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
Dual-slot 10.5 inch PCIe card with passive heat sink and full-height L-bracket

NVIDIA

NVIDIA Tesla P100 16GB HBM2 PCIe Full Height L-Bracket graphics card

SKU CH-ZPSB-XMN4V

A dual-slot 10.5-inch PCIe 3.0 x16 accelerator with 3584 CUDA cores and 16 GB HBM2 memory for high-performance computing

  • InterfacePCI Express 3.0 x16
  • Chipset ManufacturerNVIDIA
  • CUDA Cores3584
  • Memory Size16GB
  • CoolerFanless
  • Form FactorPlug-in CardFull-height
  • 16 GB HBM2 memory delivers 4.7 TFLOPS double-precision compute
  • 3584 CUDA cores provide 9.3 TFLOPS single-precision performance
  • Passive cooling enables dense server deployment without fan noise
  • PCIe 3.0 x16 interface fits standard full-height slots
  • 250 W power envelope replaces half a rack of CPU nodes

Tesla P100 16GB HBM2 Accelerator

The NVIDIA Tesla P100 is a compute accelerator built for high-performance computing and deep learning workloads. It delivers 3584 CUDA cores and 16GB of HBM2 memory on a full-height PCIe card. The passive cooler requires directed chassis airflow to stay within thermal limits. It targets servers that need dense double-precision throughput.

PCIe 3.0 x16 and 250W power envelope

The card plugs into a PCI Express 3.0 x16 slot and draws up to 250W under load. The 4096-bit memory interface feeds the GP100 GPU at 9.3 TFLOPS single precision and 4.7 TFLOPS double precision. System power delivery and slot bandwidth set the practical ceiling for sustained compute. No auxiliary power connectors are present on the board.

Passive full-height installation

Installation means seating the dual-slot 10.5 inch card in a compatible full-height bracket slot with strong front-to-back airflow. The passive heatsink has no fan, so restricted air movement causes throttling or shutdown. Common issues are insufficient case ventilation, missing airflow ducting, or placing the card in a slot with inadequate clearance. Verify server fan profiles before deployment.

Highlights

  • 16 GB HBM2 memory delivers 4.7 TFLOPS double-precision compute
  • 3584 CUDA cores provide 9.3 TFLOPS single-precision performance
  • Passive cooling enables dense server deployment without fan noise
  • PCIe 3.0 x16 interface fits standard full-height slots
  • 250 W power envelope replaces half a rack of CPU nodes

Specifications

BrandNVIDIA
ModelTesla P100
InterfacePCI Express 3.0 x16
Chipset ManufacturerNVIDIA
CUDA Cores3584
Memory Size16GB
CoolerFanless
Form FactorPlug-in CardFull-height

Questions about this item

Should I pick the 16GB HBM2 version or is there a smaller memory option?

This card ships with 16GB of HBM2 memory on a 4096-bit interface; no lower-capacity variant is listed.

What does the 4.7 TFLOPS double-precision figure mean when the card is installed in a server?

It represents peak FP64 throughput of 4.7 TFLOPS at the 250W power limit, achievable when the workload keeps all 3584 CUDA cores saturated.

How does the PCIe 3.0 x16 interface cap real-world bandwidth?

The Gen3 x16 link provides up to 32 GB/s theoretical bandwidth, which limits host-to-device transfer rates for large datasets.

Also in this aisle

People compared these

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