| 04_knowledge_base/GPU | 89 |
| 04_knowledge_base/InferenceX | 57 |
| 02_companies/NVDA | 48 |
| 02_companies/AMD | 47 |
| 04_knowledge_base/vLLM | 46 |
| 04_knowledge_base/SGLang | 42 |
| 04_knowledge_base/NVIDIA B200 | 41 |
| 04_knowledge_base/MI355X | 36 |
| 04_knowledge_base/Large language model | 25 |
| 04_knowledge_base/Blackwell | 20 |
| 04_knowledge_base/GB200 NVL72 | 19 |
| 04_knowledge_base/H200 | 18 |
| 02_companies/DeepSeek | 17 |
| 04_knowledge_base/NVIDIA Collective Communications Library (NCCL) | 17 |
| 04_knowledge_base/MI325X | 15 |
| 04_knowledge_base/FP4 | 14 |
| 04_knowledge_base/GPT-OSS | 13 |
| 04_knowledge_base/H100 | 11 |
| 04_knowledge_base/DeepSeek R1 | 11 |
| 04_knowledge_base/FP8 | 11 |
| 04_knowledge_base/MI300X | 11 |
| 02_companies/GitHub | 11 |
| 04_knowledge_base/Hopper | 9 |
| 04_knowledge_base/ROCm | 9 |
| 04_knowledge_base/Expert Parallelism | 7 |
| 04_knowledge_base/Mixture of Experts | 7 |
| 04_knowledge_base/Llama | 7 |
| 04_knowledge_base/CUDA | 7 |
| 04_knowledge_base/PTX | 6 |
| 04_knowledge_base/Prefill | 6 |
| 04_knowledge_base/Disaggregated serving | 6 |
| 04_knowledge_base/MI355 | 5 |
| 04_knowledge_base/NVIDIA SASS Assembly | 5 |
| 04_knowledge_base/AMD AITER | 5 |
| 04_knowledge_base/PyTorch | 5 |
| 04_knowledge_base/Trainium | 4 |
| 02_companies/MSFT | 4 |
| 02_companies/OpenAI | 4 |
| 04_knowledge_base/MI300 | 4 |
| 04_knowledge_base/TPU | 4 |
| 04_knowledge_base/CDNA3 | 4 |
| 04_knowledge_base/Software Stack | 4 |
| 04_knowledge_base/SM100 | 4 |
| 04_knowledge_base/CDNA4 | 4 |
| 04_knowledge_base/Decode | 4 |
| 04_knowledge_base/Microsoft Azure | 3 |
| 02_companies/Crusoe | 3 |
| 04_knowledge_base/Hyperscaler | 3 |
| 02_companies/CRWV | 3 |
| 02_companies/Nebius | 3 |
| 04_knowledge_base/Prefix Caching | 3 |
| 04_knowledge_base/HGX | 3 |
| 02_companies/Together AI | 3 |
| 03_industries/Neocloud產業 | 3 |
| 02_companies/Tensorwave | 3 |
| 02_companies/GOOG | 3 |
| 02_companies/ORCL | 3 |
| 04_knowledge_base/Disaggregated prefill | 3 |
| 04_knowledge_base/Multi-Token Prediction | 3 |
| 04_knowledge_base/CI-CD | 3 |
| 04_knowledge_base/GB200 | 3 |
| 04_knowledge_base/Arithmetic intensity | 2 |
| 04_knowledge_base/InfiniBand | 2 |
| 04_knowledge_base/MI325 | 2 |
| 02_companies/Vultr | 2 |
| 04_knowledge_base/Grace Blackwell | 2 |
| 04_knowledge_base/KV cache | 2 |
| 04_knowledge_base/Blackwell Ultra | 2 |
| 02_companies/SMCI | 2 |
| 04_knowledge_base/Tensor Parallelism | 2 |
| 04_knowledge_base/CPU | 2 |
| 04_knowledge_base/DeepEP | 2 |
| 04_knowledge_base/Data Parallelism | 2 |
| 04_knowledge_base/Model parallelism | 2 |
| 04_knowledge_base/Expected Average Power | 2 |
| 04_knowledge_base/Ethernet | 1 |
| 02_companies/Arista | 1 |
| 02_companies/Inspire Robots | 1 |
| 04_knowledge_base/Kubernetes | 1 |
| 04_knowledge_base/AI factory | 1 |
| 04_knowledge_base/SLA | 1 |
| 04_knowledge_base/Goodput | 1 |
| 02_companies/xAI | 1 |
| 04_knowledge_base/AWS | 1 |
| 02_companies/Scale AI | 1 |
| 04_knowledge_base/NVIDIA Dynamo | 1 |
| 04_knowledge_base/AI Accelerator | 1 |
| 04_knowledge_base/WhiteBox Ethernet | 1 |
| 04_knowledge_base/NVLink | 1 |
| 02_companies/Aligned Data Centers | 1 |
| 04_knowledge_base/Multi-head Latent Attention | 1 |
| 04_knowledge_base/Flash memory | 1 |
| 04_knowledge_base/TFLOPS | 1 |
| 04_knowledge_base/MXFP4 | 1 |
| 04_knowledge_base/Chiller | 1 |
| 04_knowledge_base/NVIDIA B300 | 1 |
| 04_knowledge_base/XPU | 1 |
| 04_knowledge_base/GB300 NVL72 | 1 |
| 04_knowledge_base/Thermal Design Power | 1 |
| 04_knowledge_base/AMDGPU | 1 |
| 04_knowledge_base/Inference engine | 1 |
| 04_knowledge_base/Sequence length | 1 |
| 04_knowledge_base/Control Plane | 1 |
| 04_knowledge_base/Bare Metal | 1 |
| 04_knowledge_base/Batch size | 1 |