Running an Ethereum Node and Validator on RISC-V hardware

About 2 years ago RISC-V hardware got powerful enough to do initial tests for running an Ethereum node on such hardware. As expected, the speed wasn't quite there yet, but we started getting some clients ready, submitted PRs, got a Devcon talk and even got some core devs interested in it. There were steady improvements in the last 2 years. We were able to run nodes for larger test networks until we managed to sync mainnet about a year ago. But only barely so. Technically it stayed in sync, but practically it was always 1-2 slots behind. This changed this summer with the newest hardware iteration. I managed to run a fully synced Ethereum node on a RISC-V single board computer (Spacemit K3 CoM260). The validator running through that node attested flawlessly and correctly attested head votes, even right after epoch boundaries. The node still is a bit slower than my usual NUCs, but that is not surprising as the board has about the power of a Raspberry Pi 5. In my impression the bottleneck still is the consensus workload. Reducing the number of individual validators helped here quite a bit. The execution client has some spare power to be able to handle gas limit increases and thanks to ePBS it should get more time per slot to do its duties anyway. So I hope my node can handle the workload for 1 or 2 more years.

Currently 2 Consensus clients run out of the box (Nimbus and Lighthouse). On the execution side, geth has always just worked. Now, Ethrex is also running, even though the initial sync is a bit more involved because the board has a 'only' 32 GB of RAM. But when Ethrex works it works perfectly and is very resource efficient. It is great to see that a second execution client now runs on RISC-V hardware. Grandine builds, but fails to run. I did not have the time yet to investigate as to why. Eth-docker also works, but does not support all client pairs just yet.

As RISC-V is an open standard, I see these CPUs to be able to capture a junk of the consumer market in the long run. It already happens with a lot of lower cost applications, where RISC-V CPUs replace more expensive ARM and other chips. There are also well funded companies building servers and consumer PCs using RISC-V CPUs There are also a some who specialize on building accelerator cards using the RISC-V standard. We will have to see if RISC-V CPUs manage to capture large parts of the market. If it it happens it will take years (~ a decade). With the direction Ethereum is taking with zk proving the network, I definitely see a possibility that smaller nodes will run on low cost hardware. RISC-V CPUs have a clear advantage here. Currently with the resource usage of an Ethereum node, combined with the RAM and SSD prices, the price advantage a RISC-V CPU can have does not really matter. I expect it will in the long run though.

For people wanting to know more about the progress, here are some reddit posts in chronological order:

First post in 2024: https://www.reddit.com/r/ethfinance/comments/1ewn8dw/daily_general_discussion_august_20_2024/lj2anr4/

Announcement of the Devcon talk we gave in Bangkok in 2024: https://www.reddit.com/r/ethfinance/comments/1gn3k7p/daily_general_discussion_november_9_2024/lw85ry2/

First person to run execution and consensus client simultaneously on RISC-V hardware (two boards) in 2025 for Ethereum mainnet. They later improved Lighthouse to support RISC-V out of the box: https://www.reddit.com/r/RISCV/comments/1j5uqrs/ethereum_node_on_riscv_yes_its_possible/

Summary of the progress in 2025/2026 with first node running on one board: https://old.reddit.com/r/ethereum/comments/1t9tdqb/daily_general_discussion_may_11_2026/ol54wr4/