NVIDIA
NVIDIA H100 NVL 94GB
Dual-slot PCIe Hopper card with 94 GB HBM3, deployed in NVLink-bridged pairs for large language model inference.
| 재고 상태 | 재고 있음 |
|---|---|
| SKU | AXL-GPU-H100NVL |
| 최소 주문 수량 | 2 units |
| 상태 | 신품 |
| 가격 | 견적 문의 |
개요
The H100 NVL is the PCIe member of the Hopper family aimed at inference on large language models. Each card carries 94 GB of HBM3, and cards are deployed in NVLink-bridged pairs so a model can address 188 GB across the pair. It fits standard PCIe server chassis, which makes it a practical route into Hopper for buyers who cannot host an SXM baseboard.
Because the cards are matched and bridged, AXISLINK quotes them in pairs with both serial numbers recorded on the proforma invoice. Supply comes from verified channels, and we deliver the cards alone or already fitted and tested inside a validated server platform with the correct PCIe topology and airflow. Export paperwork and freight leave from Hong Kong, and end-user details may be requested for controlled data-center parts.
사양
| Architecture | NVIDIA Hopper |
|---|---|
| Memory | 94 GB HBM3 per card, 188 GB per NVLink-bridged pair |
| Memory bandwidth | 3.9 TB/s per card |
| Multi-Instance GPU | Up to 7 instances per card |
| Form factor | PCIe 5.0 x16, dual-slot full-height full-length, passively cooled |
|---|---|
| NVLink | 600 GB/s bridge between paired cards |
| Supplied as | Matched pairs with NVLink bridges |
| Board power | Configurable, 350 W to 400 W per card |
|---|---|
| Cooling | Passive, requires server chassis airflow |
견적 요청
NVIDIA H100 NVL 94GB 견적 요청입니다. 보통 24시간 이내에 정식 견적송장으로 회신합니다.
GPU 및 가속기의 다른 제품
카테고리 보기
NVIDIA
NVIDIA H200 141GB SXM
Hopper SXM module with 141 GB HBM3e for large-model training and memory-bound inference on HGX baseboards.
NVIDIA
NVIDIA L40S 48GB
Ada Lovelace data-center card with 48 GB GDDR6 ECC for inference, fine-tuning, rendering and virtual workstations.
NVIDIA
NVIDIA H100 80GB
Hopper-architecture data-center GPU for large-scale AI training and inference. PCIe and SXM variants.