PNY NVIDIA H100 NVL 94GB HBM3 PCIe Accelerator

Brand: NVIDIA | Category: GPUs

SKU: NVH100NVLTCGPU-KIT | Part #: NVH100NVLTCGPU-KIT | MPN: NVH100NVLTCGPU-KIT

Contact for Pricing — Request a Quote

Request a Quote Contact Us

About the PNY NVIDIA H100 NVL 94GB HBM3 PCIe Accelerator

The PNY NVIDIA H100 NVL 94GB HBM3 PCIe Accelerator from NVIDIA is enterprise-grade GPUs hardware built for data centers. The PNY NVIDIA H100 NVL 94GB HBM3 PCIe Accelerator (MPN: NVH100NVLTCGPU-KIT) is built on NVIDIA's Hopper GPU architecture and is specifically engineered for large language model (LLM) inference and training at scale. The H100 NVL variant ships in a dual-GPU NVLink configuration delivering a combined 188GB of HBM3 memory and 7.8 TB/s aggregate memory bandwidth, enabling the acceleration of models that cannot fit within the memory envelope of a single GPU. The PCIe Gen5 host in Available in brand new condition. NVIDIA gpus products are trusted by enterprises and data centers worldwide. Available from Omnixon Global. Contact our sales team to request a quote.

Technical Specifications

BrandNVIDIA
CategoryGPUs
SKUNVH100NVLTCGPU-KIT
Part NumberNVH100NVLTCGPU-KIT
ConditionNew
Manufacturer Part NumberNVH100NVLTCGPU-KIT
GPU ArchitectureNVIDIA Hopper (GH100)
GPU Memory per Card94 GB HBM3
GPU Memory ConfigurationDual-GPU NVLink pair — 188 GB HBM3 combined
Memory Bandwidth per Card3.9 TB/s
Aggregate NVLink Pair Memory Bandwidth7.8 TB/s
NVLink Interconnect Bandwidth600 GB/s bidirectional (NVLink 4.0)
Host InterfacePCIe Gen5 x16
Tensor Core Generation4th Generation (FP8, FP16, BF16, TF32, FP64)
Transformer EngineYes (2nd Generation)
Supported PrecisionsFP8, FP16, BF16, TF32, FP32, FP64, INT8
Confidential ComputingYes (NVIDIA Hopper Confidential Computing)
Form FactorPCIe Dual-slot (per GPU card in NVLink pair)
Thermal DesignPassive cooling (requires adequate chassis airflow)
Max Thermal Design Power (per GPU)400 W
ECC Memory SupportYes
CUDA Compute Capability9.0
Launch QuarterQ1 2024