NVIDIA Blackwell Platform Arrives to Power a New Era of …
Mar 18, 2024 · NVIDIA announces the Blackwell platform, a new GPU architecture that enables real-time generative AI on trillion-parameter large language models. The platform features six technologies, including a 208-billion-transistor chip, a second-generation transformer engine, and a fifth-generation NVLink.
NVIDIA Blackwell B200 GPU Achieves 2.2x Performance Increase …
10 hours ago · Notably, the Blackwell GPUs delivered a twofold performance increase for GPT-3 pre-training and a 2.2 times boost for Llama 2 70B low-rank adaptation (LoRA) fine-tuning per GPU.
Nvidia Blackwell and GeForce RTX 50-Series GPUs: Rumors, …
Aug 9, 2024 · Learn about the next generation Nvidia Blackwell GPU architecture and RTX 50-series GPUs, expected to launch by the end of 2024 or early 2025. Find out the latest rumors, specifications,...
Nvidia reveals Blackwell B200 GPU, the ‘world’s most ... - The Verge
Mar 18, 2024 · Nvidia unveils the Blackwell B200 GPU, a 208-billion-transistor chip that offers up to 20 petaflops of FP4 horsepower for AI training and inference. It also introduces the GB200 superchip,...
NVIDIA Contributes Blackwell Platform Design to Open ... - NVIDIA …
Oct 15, 2024 · NVIDIA contributes elements of its Blackwell accelerated computing platform design to the Open Compute Project and broadens Spectrum-X support for OCP standards. The platform connects 36 Grace CPUs and 72 Blackwell GPUs in a rack-scale design for AI factories.
NVIDIA Launches Blackwell-Powered DGX SuperPOD for ... - NVIDIA …
Mar 18, 2024 · NVIDIA DGX SuperPOD is a new AI supercomputer powered by Grace Blackwell Superchips, each with 72 Blackwell GPUs connected via NVLink. It can process trillion-parameter models for superscale generative AI training and inference with constant uptime and fast memory.
NVIDIA Blackwell Delivers Next-Level MLPerf Training …
1 day ago · The Blackwell training results follow an earlier submission to MLPerf Inference 4.1, where Blackwell delivered up to 4x more LLM inference performance versus the Hopper generation. Taking advantage of the Blackwell architecture’s FP4 precision, along with the NVIDIA QUASAR Quantization System, the submission revealed powerful performance ...
‘We Created a Processor for the Generative AI Era,’ NVIDIA CEO …
Mar 18, 2024 · NVIDIA founder and CEO Jensen Huang unveils NVIDIA Blackwell, a new computing platform for real-time generative AI on trillion-parameter large language models. He also introduces NVIDIA NIM, Omniverse Cloud APIs and other advances at GTC 2024.
Blackwell Architecture for Generative AI - NVIDIA
NVIDIA Blackwell is a new class of AI superchip that powers generative AI and accelerated computing with unparalleled performance, efficiency, and scale. Learn about its technological breakthroughs, such as the second-generation Transformer Engine, Confidential Computing, NVLink Switch, Decompression Engine, and RAS Engine.
NVIDIA Blackwell Platform Pushes the Boundaries of Scientific …
Learn how NVIDIA Blackwell platform, unveiled at GTC in March 2024, improves performance and efficiency for scientific computing, AI, quantum computing and data analytics. See examples of applications, such as physics-based simulations, analog circuit design, generative AI and quantum simulation.