| 02_companies/AMD | 227 |
| 02_companies/NVDA | 97 |
| 04_knowledge_base/GPU | 89 |
| 04_knowledge_base/MI300X | 87 |
| 04_knowledge_base/PyTorch | 71 |
| 04_knowledge_base/H100 | 69 |
| 04_knowledge_base/H200 | 52 |
| 04_knowledge_base/InfiniBand | 27 |
| 04_knowledge_base/Ethernet | 22 |
| 04_knowledge_base/FP8 | 18 |
| 04_knowledge_base/MLPerf | 18 |
| 02_companies/META | 17 |
| 03_industries/Neocloud產業 | 14 |
| 04_knowledge_base/BF16 | 14 |
| 04_knowledge_base/Llama | 12 |
| 04_knowledge_base/HBM | 11 |
| 02_companies/Switch | 11 |
| 04_knowledge_base/NVIDIA Collective Communications Library (NCCL) | 9 |
| 04_knowledge_base/GPT-3 | 9 |
| 04_knowledge_base/FLOP | 9 |
| 04_knowledge_base/FP16 | 8 |
| 04_knowledge_base/NVLink | 7 |
| 04_knowledge_base/Large language model | 7 |
| 04_knowledge_base/ROCm Communication Collectives Library | 7 |
| 04_knowledge_base/L2 cache | 7 |
| 02_companies/Tensorwave | 6 |
| 04_knowledge_base/CUDA | 6 |
| 04_knowledge_base/Fully Sharded Data Parallel | 6 |
| 02_companies/Together AI | 6 |
| 04_knowledge_base/Transformer architecture | 6 |
| 04_knowledge_base/NIC | 6 |
| 04_knowledge_base/Software Stack | 6 |
| 04_knowledge_base/Transistor | 6 |
| 04_knowledge_base/ROCm | 5 |
| 04_knowledge_base/Hyperscaler | 5 |
| 04_knowledge_base/Flash memory | 5 |
| 04_knowledge_base/AWS | 5 |
| 04_knowledge_base/NVIDIA Spectrum-X | 4 |
| 04_knowledge_base/Sparsity | 4 |
| 04_knowledge_base/Tensor Core | 4 |
| 02_companies/OpenAI | 4 |
| 04_knowledge_base/RoCEv2 | 4 |
| 04_knowledge_base/GCP | 4 |
| 04_knowledge_base/CI-CD | 4 |
| 04_knowledge_base/Batch size | 4 |
| 04_knowledge_base/MI300 | 3 |
| 04_knowledge_base/Blackwell | 3 |
| 04_knowledge_base/AMD HIP | 3 |
| 04_knowledge_base/Tensor Parallelism | 3 |
| 04_knowledge_base/Tomahawk 5 | 3 |
| 04_knowledge_base/ChatGPT | 3 |
| 04_knowledge_base/400G | 3 |
| 04_knowledge_base/Quantum-2 | 3 |
| 04_knowledge_base/ClusterMAX | 3 |
| 02_companies/Nebius | 2 |
| 04_knowledge_base/Feed-forward Network | 2 |
| 04_knowledge_base/Scale-out networking | 2 |
| 04_knowledge_base/Pollara | 2 |
| 04_knowledge_base/Anthropic Claude | 2 |
| 02_companies/Crusoe | 2 |
| 02_companies/Lambda | 2 |
| 04_knowledge_base/HBM3 | 2 |
| 04_knowledge_base/2-of-4 Sparsity | 2 |
| 02_companies/Arista | 2 |
| 02_companies/AVGO | 2 |
| 04_knowledge_base/torch.compile | 2 |
| 04_knowledge_base/Reference Design | 2 |
| 04_knowledge_base/CoWoS | 1 |
| 04_knowledge_base/800G | 1 |
| 04_knowledge_base/IC design verification | 1 |
| 02_companies/GOOG | 1 |
| 04_knowledge_base/ConnectX-8 | 1 |
| 04_knowledge_base/Triton | 1 |
| 04_knowledge_base/SIMT | 1 |
| 02_companies/Anthropic | 1 |
| 02_companies/xAI | 1 |
| 04_knowledge_base/Expert Parallelism | 1 |
| 04_knowledge_base/WhiteBox Ethernet | 1 |
| 02_companies/MSFT | 1 |
| 04_knowledge_base/MI325X | 1 |
| 04_knowledge_base/液冷散熱 | 1 |
| 04_knowledge_base/ASIC | 1 |
| 04_knowledge_base/MI355X | 1 |
| 04_knowledge_base/Pipeline parallelism | 1 |
| 04_knowledge_base/NVSwitch | 1 |
| 02_companies/GitHub | 1 |
| 04_knowledge_base/Grok | 1 |
| 04_knowledge_base/Data Parallelism | 1 |
| 04_knowledge_base/Scale-up network | 1 |
| 04_knowledge_base/Checkpointing | 1 |
| 04_knowledge_base/Network Topology | 1 |
| 04_knowledge_base/CPU | 1 |
| 04_knowledge_base/Fat Tree | 1 |
| 04_knowledge_base/CUTLASS | 1 |
| 04_knowledge_base/NVIDIA Nsight Systems | 1 |
| 04_knowledge_base/EFAv3 | 1 |
| 04_knowledge_base/BlueField-3 | 1 |
| 04_knowledge_base/EFAv2 | 1 |
| 04_knowledge_base/ConnectX-7 | 1 |
| 04_knowledge_base/Direct Current | 1 |
| 04_knowledge_base/GB200 | 1 |
| 04_knowledge_base/Thermal Design Power | 1 |
| 04_knowledge_base/Congestion Control | 1 |
| 04_knowledge_base/QM9700 | 1 |
| 04_knowledge_base/200G | 1 |
| 04_knowledge_base/SN5600 | 1 |
| 04_knowledge_base/NVIDIA Nsight Compute | 1 |
| 04_knowledge_base/NVIDIA B200 | 1 |
| 04_knowledge_base/Distributed Data Parallel | 1 |