Member-only story
Nvidia GPUs — H100 and A100
Let the cores burn!
Many companies are burning cash for GPUs with the hope of building an AGI. They have built data centers with thousands of GPUs running in parallel to train a large language model or other deep learning models, thus making these massive computing devices the backbone of modern AI.
In this article, we’ll discuss the widely used GPUs H100 and A100 from Nvidia, which are being used in Modern Deep Learning setups at massive data centers.
What are tensor and tensor core?
Both A100 and H100 are tensor core GPUs, these are processing units that help in performing matrix-multiplication efficiently. A tensor in deep learning refers to a data type that stores multi-dimensional data.
All microprocessors are designed to perform logical and arithmetic operations, but when it comes to deep learning or AI in general, matrix-multiplication plays a key role.
Multiplying two 4×4 matrices involves 64 multiplications and 48 additions. Convolution and Multiplication are the areas where the new cores shine.
