performance of DML algorithms is presence of stragglers.…
When modeling related tasks in computer vision, Multi-Task Learning (MTL) can outperform Single-Task Learning (STL) due to its ability to capture intrinsic relatedness among tasks. However, MTL may encounter the insufficient training…