Beranda › Chip › NVIDIA H100

NVIDIA H100

Chip GPU Pengiriman

The GPU of the ChatGPT era — the unit of account for the 2023–25 AI build-out.

Anda sedang melihat entri chip: apa itu prosesor ini, spesifikasi utamanya, siapa yang merancang, memproduksi, dan memasok komponennya, serta di mana chip ini digunakan.

Hopper's H100 became the defining commodity of the AI boom: clusters were sized in H100s, startups raised funding denominated in them, and export policy was written around them. Its Transformer Engine pairs fourth-generation tensor cores with FP8 precision, roughly tripling A100 training throughput per chip.

xAI's Colossus assembled roughly 100,000 of them in 122 days; Meta, Microsoft, OpenAI, Google and every major cloud deployed them in the hundreds of thousands. Blackwell succeeded it in 2025, but H100/H200 fleets still carry a large share of world AI compute.

Spesifikasi & fakta utama

Setiap angka diberi label berdasarkan sumbernya: spesifikasi vendor, klaim vendor, dilaporkan, estimasi, atau resmi. "Tidak diungkapkan secara publik" berarti tepat seperti itu — kami tidak pernah mengarang nilai.

Overview

Designer NVIDIA

Architecture Hopper (GH100)

Announced March 2022 (GTC); shipping volume from early 2023

Status Shipping — succeeded by Blackwell at the frontier

Manufacturing

Foundry / process TSMC 4N (custom 5 nm-class)Vendor spec

Transistors 80 billionVendor spec

Die size 814 mm²Vendor spec

Packaging CoWoS-S with 5 or 6 HBM stacksVendor spec

Memori

Memory 80 GB HBM3 (SXM); 94 GB HBM3 (NVL); 80 GB HBM2e (PCIe)Vendor spec

Memory bandwidth 3.35 (SXM) / 3.9 (NVL) TB/sVendor spec

Komputasi

TF32 tensor (dense) 495 TFLOPSVendor spec NVIDIA datasheets often headline throughput with structured sparsity; figures here are dense unless noted.

BF16/FP16 tensor (dense) 990 TFLOPSVendor spec

FP8 tensor (dense) 1,979 TFLOPSVendor spec FP8 via the Transformer Engine was Hopper's signature feature

INT8 tensor (dense) 1,979 TOPSVendor spec

FP64 tensor 67 TFLOPSVendor spec

AI Performance

MLPerf Set the reference results for training and inference benchmarks across 2023–24 submissionsMLPerf

Daya

TDP up to 700 (SXM) / 350 (PCIe) WVendor spec

Physical

Form factors SXM5, PCIe, NVL (paired PCIe)

Interconnect NVLink 4 — 900 GB/s; PCIe 5.0Vendor spec

Harga

Launch price Not publicly disclosed

Market price ~$25,000–40,000 per GPU at the 2023–24 peakReported

Cloud rental Roughly $2–7 per GPU-hour across major and specialist clouds during 2025 — falling steadilyReported varies with commitment and provider

Bagaimana ini terhubung ke bagian lain dari stack

Tautan catatan ini sekilas — klik grup untuk langsung menuju ke sana. +11 lebih banyak grup di bawah.

Dirancang oleh

NVIDIA

Diproduksi oleh

TSMC

Node proses

TSMC N5 / 4N (5 nm class)

Pengemasan

CoWoS

Penyedia pengemasan

TSMC

Teknologi memori

HBM3

Memori disediakan oleh

SK hynix

Interkoneksi & jaringan

NVLink & NVSwitch InfiniBand

Ditemukan di dalam

Dell PowerEdge XE9680/XE9712 Supermicro NVL72 & 4U Liquid-Cooled Systems NVIDIA HGX H100/H200

Menjalankan

Triton (OpenAI) PyTorch Megatron-Core vLLM

Didukung oleh

CUDA

Digunakan untuk melatih

Llama 3.1 405B Llama 4 Grok 3

Tersedia dari

Amazon Web Services Microsoft Google (Alphabet) Oracle CoreWeave Lambda

Digunakan di

xAI Colossus

Dasar untuk

NVIDIA H800

Digantikan oleh

NVIDIA H200

Penerus dari

NVIDIA A100

Bersaing dengan

AMD Instinct MI300X Google TPU v5p Groq LPU AWS Trainium2 Intel Gaudi 3 Huawei Ascend 910C

Terkait

Export Controls & the Chip War Training & Pre-training MLPerf

Sumber

NVIDIA H100 page ↗ · Hopper architecture whitepaper ↗

Rekaman ditinjau per 2026-01-15 — lihat Metodologi.