Specifications
System

System

SoC T-Head 1520
  • CPU (C910) RV64GC 4 cores 1,85 GHz
  • GPU (OpenCL 1.1/1.2/2.0, OpenGL ES 3.0/3.1/3.2, Vulkan 1.1/1.2, Android NN HAL)
  • VPU (H.265/H.264/VP9 video encoding/decoding)
  • NPU (4TOPS@INT8 1GHz, Tensorflow, ONNX, Caffe)

memory

Memory

16GB LPDDR4

volume

Storage

128 GB eMMC

network

Network

100 Mbit/s ethernet network card
Public IPv4 and IPv6 addresses included

os

Operating System

Debian, Ubuntu, Alpine

Energie

Energy

An average power consumption of 1.3W per core

Models and prices
NameCPURAMStorageEthernet NetworkPrice excl. VAT/hourPrice excl. VAT/month
EM-RV1-C4M16S128-A4x C910 RISC-V 64GC 1.85GHZ16 GB128 GB100 Mb/s0,042 €15,99 €
B stands for byte. b stands for bit.
Benchmarks

The recent RISC-V architecture, which is currently being optimized, offers encouraging performance thanks to its modern, open design, making it a credible alternative to established architectures such as x86 and ARM.

Use cases

RISC-V EM-RV1 servers, available now in the Scaleway cloud, offer a unique opportunity to explore this new open architecture at an attractive price. They are available by the hour or by the month, with no installation costs.

Developers can deploy, test and debug their RISC-V applications in a cloud environment, reducing the obstacles associated with installing and configuring specific hardware. This enables rapid iterations, fostering innovation and wider adoption of RISC-V architecture.

By using EM-RV1, CI/CDs can adapt to demand and benefit from native performance, boosting the speed and reliability of the development process on RISC-V architecture.

The EM-RV1s feature a neural network accelerator chip (NPU), optimizing specialized computing capabilities for AI/ML workloads such as image recognition, machine translation and many others.

dropdown illustration
In-house design

The EM-RV1 servers are the fruit of months of intense research and development aimed at creating RISC-V servers in the cloud. The limited hardware offering, unsuitable for data centers, and the maturity of the software ecosystem, presented us with a number of challenges.

Every aspect of their design, from the initial iterations with suppliers, through the soldering of electronic components, the development of specific firmware, to the manufacture of the enclosures using 3D printing, bears witness to an iterative, artisanal approach.

Our goal was to offer you as soon as possible a range of RISC-V servers meticulously designed to meet the demands of a cloud environment.

A world premiere

Scaleway's EM-RV1 offering marks a turning point in the server industry as the world's first RISC-V dedicated server offering. This breakthrough testifies to Scaleway's unwavering commitment to being a leader in technological innovation. By introducing this revolution in the world of processors to the cloud market, Scaleway is positioning itself as an activator of the RISC-V ecosystem, accelerating its development and popularization.

EM-RV1 servers are more than just an innovation; they represent a bold vision for the future of computing. A vision that advocates the opening up of processor design standards, offering the possibility of the emergence of local players enabling nations to eventually control the manufacture of their processors.

From laboratory to production

The conclusion of an iterative R&D process in our Paris laboratories, the EM-RV1 is a bold, pragmatic approach to popularizing RISC-V processor architecture by making economical servers immediately available in the cloud.

Our engineers have worked tirelessly to obtain the best performance and guarantee maximum longevity. Particular attention has been paid to the maintainability of the servers, so as to be able to maintain each server with minimum impact on its neighbors, while maintaining extremely high density.

As these offers are particularly innovative, pushing back the limits of our know-how in many fields, they induce a degree of unpredictability. EM-RV1s are considered as "Labs" services, with a service level agreement (SLA) contractually defined in our special conditions for BETA services.

FAQ

The kernel installed by default is not vulnerable to GhostWrite, we worked with CISPA security researchers to provide a fix before the disclosure. Installations made before the 6th of June 2024 are vulnerable, instruction to update your kernel are available here: EM-RV1 Guidelines: Update the kernel.

RISC-V is an open source instruction set architecture (ISA) developed at the University of California at Berkeley.

Unlike proprietary architectures, RISC-V is freely accessible, enabling anyone to design processors based on this architecture without having to pay royalties.

