GPU Hardware: Topics Coming Soon
Short, single-idea pages on what a GPU physically is. Here is what is queued next for this section.
TopNotchNote
Notes that matter
queued next
- PCIe generations — the host-device bottleneck, and how to tell what your card actually negotiated.
- HBM generations — HBM2 through HBM3E, and why bandwidth keeps climbing.
- NVLink vs Infinity Fabric — the fast peer network that skips the host entirely.
already published in this section
CPU vs GPU: Why GPUs Exist
Anatomy of a GPU: Die, Package and Memory
What a Kernel Launch Physically Does
SM vs CU: The Core Compute Block
FLOPs and TFLOPS
Consumer vs Datacenter GPUs
GPU History
Driver, Toolkit and Runtime
Warps vs Wavefronts
The GPU Memory Hierarchy
Tensor Cores vs Matrix Cores
Nvidia ↔ AMD GPU Glossary
Anatomy of a GPU: Die, Package and Memory
What a Kernel Launch Physically Does
SM vs CU: The Core Compute Block
FLOPs and TFLOPS
Consumer vs Datacenter GPUs
GPU History
Driver, Toolkit and Runtime
Warps vs Wavefronts
The GPU Memory Hierarchy
Tensor Cores vs Matrix Cores
Nvidia ↔ AMD GPU Glossary