NTHRYS › PhD Assistance › Parallel Computing

Parallel Computing

Field

▾

Parallel Computing

Select a category to explore research frontiers

Loading categories...

Research Frontiers in GPU-accelerated Deep Learning Optimization

Research on leveraging GPU parallelism to optimize training and inference of deep neural networks through advanced algorithmic and architectural techniques.

Memory-Compute Decoupling in GPU Neural Architectures

Heterogeneous Precision Training Across Distributed GPUs

Tensor Core Saturation and Latency Hiding Strategies

Dynamic Graph Compilation for GPU Kernel Fusion

Communication-Avoiding Algorithms in Multi-GPU Learning

Sparsity Exploitation in GPU-Accelerated Neural Networks

Adaptive Batch Scheduling for Heterogeneous GPU Clusters

Memory Bandwidth Saturation and Model Scaling Limits

Overlapping Computation and Synchronization in Distributed Training

Quantization-Aware Optimization for GPU Inference Pipelines

All Parallel Computing PhD categories