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
Full-height dual-slot GPU card with passive cooler and gold PCIe 3.0 x16 edge connector

NVIDIA

NVIDIA Tesla P40 24GB GDDR5 GPU Accelerator Card Dual PCIe 3.0 x16

SKU CH-ZGQ8-52YBH

24 GB GDDR5 GPU accelerator delivering 12 teraFLOPS single-precision and 47 TOPS INT8 for server inference workloads

  • Memory TypeGDDR5
  • 3840 CUDA cores deliver 12 teraFLOPS single-precision compute
  • 47 TOPS INT8 throughput accelerates inference workloads
  • 24 GB GDDR5 memory with 346 GB/s bandwidth handles large models
  • PCI Express 3.0 x16 interface fits standard server slots
  • 250 W maximum power fits dense data-centre deployments

NVIDIA Tesla P40 24GB GDDR5 Accelerator

This NVIDIA Tesla P40 is a GPU accelerator card built for server deployment. It delivers 12 teraFLOPS of single-precision performance and 47 TOPS for integer operations. The card targets data centre workloads such as inference and high-performance computing. It is not intended for home PC use.

24GB GDDR5 Memory and PCIe 3.0 x16 Interface

The card provides 24 GB of GDDR5 memory with 346 GB/s of bandwidth. It connects through a PCI Express 3.0 x16 system interface. These two specifications determine whether the accelerator fits your server platform and memory requirements.

250 W Power Draw in Dual-Slot Form Factor

Maximum power consumption is rated at 250 W within a 4.4 H x 10.5 L dual-slot full-height design. Server chassis airflow handles thermal dissipation for this form factor. The hardware includes ECC protection and a page migration engine for reliability.

Highlights

  • 3840 CUDA cores deliver 12 teraFLOPS single-precision compute
  • 47 TOPS INT8 throughput accelerates inference workloads
  • 24 GB GDDR5 memory with 346 GB/s bandwidth handles large models
  • PCI Express 3.0 x16 interface fits standard server slots
  • 250 W maximum power fits dense data-centre deployments

Specifications

Memory TypeGDDR5

Questions about this item

How does the 12 TeraFLOPS single-precision figure translate to actual server workloads?

The 12 TeraFLOPS rating represents peak single-precision compute throughput from the Pascal architecture and 3840 CUDA cores, enabling high-density inference and HPC tasks in supported server environments.

What does the PCI Express 3.0 x16 system interface limit in practice?

The PCIe 3.0 x16 link caps host-to-device bandwidth at the Gen3 x16 rate, which can become a bottleneck for large model transfers compared to newer interconnects.

Will the 4.4 H by 10.5 L dual-slot form factor fit in a standard 2U server chassis?

The full-height, dual-slot dimensions of 4.4 inches high by 10.5 inches long require a compatible server chassis with adequate clearance and airflow for 250 W thermal design power.

Also in this aisle

People compared these

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