NVIDIA
NVIDIA Tesla M10 graphics card 4 GPUs 32 GB GDDR5
Multi-GPU accelerator with 32 GB GDDR5 memory supporting 64 virtual desktops per board and 128 per server
- Memory TypeGDDR5
- Four GPUs deliver 64 virtual desktops per board
- 128 virtual desktops per server maximise user density
- 32 GB GDDR5 memory supports demanding workloads
- Passive cooler fits dual-slot server chassis
- NVIDIA GRID software enables scalable virtualization
Multi-GPU virtualization board
The NVIDIA Tesla M10 is a passive plug-in card that integrates four GPUs and 32 GB of GDDR5 memory. It is built for data-centre servers that run NVIDIA GRID software to deliver virtual desktops and applications. Each board can serve up to 64 desktops, and a server can scale to 128 desktops.
GRID density limits
The accelerator supports 64 desktops per board and 128 desktops per server when paired with GRID software. This ceiling defines how many concurrent users a single card or host can serve without degrading the experience. Planning capacity means matching the number of boards to the expected peak desktop count.
Passive dual-slot cooler
A passive heatsink occupies two expansion slots and relies on server chassis airflow for cooling. The design has no moving parts, which reduces maintenance in rack environments. Installers must ensure the host provides sufficient directed airflow across the dual-slot profile.
Highlights
- Four GPUs deliver 64 virtual desktops per board
- 128 virtual desktops per server maximise user density
- 32 GB GDDR5 memory supports demanding workloads
- Passive cooler fits dual-slot server chassis
- NVIDIA GRID software enables scalable virtualization
Specifications
| Memory Type | GDDR5 |
|---|
Questions about this item
How many virtual desktops can a single M10 board support?
A single M10 board supports up to 64 virtual desktops when paired with NVIDIA GRID software, enabling high user density for VDI deployments.
What cooling method does the card use and how does it affect server installation?
The card uses a passive cooler with no onboard fan, so it relies on directed server airflow and fits in a standard dual-slot form factor for rackmount chassis.
What is the total onboard memory and how is it distributed across the GPUs?
The board provides 32 GB of GDDR5 memory shared across four GPUs, giving each GPU 8 GB for virtual desktop or application workloads.
New to this? How to buy a graphics card with Bitcoin — what to check, what it costs to pay, and what is never asked for.
Also in this aisle
People compared these
Same shelf, same checkout — eight coins and a 30-minute rate lock.
minisforum MF-SRP01 SSD Heatsink accessories barebone
- SSD Heatsink
Intel HNS2600BPS24R 2U barebone pc dual Xeon DDR4 2.80 TB C622
- Xeon Bronze Xeon Go…
- DDR4-2933/PC4-23466



