Sifive Layoffs?

Lets hope this does not impact development on the VisionFive2.

See this and other articles.


Capitalism, amirite. Anything to save a buck, even at the expense of the products being produced and the people producing them. How… unsurprising. They are just taking from Google’s and Facebook’s playbook here.

1 Like

I’ve done some reading around this and it looks like somewhere between 100 and 300 of their core engineering, marketing and sales people. There could be a major fire sale to maximise profit in the shortest timeline possible.

The unfortunate reality is that the higher the valuation of a company in relation to actual number of direct employees, if the plan is to float the company, the higher the initial stock price. So this could be the first step in changing SiFive Inc from a privately held company to a publicly listed one on the stock market.

One, intended or unintended, side effect of this could be to suppress their direct competition in the RISC-V space and produce a very large number of RISC-V SOC’s, for at least a few of years, using SiFive IP.

1 Like

RISC-V is open source hardware.
This means everyone can desgin his self cpu without bought a IP from SiFive.

Specialy, RISC-V is simple enought to become a university book.
you got a pre-desgin IP to “bootstrap” is enought.
Or, you can start from “empty” if you are Chinese Academy of Sciences …
Or, you can empoly some graduates who graduated from CAS …
I remember that some chinese riscv develper is jump from loongson. :thinking:

SiFive’s business is limited by risc-v’s spec desgin.
RISC-V specification desgin is ssssssssslllllllllllllllloooooooowwwwwwwww…

To make an effective application processor from the RISC-V specifications, you need to have a comprehensive understanding of processor design. To make a Microcontroller is dead easy for a university student. Making an efficient superscalar processor with out of order execution isn’t.

1 Like

But I sad:

you can empoly some graduates who graduated from CAS …

“out of order execution” is not a necessary feature, RVA22 is published half-year ago.
RISC-V cpu is running in alot of IoT devices.They do not depend OoO.
RVA22’s target is ARM A78, not for IoT devices like ARM M series.

And , I sad:

I remember that some chinese riscv develper is jump from loongson. :thinking:

Not ONLY student can desing a new CPU. There have alot of old-birds are working in new cpu .
SiFive is not have “some mysterious magic” to design riscv.

1 Like

I don’t imagine this should impact VF2 development thankfully, but StarFive may look to a different core designer for their next SoC in the future to be used in the (hopefully) eventual VF3. I hope all those former employees can find better work elsewhere :frowning:


To be honest, after their over-promised, under delivered board; slow updates and insular community, lets hope they sink. I Read the thread about people whining about their stock options. Stock is worth what it’s perceived as, which partially comes down to how ready for market your products are. The VF2 is not ready for a consumer audience, and the communication even for “expert developers” to use it, was poor at best.

Sorry, but I can not agree on that. Hardware and basic software support is ok in my eyes. I did not expect a feature complete software stack and wanted a board for a reasonable price which allows me to experiment with RISC-V and that has been delivered. No offense here, we can agree to disagree on that one :wink:
There is only one thing which annoys me: that there is no repository where every patch and every source package they are allowed to deliver (for my knowledge everything besides the kernel driver for the Rogue chip) really is available, matching the official debian build. E.g. in the github repo there is the build 3.7.5 available but when it comes to mesa, it uses a version 22.1.3 and all patches are for that one. But the debian version uses a far more current 23.x.y version of mesa.
If that would have been properly documented, I would not have to guess and poke around building a fully working Archlinux installation, based on @cwt 's work. That one is working but has no good support for mesa from the reasons, I stated above. And that is unneccessary work which has already been done by the guys at Starfive, I would love to concentrate on other things where really work has to be done.