NVIDIA to discuss building AI to build chips for AI at upcoming Hot Chips event

NVIDIA will discuss using AI to build next-generation chips for AI at the upcoming Hot Chips 2024 event next week.

11

The company will discuss how NVIDIA designs some of the most complex products on the planet, with its new Blackwell B200 AI GPU featuring 208 billion transistors, made on TSMC's new 4NP process node. At Hot Chips, NVIDIA will discuss using generative AI to generative optimized Verilog code.

What's Verilog code? Verilog is a hardware description language that describes circuits in the form of code. It's used for design and verification of processors, with NVIDIA building an Agentic AI LLM application to accelerate Computer Aided Engineering (CAE) that generates Verilog code, which can:

• Speed up design generation, debug, review/analysis, and optimization
• Automate a wide variety of manual design tasks
• Enable more complex chip architectures and maintain one-year design cadence

11

Inside of NVIDIA's new Blackwell B200 AI GPU we have 208 billion transistors (104 billion with two reticle-limited GPUs merged into one) on the TSMC 4NP process node. There's 192GB of HBM3E memory with 8TB/sec of memory bandwidth, and 1.8TB/sec bi-directional NVLink bandwidth with high-speed NVLink-C2C Link to Grace CPU.