[Grants] Grant Application from David M Williams for RISC-V hardware

[David M Williams]

First name: David
Last name: Williams
Your email address: david at davidmwilliams.com<mailto:david at davidmwilliams.com>
Project name: Advancing open-source database support for RISC-V hardware
Grant type: Project grant
Project team details: David M Williams


A statement including a willingness to provide regular project updates on the project: Yes, I will provide regular updates via various online, freely accessible media, and as many other updates as people are interested in.


Aim of the project:
*** Please excuse any roughness or imprecision in my aim and costings; I am submitting this conscious the deadline for grants approaches, and I hope the concepts expressed here can be refined with the help of the community ***

The primary aim is to improve the free software ecosystem for RISC-V around support for open-source database systems.

For clarity, this project received inspiration from Russell Coker’s existing work to improve the free software ecosystem around RISC-V. I have discussed my interests with Russell and as our interests were in different areas it didn’t fit well in a single request. However, as there is overlap, we will be exchanging ideas. I trust my proposal does not in any way negatively impact Russell’s as that is not my intention, and I appreciate all he is doing for the community, and for the information and support he has provided me.

As this group knows, any non-trivial app relies on data persistence, and open-source databases such as MySQL, etc., power much of the Internet. Data/database systems have always been relevant, but increasingly so in this era of big data, of AI and ML, etc. The RISC-V ecosystem offers a tremendous opportunity for organisations and individuals to have a completely open platform, but yet as this is still an emerging space, software support is lagging.

I have a particular passion for data; in all my 30 years of professional IT experience, it’s always been giving people actionable insights to properly manage their own business area where I have seen the greatest transformation. As much as I might think I’ve written clever software, or built robust networks, or even implemented a great ERP (😊) it’s been data, insights, and information that have had the greatest effect. I say all this simply to say data is an area I love.

Previous explorations into RISC-V database engine support found that MySQL will not natively compile for RISC-V (but can with some patching), while MongoDB will not compile at all, nor will TiDB. Neither are available through APT installation either. MariaDB, Redis, and PostgreSQL fared better. However, there are other databases, and there are other tools that are vital and critical – Apache Kafka, AirByte, and more, and which the whole field of data engineering relies on.

In my view, it’s essential these products are all as easily available on RISC-V as they are on Intel or other current, popular hardware.

Thus, my goal here is to identify key free software database and data-related products, establish their current capacity to compile, execute, and be distributed on and for RISC-V platforms, and then to explicitly work with the relevant project teams to highlight problems and seek resolution. At best, this project can help achieve quality software outcomes for this section of free software; at worst, it will result in a documented, current state of play of such products identifying the tools that are and are not working on RISC-V, and what issues exist for those that do not.


Estimated cost breakdown of the project:
This section is fluid, and my apologies for the imprecision. Identifying appropriate hardware that is sufficiently powerful but cost-effective and, importantly, which exists to ship has not been simple or straightforward. Further, shipping fees are somewhat unclear on certain items.

That said, the LicheeCluster 4A with chassis (because it also ships without!) looks to be a suitable device -> https://sipeed.com/licheepi4a however, the cost may be prohibitive for this esteemed group. According to AliExpress it begins at $AU 1,520.87 and ranges up to $AU 2,029.92, plus $AU 87.48 shipping, or a total of $1,608.35 to $2,117.40. Ref -> https://www.aliexpress.com/item/1005006024510948.html The pricing difference represents the RAM and storage, ranging from 7*(8GB + 32GB) to 7*(16GB + 128GB).

Alternatively, there is also the Lichee Console 4A which says it is a portable RISC-V development terminal. It sounds cool, but is 7” and tiny though it does offer an M.2 (NGFF) 2422 socket. Pricing ranges based on capacity, but for the higher end model of 16GB DDR + 128GB eMMC + 1TB SSD it is $AU 698.60 and $41.72 shipping, for a total of $740.32 -> https://www.aliexpress.us/item/1005006174268912.html?gatewayAdapt=4itemAdapt

Another device is the Lichee Pad Max 4A which is a 14” device, but it is pre-order only and I was not able to find the release date or pricing.

Finally, there is the DeepComputing DC ROMA RISC-V laptop -> https://deepcomputing.io/product/dc-roma-risc-v-laptop/ which appears to be a robust model and costs are currently $AU 1,159 for the laptop plus $AU 198 shipping, or a total of $1,357. While I like the portability of a laptop, and the design and screen size, it has lower specs for a higher price than other options. I feel this unit has the advantage of aesthetics and design but ultimately does not represent good value.

Consequently, I would be hoping this esteemed group may be willing to consider the costs for a LicheeCluster 4A with chassis with specs considered acceptable to the group for its costs, or the higher end Lichee Console 4A.

This project does not seek funding for software (of course), or for peripherals such as keyboard, mouse, monitor, etc. – purely for the RISC-V unit itself.

I hope this request is (a) interesting, (b) demonstrates a valuable contribution to the community’s goals and aims, and (c) is sufficiently worded and fleshed out for consideration, subject to refinement through discussion.

Sincere regards,
David W

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.org.au/pipermail/grants/attachments/20231130/d40eaf1b/attachment-0001.html>