Raspberry Pi Powered Compute Blade Makes the Cut
We’ve been tracking this project since mid-2021 and have spent plenty of time on it. Ivan Kuleshov’s Compute Blade is a thin PCB with numerous storage options for a Raspberry Pi Compute Module 4 (or compatible). Kickstarter for Kuleshov We have beaten our $522,209 fundraising goal and have reached $673,365 at the time of writing.
The Compute Blade is a rack-mountable carrier board for the Raspberry Pi Compute Module 4 designed for high-density clusters. There’s a lot of functionality packed into the PCB, but the eye is drawn to a red anodized aluminum heatsink that fits into a Compute Module 4 (or compatible) and provides a passive means of cooling the Pi. This is useful if you want to overclock.
row 0 – cell 0 | Uptime Compute Blade Basic | Uptime Compute Blade TPM | Uptime Compute Blade Development |
Raspberry Pi CM4 connector | yes | yes | yes |
M.2 (M key for up to 22110 NVMe drives) | yes | yes | yes |
Gigabit Ethernet with PoE | yes | yes | yes |
UARTs | yes | yes | yes |
Blade Header Calculation | yes | yes | yes |
stealth mode | yes | yes | yes |
LED | yes | yes | yes |
USB-A port | no | yes | yes |
TPMs | no | yes | yes |
HDMI | no | no | yes |
USB-C (bootloader update) | no | no | yes |
Micro SD card slot | no | no | yes |
Wi-Fi, BT, nRPIBOOT | locked | locked | switchable |
The PCB size is 27 x 4 x 1.6 cm and Kuleshov has optimized the board for different functions. Going from the Ethernet port on the left, we have Gigabit Ethernet with Power over Ethernet (PoE) support, and our development unit has his HDMI port. Next we reach the anodized heat sink which is secured using T7 hex screws. after that. There is a USB-A port, followed by an expansion M.2 slot. This slot supports his NVMe drives from 2230 to 22110 or can be used with other modules such as Google’s Coral TPU.
If you need to access GPIO, you have some GPIO pins at your disposal. You can’t get the full 40 pins. This means you can’t connect HATs, but you can access a small number of GPIO pins. If you need to connect sensors, the I2C pins are exposed and can be used by components that support the communication protocol, including Stemma QT components.
Compute Blade’s strength lies in numbers, or more specifically, “clusters.” Due to the unit’s small size and blade design, it can be easily embedded into a blade server, and as long as you have plenty of Raspberry Pis, you can use a powerful Arm compute cluster.
Compute Blade Basic starts at $65, but the version on the bench is the $107 Dev version with all the features. If you like what you see, kickstarter page to make your vows. Live chat with Kuleshov every week on February 14th at 2:00pm / 7:00pm GMT. Raspberry Pi Show, The Pi Cast.
Please note that crowdfunding a project does not guarantee that you will receive a finished product. You believe in the project and want it to succeed. You are not purchasing a retail product.