Supported compute modules

Turing Pi V2 Board contains four vertical 260-pin SO-DIMM sockets. This allows us to support not only the Raspberry Pi compute Module (with CM4 Adapter Board), directly attachable Nvidia Jetson compute Modules, but also our upcoming Turing Pi compute module!

You are not limited to using only one model for every slot, or even one vendor. Feel free to mix the Compute Modules however you like.

You may want an all-purpose Kubernetes cluster made out of 4 Raspberry Pi modules, but you can replace one or more RPI modules with Nvidia Jetson and suddenly your cluster has Machine Learning capable nodes.


Supported Modules

Vendor Model CPU Cores RAM, GB
Turing Pi RK1 Rockchip RK3588 8 8, 16, 32
Raspberry Pi CM4 Broadcom BCM2711 4 2, 4, 8
Nvidia Jetson Nano Quad-core ARM Cortex-A57 MPCore 4 2, 4
Nvidia Jetson TX2 NX Dual-core NVIDIA Denver 2
Quad-core ARM A57 Complex
6 4
Nvidia Jetson Xavier NX NVIDIA Carmel ARM®v8.2 6 8


However, keep in mind that not all Compute Modules offer the same features.


    PCIe Lanes        
Vendor Model Gen 2 Gen 3 Gen 4

NVMe speed

up to*

Mini PCIe

Gen 2


USB 3.0

Raspberry CM4 x1     N/A Yes Yes Yes
Turing RK1 x1 x4   4 GB/s Yes Yes Yes
Nvidia Jetson TX2 NX x1 x2   2 GB/s Yes Yes Yes
Nvidia Jetson Xavier NX   x1 x4 8 GB/s Yes Yes Yes
Nvidia Jetson Nano x1     1 GB/s N/A N/A N/A
  • Mini PCIe top side of the board
  • M.2 M-key on the bottom side of the board.

Raspberry Pi Compute Module Adapter Board


Nvidia Jetson Compute Module



Additional modules tested

In addition to the Raspberry Compute Modules, we have evaluated various alternative options that could serve as drop-in replacements. However, it's important to note that most of these alternatives utilize different types of chips that may not be compatible with the peripheral devices on the Turing Pi V2. This means that if you choose to use a different module, it may not work seamlessly with the existing hardware on the Turing Pi V2, potentially affecting the performance or functionality of the system as a whole.


We had the opportunity to test the Soquartz CM 2GB and found that it does successfully boot up. However, we encountered an issue where only the network was visible in the operating system. We used DietPi OS, specifically designed for this compute module, during our testing. Despite this limitation, it is still possible to use the Soquartz CM 2GB as a compute module in a Kubernetes cluster.

If you choose to purchase the module without the built-in eMMC, the SD card on the CM4 adapter board can be utilized, and the operating system can be booted from it. However, if you opt for the version with the eMMC extension, you will need to use the USB eMMC adapter tool provided by Pine64 to flash the operating system onto the module.




Was this article helpful?

40 out of 44 found this helpful
Have more questions? Submit a request

Comments (0 comments)

Article is closed for comments.