Infrastructure / Edge AI hardware

Rockchip RK3588

An application-processor SoC/chipset (not a board) made by Fuzhou Rockchip. It is an octa-core ARM SoC pairing 4x Cortex-A76 with 4x Cortex-A55, a Mali-G610 MC4 GPU, and a 6 TOPS triple-core NPU supporting INT4/INT8/INT16/FP16/BF16/TF32, with LPDDR4x/LPDDR5 up to 32GB and PCIe 3.0. It is the chipset behind a broad range of edge/SBC products (Orange Pi 5, Radxa Rock 5B, Khadas Edge2), making it the de facto high-end ARM SBC SoC of its generation.

Verified live 2026-06-22 via primary sources. Documentation is unusually open (full TRM and datasheet public) and the RKNN toolkit is openly distributed, but the SDK ships as prebuilt binaries under a proprietary license with closed firmware.

Openness

3 high confidence

3.0

toolchain: partial (NPU SDK on GitHub but proprietary RKNN license + prebuilt blobs, rknpu kernel driver open)
datasheets: public (full TRM + datasheet public, no NDA)
blobs: required (proprietary GPU/NPU firmware)

Documentation is unusually open (full TRM and datasheet public) and the RKNN toolkit is openly distributed, but the SDK ships as prebuilt binaries under a proprietary license with closed firmware.

https://www.rock-chips.com/a/en/products/RK35_Series/2022/0926/1660.html recorded 2026-06-22
official RK3588 page: 8-core A76+A55, 6 TOPS triple-core NPU, Mali-G610 MC4
https://github.com/airockchip/rknn-toolkit2/blob/master/LICENSE recorded 2026-06-22
SDK under a proprietary RKNN SDK License

Adoption

5 high confidence

5.0

Powers a very large multi-vendor set of SBCs and modules (Orange Pi, Radxa, Khadas, FriendlyELEC, Firefly), widely available at retail and broadly supported in community Linux.

https://forum.armbian.com/topic/26398-video-comparing-rk3588-sbcs-nanopi-r6s-khadas-edge2-radxa-rock5b-mekotronics-r58-mini-r58x-4g-orange-pi-5/ recorded 2026-06-22
independent comparison of many RK3588 SBCs from multiple vendors

Capability

4 high confidence

4.0

The 6 TOPS NPU is genuine edge-AI capability and the RKLLM SDK runs quantized LLMs on-chip, but modest versus dedicated accelerators.

https://github.com/airockchip/rknn-llm recorded 2026-06-22
official RKLLM SDK running LLMs on RK3588-series NPU

Unchanged since 2026-08-01 (last edited, not re-checked)