Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Why HPSC Is a Big Deal for Space Exploration (windriver.com)
23 points by mooreds 7 hours ago | hide | past | favorite | 8 comments
 help



The PDF linked in the article provides much more information on the technical details: https://www.nasa.gov/wp-content/uploads/2024/07/hpsc-white-p...

> The PIC64-HPSC series has a built-in 240 Gbps, 16-port TSN Ethernet switch with Remote Direct Memory Access (RDMA) that uses the RDMA over Converged Ethernet (RoCE) version 2 standard

It seems like this is what the article is referencing. However, the ethernet embedded CPUs have "GigE" or even 10/100 capability. Why the need for 240 Gbps with 16 GigE ports? Are the ports 10G fiber capable and they are just future-proofing or is there something I missed with the time-slicing and RDMA?


Looks like the press release is mixing up the chip level with some board or assembly. The MPUs in the family look like 10G by themselves, so maybe the board includes the switch and shared memory.


There's never been much of a "market" for systems like this. Even at SpaceX scales, nobody would ever be able to sell enough chips to make the effort worthwhile. The only reason to do it is to fulfill a need in a niche market.

Is this the one with the SiFive x280 in it?

I've always felt bad for windriver; they do a lot of good work and there are clear benefits but because linux is free people keep moving away from their products. It's cool to see them doing more RISC-V stuff though

There's also Green Hills, which has a far better product (Integrity) vs. WRS VxWorks. Somehow VxWorks has better marketing and more sales, because their product is cheaper.

Both are commonly used for space-borne platforms, and I've done development for said platforms with both.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: