#
Ai-Training
AI tools tagged Ai-Training.
AMD Instinct MI100
First-generation CDNA datacenter accelerator
AMD Instinct MI210
Single-die PCIe CDNA 2 accelerator
AMD Instinct MI250
Dual-die CDNA 2 accelerator, cut-down MI250X
AMD Instinct MI300A
APU combining CDNA 3 GPU and Zen 4 CPU cores
AMD Instinct MI300X
AMD flagship datacenter accelerator for large-model AI training
AMD Instinct MI355X
AMD's latest flagship AI training and inference accelerator
AMD Instinct MI450X
CDNA 5 flagship of the MI400 series, built for the Helios rack
AWS Trainium
AWS's first purpose-built training chip
AWS Trainium2
AWS's second-generation custom training chip
Baidu Kunlun II
Second-generation Baidu cloud AI chip
Biren BR100
Chinese chiplet-based general-purpose GPU for AI training
Cambricon MLU370-X8
Dual-die AI training and inference card
Cambricon MLU590
Cambricon's flagship large-model training chip
Cerebras WSE-2
Second-generation wafer-scale AI chip
Cerebras WSE-3
Largest chip ever built, with 4 trillion transistors
Google TPU 8t
Google's eighth-generation TPU, split off to focus purely on training
Google TPU v2
First TPU generation to support training, not just inference
Google TPU v3
Liquid-cooled third-generation TPU
Google TPU v4
Optically-reconfigurable TPU pod generation
Google TPU v5p
Google custom AI accelerator for training large models on Cloud TPU pods
Google TPU v6e (Trillium)
Fourfold performance-per-chip jump over TPU v5e
Graphcore Bow IPU
First wafer-on-wafer AI processor
Graphcore GC200 IPU
Second-generation Intelligence Processing Unit
Huawei Ascend 910
Huawei's first Ascend AI training chip
Huawei Ascend 910B
China's leading domestic AI training accelerator
Intel Data Center GPU Max 1100
PCIe variant of Intel's Ponte Vecchio datacenter GPU
Intel Gaudi
Habana Labs' first-generation training accelerator
Intel Gaudi 2
Second-generation Habana Labs training accelerator
Intel Gaudi 3
Intel third-generation AI training and inference accelerator
Microsoft Maia 100
Microsoft's first custom AI accelerator for Azure
Moore Threads MTT S4000
Chinese full-function GPU for AI training and inference
NVIDIA A100 40GB
Original 40GB Ampere datacenter GPU
NVIDIA A100 80GB
Ampere-generation datacenter GPU, still widely deployed
NVIDIA A800
China-market Ampere GPU with reduced NVLink bandwidth
NVIDIA B100
Drop-in Blackwell successor to H100 for HGX systems
NVIDIA B200
NVIDIA Blackwell flagship datacenter GPU for frontier model training
NVIDIA GB200 NVL72
Rack-scale Grace Blackwell superchip for trillion-parameter models
NVIDIA GB300 NVL72
Rack-scale Blackwell Ultra superchip with expanded memory
NVIDIA H100 PCIe
PCIe-form-factor Hopper datacenter GPU
NVIDIA H100 SXM
Hopper-architecture datacenter GPU for large-scale AI training
NVIDIA H800
China-market Hopper GPU with reduced NVLink bandwidth
NVIDIA RTX PRO 6000 Blackwell
Flagship Blackwell workstation GPU with 96GB VRAM
NVIDIA Rubin R100
Successor to Blackwell built for agentic-scale AI
NVIDIA Tesla P100
First HBM2 datacenter GPU, pre-dates Tensor Cores
NVIDIA V100 32GB PCIe
PCIe-form-factor Volta datacenter GPU
NVIDIA V100 SXM2
First-generation Tensor Core datacenter GPU
SambaNova SN30
Second-generation reconfigurable dataflow chip