With its modular, extensible design, RISC-V offers exceptional flexibility, enabling the creation of tailor-made processors suited to a wide range of applications, making it an attractive choice for researchers and companies looking to innovate in the hardware field.

You can order your Elastic Metal RV1 server from:
  • the web interface, directly on the order page. If you need help, see the dedicated documentation page;
  • the command-line interface. If you need help, see the dedicated documentation page.
  • These servers are only available in the fr-par-2 zone.

The initial allocation of an Elastic Metal RV1 server to your account takes a few seconds, and the installation of an operating system takes a few minutes.

We support 3 Linux distributions:
  • Ubuntu
  • Debian
  • Alpine Linux

Sure. Access to the server's serial console is available for installing the most exotic operating systems.

EM-RV1s feature the T-HEAD C910 processor, which supports RV64IMAFDC_Zicsr_Zifencei_Zfh_XTheadc instruction set modules.

Note that the processor also supports XuanTie proprietary extensions (XIE, XMAE).

A description of RISC-V instructions and their meanings can be found on Wikipedia.

We are committed to providing you with the best possible experience, but please note that the artisanal design and the youth of RISC-V processor architectures do not allow us to guarantee a specific level of service, resulting in an SLA of 0%.

Contractual guarantees for "Labs" services are detailed in our special conditions for BETA services.

Yes.

Yes

Yes, a serial console accessible via SSH is activatable on your account.

The sources of modified binaries are available in /usr/lib/modules/$(uname -r)/source

An NPU, or Neural Processing Unit, is a specialized chip designed to perform information processing operations based on neural models or artificial neural networks. These chips are optimized for accelerating tasks related to machine learning, natural language processing, computer vision and other tasks associated with artificial intelligence.

Unlike traditional processors that execute sequential instructions, NPUs are specifically designed to simultaneously handle large amounts of data and perform massively parallel calculations, making them particularly effective for tasks related to deep learning and deep neural networks.

NPUs are widely used in applications such as speech recognition, image classification, machine translation, object detection, sequence prediction and many other fields where massive and complex data processing is required. They help speed up these tasks and improve the performance of artificial intelligence systems, both in terms of processing speed and energy efficiency.

If you have lost network access to your server due to a configuration error, you can try to:
  • restart it. A guide is available here;
  • restart it on a rescue image provided by Scaleway. A guide is available here.
  • reinstall it with an image supplied by Scaleway. A guide is available here. Warning: this will erase your data on the local disk.

RISC-V is not yet officially supported in Kubernetes, but to let you test this technology with kube containers, we've prepared an image with the necessary pre-built binaries.

Thanks to Scaleway Kosmos and our Elastic Metal RV1, you can follow this guide to launch RISC-V containers on Kubernetes effortlessly and in just a few minutes.

All you need is a Kosmos cluster (currently only version 1.29.1), configured with a pool of external nodes and an Elastic Metal RV1 server.

If you don't have a Kosmos cluster, you can follow this guide to create your first Kosmos cluster.

If you don't have an Elastic Metal RV1, you can follow this guide.
Note that Elastic Metal RV1s are only available in the `FR-PAR-2` region in the `Labs` section.

  • 1. Install the server with the `Kosmos` image.
  • 2. Go to the Nodes tab of your Kosmos cluster and click Add external node.
  • 3. Do not follow the instructions on the page to connect your server into the pool.
    Note only the `POOL_ID` and `POOL_REGION`.
  • 4. Create a Scaleway secret key with the `KubernetesExternalNodeRegister` IAM permission set. How to create a secret key
  • 5. Connect via SSH to your Elastic Metal RV1 server
  • 6. Export noted env variables and secret key
    scaleway labs

    $ export POOL_ID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx POOL_REGION=xx-xxx SCW_SECRET_KEY=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx

    • 8. Run node agent
      scaleway labs

      $ sudo -E ./node-agent_linux_riscv64 -loglevel 0 -no-controller && sleep 10

      • 9. Restart your server
      Your server is now available in your Kubernetes cluster.

      All Q&A related to Scaleway servers and specifically to Elastic Metal products are available in the FAQ section: Go to our full FAQ.

      Otherwise, you can contact us officially via our support center, or informally via the Scaleway Community Slack.