We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in
NVIDIA Tesla P100 dual-slot PCIe card with 16 GB HBM2 memory

NVIDIA

NVIDIA Tesla P100 16 GB HBM2 PCIe 3.0 x16 Workstation Graphics Card

SKU CH-A6TM-TEGKZ

16 GB HBM2 accelerator with 3584 CUDA cores and 732 GB/s bandwidth for HPC and deep learning workloads

  • InterfacePCI Express 3.0 x16
  • Chipset ManufacturerNVIDIA
  • GPUTesla P100
  • Core Clock1190 MHz
  • CUDA Cores3584
  • Memory Clock715 MHz
  • 16 GB HBM2 memory with 732 GB/s bandwidth accelerates data-intensive workloads
  • 3584 CUDA cores deliver 9.3 teraFLOPS single-precision compute for HPC tasks
  • Pascal architecture provides 4.7 teraFLOPS double-precision performance for scientific computing
  • PCI Express 3.0 x16 interface ensures high-throughput host connectivity
  • Dual-slot 267 mm form factor fits standard workstation chassis

NVIDIA Tesla P100 16 GB HBM2 Accelerator

This workstation graphics card delivers 3584 CUDA cores and 16 GB of HBM2 memory on a 4096-bit interface. It connects via PCI Express 3.0 x16 and has a maximum power draw of 250 W. The Pascal architecture provides 9.3 teraFLOPS single-precision and 4.7 teraFLOPS double-precision performance.

Compute-Focused Users Not Needing Display Output

The card targets high-performance computing, deep learning and scientific simulation workloads that benefit from 732 GB/s memory bandwidth. It is not suited for users who require video outputs for monitor connectivity or consumer graphics features. Buyers needing certified drivers for specific ISV applications should verify support independently.

16 GB HBM2 Memory Capacity

The 16 GB HBM2 frame buffer with 4096-bit width determines the maximum model and dataset size that fits in local memory. This capacity is the primary factor when deciding if the card can handle a given workload without host memory paging.

Highlights

  • 16 GB HBM2 memory with 732 GB/s bandwidth accelerates data-intensive workloads
  • 3584 CUDA cores deliver 9.3 teraFLOPS single-precision compute for HPC tasks
  • Pascal architecture provides 4.7 teraFLOPS double-precision performance for scientific computing
  • PCI Express 3.0 x16 interface ensures high-throughput host connectivity
  • Dual-slot 267 mm form factor fits standard workstation chassis

Specifications

BrandNVIDIA
Model900-2H400-0000-000
InterfacePCI Express 3.0 x16
Chipset ManufacturerNVIDIA
GPUTesla P100
Core Clock1190 MHz
CUDA Cores3584
Memory Clock715 MHz
Memory Size16GB
Memory Interface4096-bit
Memory TypeHBM2
DirectXDirectX 12.1
OpenGLOpenGL 4.6
System RequirementsMax Power Consumption 250W
Dimensions (L x H)Length: 267 mm(10.5 inches)
Slot WidthDual Slot

Questions about this item

What does 9.3 teraFLOPS single-precision performance mean for my workloads?

It means the GPU can theoretically execute 9.3 trillion single-precision floating-point operations per second, which determines peak throughput for compute kernels and deep-learning training tasks.

What bandwidth limit does the PCI Express 3.0 x16 interface impose?

The PCI Express 3.0 x16 link caps host-to-device transfer rates at roughly 16 GB/s, so data movement between system memory and the 16 GB HBM2 frame buffer can become the bottleneck in data-heavy pipelines.

Will this 267 mm dual-slot card fit in my chassis?

The board is 267 mm long and occupies two expansion slots, so it requires a case with at least 267 mm of clear length and two adjacent rear brackets; verify your chassis clearance and power-supply cables before installing.

Also in this aisle

People compared these

Same shelf, same checkout — eight coins and a 30-minute rate lock.