| Name |
Transistors |
Die size |
manufacturing |
| Tesla GV100 |
21.1 billion |
815 mm^2^ |
TSMC 12nm |
Key features
New Streaming Multiprocessor(New volta SM) Architecture for Deep learning
- New Tensor cores designed for deep learning
- With independent parallel integer and floating-point data paths
- Volta’s new independent thread scheduling
- A new combined L1 data cache and shared memory unit
Second-Generation NVIDIA NVLink
HBM2 Memory: Faster, Higher Efficiency
- 16GB HBM2 memory delivers 900GB/Sec peak memory bandwidth
- 95% memory bandwidth utilization
Enhanced Unified Memory and Addres Tranlation Services
Reference
