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-length dual-slot graphics card with matte black cooler and gold PCIe 3.0 x16 connector

NVIDIA

NVIDIA Tesla M40 839949-001 graphics card

SKU CH-5K2M-BYFG4

A GPU for machine learning with 3072 CUDA cores, GDDR5 memory on a 384-bit bus and a 1140 MHz boost clock

  • Memory TypeGDDR5
  • 3072 CUDA cores accelerate machine-learning workloads
  • Boost clock reaches 1140 MHz for fast compute throughput
  • 6 GHz GDDR5 memory delivers high bandwidth
  • 384-bit memory interface supports large data sets
  • PCI Express 3.0 x16 connection fits standard server slots

Tesla M40 for Machine Learning Workloads

The NVIDIA Tesla M40 is a compute-focused graphics card built for machine learning tasks. It delivers 3072 CUDA cores and a boost clock of 1140 MHz. The board connects via PCI Express 3.0 x16 and uses GDDR5 memory on a 384-bit interface clocked at 6 GHz.

Suited for Inference Not Training at Scale

This card suits engineers running inference or smaller model development on a single GPU. The memory capacity fits many classification and language models. Teams needing multi-GPU training throughput or larger model support should look at higher-capacity accelerators.

3072 CUDA Cores Define Parallel Throughput

The 3072 CUDA cores set the parallel compute ceiling for every workload on this card. If your kernels saturate this core count the M40 will perform well. If you need substantially more concurrent threads a different accelerator is required.

Highlights

  • 3072 CUDA cores accelerate machine-learning workloads
  • Boost clock reaches 1140 MHz for fast compute throughput
  • 6 GHz GDDR5 memory delivers high bandwidth
  • 384-bit memory interface supports large data sets
  • PCI Express 3.0 x16 connection fits standard server slots

Specifications

Memory TypeGDDR5

Questions about this item

How many CUDA cores does the Tesla M40 provide for parallel compute?

The card has 3072 CUDA cores running at a boost clock of ~1140 MHz for high-throughput workloads.

What memory bandwidth does the 384-bit GDDR5 interface deliver?

Memory runs at 6 GHz on a 384-bit bus, giving ample bandwidth for large-model inference.

Which PCI Express slot is required for full-speed operation?

A PCIe 3.0 x16 slot provides the full 16-lane link the board is designed for.

Also in this aisle

People compared these

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