TensorCore.si explains the hardware behind artificial intelligence in plain language. We help students, developers and curious readers understand how Tensor Cores, GPUs and AI chips actually work.
Our mission
Every AI model, from chatbots to image generators, runs on billions of matrix multiplications. Tensor Cores are the specialised units that make that math fast. Most explanations of them are written for chip engineers. We write for everyone else, using simulations, diagrams and step-by-step guides so the ideas are easy to see, not just read.
What we cover
- Tensor Cores: how matrix engines work, generation by generation.
- GPU Architecture: streaming multiprocessors, CUDA cores, memory and caches.
- AI Chips: data-center accelerators, their specs and where each one fits.
- Simulations: animated visuals of the math running inside the hardware.
- Guides: beginner-friendly explainers on precision formats, sparsity and more.
- News & Analysis: launches and benchmarks from across the AI hardware world.
Free interactive tools
- Exploded chip: take a GPU package apart layer by layer.
- Live simulation: watch a matrix multiply-accumulate run on a CUDA core and on a Tensor Core.
- TensorCore.si Challenge: randomized math problems with verifiable certificates.
- AI Model Explorer: size the memory and compute that real AI models need.
- Glossary: every AI hardware term defined in plain language.
How we write
- Primary sources first. We rely on official architecture documents, research papers and published specifications, and we link to them at the end of every guide.
- Clear over clever. Every technical term is explained the first time it appears.
- Honest about limits. Simulations and diagrams are simplified to teach a concept, and we say so when they are. Vendor performance claims are labelled as vendor claims.
- Kept up to date. Hardware changes fast. Each article shows the date it was last updated.
- Corrections. If we get something wrong, we fix it. Spot an error? Tell us.
Who we are
TensorCore.si is an independent site. Every article, tool and dataset is edited and fact-checked by our editor, Sami Ullah Khan, a geologist who spent more than seven years monitoring real-time drilling data in the oil and gas industry and now writes about AI tools. He checks every specification against vendor datasheets and research papers before it is published.
Read our editorial policy to see where our numbers come from and how we use AI tools, and our site updates page for every significant change and correction. Our LLM benchmarks page explains how we test the assumptions behind our calculators.
Independence
TensorCore.si is an independent educational website. We are not affiliated with, sponsored by or endorsed by NVIDIA or any other hardware manufacturer. Read our Disclaimer for details.
Questions, suggestions or partnership enquiries are welcome on our Contact Us page or by email at hi@tensorcore.si.