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 PCIe card, silver-grey metal shroud, single blower fan, 24 GB GDDR5

NVIDIA

NVIDIA 900-2g600-0310-030 Tesla M40 24GB GDDR5 graphics card

SKU CH-9XMN-VJDNC MPN 900-2g600-0310-030

A 24 GB GDDR5 GPU computing processor with 3072 CUDA cores on a PCI Express 3.0 x16 bus delivering 288 GBps memory bandwidth

  • Memory Size24GB
  • Memory TypeGDDR5
  • 24 GB GDDR5 memory handles large datasets
  • 3072 CUDA cores accelerate parallel workloads
  • 288 GBps bandwidth feeds compute engines fast
  • PCI Express 3.0 x16 slot fits standard servers
  • Maxwell architecture with CUDA, Dynamic Parallelism, Hyper-Q

24 GB GDDR5 GPU compute card

The NVIDIA Tesla M40 provides 24 GB of GDDR5 memory and 3072 CUDA cores on a PCI Express 3.0 x16 interface. It targets deep learning training, inference and high-performance computing workloads that require large memory capacity. The Maxwell architecture delivers parallel processing power for data-centre and workstation environments.

Passive cooling for rack density

The card uses a passive heatsink designed for server airflow, so it adds no fan noise to the chassis. Thermal performance depends on the host system supplying sufficient directed airflow across the 26.7 cm length. Power consumption is managed by the platform within standard PCIe limits.

CUDA software stack enables acceleration

NVIDIA CUDA toolkits and libraries let developers compile and optimise code for the 3072 cores without rewriting algorithms. Supported frameworks include TensorFlow, PyTorch and custom HPC applications that leverage the 288 GBps memory bandwidth. The PG600 board design ensures compatibility with qualified HP and OEM server platforms.

Highlights

  • 24 GB GDDR5 memory handles large datasets
  • 3072 CUDA cores accelerate parallel workloads
  • 288 GBps bandwidth feeds compute engines fast
  • PCI Express 3.0 x16 slot fits standard servers
  • Maxwell architecture with CUDA, Dynamic Parallelism, Hyper-Q

Specifications

BrandNVIDIA
Model900-2g600-0310-030
Memory Size24GB
Memory TypeGDDR5

Questions about this item

How is the PCI Express 3.0 x16 card seated, and what step do builders often miss?

Insert the card fully into a PCIe 3.0 x16 slot until the retention latch clicks, then secure the bracket screw; the step often missed is connecting the required auxiliary power cables before closing the case.

What maintenance will the Tesla M40 24GB GDDR5 need over time?

Periodically remove dust from the heatsink fins and fan blades using compressed air; the thermal paste under the cooler may need replacement after several years of continuous compute workloads.

Which limit appears first when the 3072 CUDA core GPU runs sustained loads at 288 GBps memory bandwidth?

The thermal envelope of the dual-slot cooler typically reaches its limit first, causing the card to throttle clock speeds before the 24 GB GDDR5 memory or PCIe 3.0 x16 bus becomes saturated.

Also in this aisle

People compared these

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