GPU Memory Hierarchy Part 2: Global and Specialized Memory Types
Deep dive into global memory, HBM, texture memory, and unified memory with comprehensive comparisons
SriHarsha Paladugula
GPU Memory Hierarchy Part 1: Understanding the Foundations
Introduction to GPU memory hierarchy and the fastest memory types - registers, shared memory, and L2 cache
CPU Memory Architecture: Foundations and GPU Differences
Explore CPU memory architecture, its importance in deep learning, and key differences from GPUs.
Efficiency Metrics Part 3: Computation Metrics (MACs and FLOPs)
Understand MACs and FLOPs - the fundamental computation metrics for measuring neural network complexity. Learn formulas, layer-wise calculations, and how to compare model efficiency.
Efficiency Metrics Part 2: Memory Metrics for Deep Learning
Understanding parameters, model size, and activation memory - critical metrics for resource-constrained deployment
Efficiency Metrics Part 1: Performance Metrics for Deep Learning
Understanding latency and throughput - the key performance metrics for evaluating neural network efficiency
Why Do We Need Efficient Deep Learning Computing?
Efficient deep learning computing speeds training, reduces costs, and enhances scalability