In a event held at the SAP Center in San Jose, California, Nvidia’s CEO Jensen Huang took the stage to reveal the company’s latest breakthrough in artificial intelligence (AI) technology. The announcement centered around the unveiling of Nvidia’s newest chip, the Blackwell B200, the world’s most powerful chip for AI applications.
Also Read: Apple in Talks with Google to Integrate Gemini AI for iPhone Features
The Blackwell B200 GPU represents a quantum leap in AI processing power has a 20 petaflops of FP4 horsepower derived from its 208 billion transistors.
This achievement shows an advancement over its predecessor, the H100, offering up to 30 times the performance for LLM inference workloads while consuming less energy.
Nvidia claims that the B200 GPU can reduce both cost and energy consumption by up to 25 times compared to the H100.
One of the key highlights of the Blackwell B200 GPU is its capability to train massive AI models with speed and efficiency.
Training a 1.8 trillion parameter model, which previously required 8,000 Hopper GPUs and 15 megawatts of power, can now be accomplished with just 2,000 Blackwell GPUs consuming four megawatts.
This advancement underlines Nvidia’s commitment to democratizing trillion-parameter AI and making AI technologies more accessible for businesses worldwide.
The Blackwell B200 GPU introduces several features that set it apart from its predecessors. A second-gen transformer engine doubles the compute, bandwidth, and model size by utilizing four bits for each neuron instead of eight, resulting in the 20 petaflops of FP4 performance.
A next-gen NVLink switch facilitates communication between GPUs, enabling 576 GPUs to interact with each other with 1.8 terabytes per second of bidirectional bandwidth.
Also Read: Sony PlayStation 5 Pro: 3x Faster GPU, With PSSR Support and More
Has a colossal 208 billion transistors, this powerhouse chip offers up to 20 petaflops of FP4 horsepower. Nvidia CEO Jensen Huang underlined its importance positioning it as a game-changer in the world of AI computing.
Nvidia’s pursuit of innovation has culminated in the Blackwell B200 GPU, delivering performance metrics. Compared to its predecessor, the H100, the B200 offers a 30 times the performance for LLM inference workloads.
The Blackwell B200 GPU is about raw power and efficiency. Nvidia’s chip reduces cost and energy consumption by up to 25 times over the H100.
With the introduction of the GB200 “superchip,” Nvidia is ushering in a new era of AI infrastructure. By combining two B200 GPUs with a single Grace CPU, the GB200 superchip offers levels of performance and efficiency.
The impact of Nvidia’s Blackwell B200 GPU is beyond hardware, it’s about reshaping the cloud computing. Major players including Amazon, Google, Microsoft, and Oracle have already committed to integrating the GB200 into their cloud services.
With the introduction of microservices tailored for AI deployment, Nvidia is empowering developers to harness the full potential of its hardware.
Nvidia’s ambition knows no bounds, as evidenced by its foray into automotive chips and collaboration with leading design software firms. From onboard chatbots in automotive systems to partnerships in 3-D modeling software.
Also Read: Asus Zenfone 11 Ultra Review and Features
/cdn.vox-cdn.com/uploads/chorus_asset/file/25343060/NVIDIA_Blackwell_Architecture_Image.jpg)
