Specifications
Graphics Chipset NVIDIA H100
Edition PNY
Manufacturing process 4 nm
Microarchitecture Hopper
GPU Name
Cores & Clocks
NVIDIA CUDA Cores
Core Clock
Boost Clock
Video Memory (VRAM)
Memory Size 80 GB
Memory Type HBM3
Memory Clock
Memory Bit Rate
Memory Bandwidth 2000 GB/s
ECC Technology No
Cooling
Cooling Passive
I/O & Connectivity
Interface PCIe
Interface Bandwidth 128 GB/s
Graphics Output
Visuals
Multi GPU Support 2-Way NVLink
NVLink Support N/A
Total NVLink Bandwidth 600 GB/s
Microsoft DirectX Support
HDCP Ready No
Multi Monitor Support
Concurent Users N/A
H.264 1080p30 Streams N/A
Maximum Digital Resolution
Maximum VGA Resolution
Supported Graphics APIs
Compute Performance
Supported Compute APIs
Single Precision (FP32) Processing Yes
Single Precision (FP32) Performance 51 teraFLOPS
Double Precision (FP64) Processing Yes
Double Precision (FP64) Performance 26 teraFLOPS
Tensor Processing Yes
Deep Learning (Tensor) Performance 756 teraFLOPS
NVIDIA Tensor Cores
Integer Operations (INT8) 3026 TOPS (Tensor) (Tera-Operations per Second)
Ray-Tracing
NVIDIA RT Cores
RTX-OPS
Rays Cast
Power & Thermals
Graphics Card Power Connectors
Board Power 300~350 W
Minimum Recommended PSU
Maximum GPU Temperature
Physical
Form Factor Dual Slot
Low Profile Compatible No
Low Profile Support
Dimensions
Package Type Retail
Additional Information
Scan Code LN129414
Model Number TCSH100PCIE-PB