HyperAI
Command Palette
Search for a command to run...
NVIDIA B200
Specifications
Architecture
NVIDIA Blackwell
Release Date
2024.03
CUDA Cores
37,888
Total Graphics Power
1000 W
GPU Memory
180 GB HBM3E
Memory Bandwidth
8.20 TB/s
Performance Metrics
FP8 Performance
Not supported
FP16 Performance
2,382.4 TFLOPS
FP32 Performance
74.45 TFLOPS
FP64 Performance
74.44 TFLOPS
Key Features
Built on the NVIDIA Blackwell architecture, this GPU features 37,888 CUDA cores and 180 GB HBM3E of memory.
With a TGP of 1000 W, this GPU is designed for high-performance computing and AI workloads.
Released in 2024.03, this GPU represents cutting-edge technology in the GPU market.
Want to use this GPU?
Run your AI workloads on NVIDIA B200 with our cloud platform.