Is the SiFive FE310 a Toy?

[asmi]

I personally find it reasonably well documented. For the GPIOs, sure it's not very detailed and kind of assumes that you already know all the concepts usual with GPIOs, but it looks enough IMO. Seems like for each register, one bit is to control the corresponding IO, you don't need each register detailed for this?

Yes you do need as nobody is born with this knowledge.

[brucehoult]

I guess being an ISA / compiler guy, not a hardware guy, I *expect* to have to make and test assumptions about hardware documentation that experienced hardware guys would simply know the answers to automatically. Reference to the header file and/or the Arduino library code answered any questions I had, but of course you are correct that it would be better to have everything up to and including code examples pulled together in one place.

[Sal Ammoniac]

I personally find it reasonably well documented. For the GPIOs, sure it's not very detailed and kind of assumes that you already know all the concepts usual with GPIOs, but it looks enough IMO. Seems like for each register, one bit is to control the corresponding IO, you don't need each register detailed for this?

Well-written documentation should never assume anything. Spell it out, even if 95% of readers will already know what's going on, because there's always the 5% who won't.

[Sal Ammoniac]

I guess being an ISA / compiler guy, not a hardware guy, I *expect* to have to make and test assumptions about hardware documentation that experienced hardware guys would simply know the answers to automatically. Reference to the header file and/or the Arduino library code answered any questions I had, but of course you are correct that it would be better to have everything up to and including code examples pulled together in one place.

You have different expectations than I do. When I use a part, I expect the hardware manual to explicitly and completely describe the operation of the part. I should not have to refer to header files or library code to figure things out. Anyone who expects their users to do that is both lazy and inconsiderate.

[thm_w]

What I don't get is their prices, when a +600MHz Teensy 4 is less than $20 and dual core esp32s (240MHz) are $5... Sparkfun: HiFive1 (320MHz): $67, RedV RedBoard (150MHz): $39.95, RedV Thing Plus (150MHz): $29.95

You should compare to sparkfuns own prices to understand, their cheapest ESP32 is $21: https://www.sparkfun.com/products/14689
The equivalent to that board is the red-v thing plus you mentioned, $29: https://www.sparkfun.com/products/15799

Its not unreasonable.

[srce]

How does SiFive expect to play in the professional MCU market

It doesn't.

SiFive is in the business of licensing cores to to other companies, who might in turn make and sell stand-alone MCUs, or embed them in some other product.

Exactly like ARM.

Early versions of the FE-310 are a demo of the E31 core, not a mass market general purpose MCU. The FE310-G003 (with 64 KB SRAM and other features) has been designed for a specific high volume product, the "SiFive Learn Inventor" and might also see general availability. But it's a sideline.

Hundreds of millions of SiFive cores will ship in 2020 inside Samsung's flagship phones (controlling the camera and 5G), in Qualcomm 5G SoCs, in Huami wearables, in Allianz microwave ovens (>50% world market share, OEMing for many brands such as GE) and other whiteware, and many others that may or may not ever be publicly announced.

The question is, with low licensing fees and no royalities, can SiFive ever even make back the $125m they've had invested in them?

[brucehoult]

The question is, with low licensing fees and no royalities, can SiFive ever even make back the $100m they've had invested in them?

Yes :-)

[srce]

The question is, with low licensing fees and no royalities, can SiFive ever even make back the $125m they've had invested in them?

Yes :-)

Seems unlikely from IP sales, for a few hundred k per license. There just aren't enough designs out there. You'll be needing another funding round next year, unless you start making money from design services or silicon!

Oh, edit to $125m...

[donotdespisethesnake]

According to https://www.eetimes.com/qualcomm-takes-stake-in-sifive/ sounds like SiFive are doing pretty well.

I think in future we might look back and see RISC-V as the Open Source Hardware equivalent of Linux.

[brucehoult]

According to https://www.eetimes.com/qualcomm-takes-stake-in-sifive/ sounds like SiFive are doing pretty well.

Ah, good. I always need to be careful not to reveal information that hasn't already been made public by an executive. Sooo ... I'll just point out that in this business you make bookings in one year and they are realized as revenue in the next year. So when Naveed says what revenues are expected in 2019 (say) ... that is based on contracts signed in 2018, and it's basically money in the bank if you don't totally screw up the execution.

I think in future we might look back and see RISC-V as the Open Source Hardware equivalent of Linux.

Yes, and SiFive something like a RedHat.

[SiliconWizard]

Good thing for SiFive is that RISC-V is currently a fast growing "market", and there's still very little competition overall, considering that companies willing to go for RISC-V are doing it for specific reasons beyond merely getting CPU cores.

[brucehoult]

Good thing for SiFive is that RISC-V is currently a fast growing "market", and there's still very little competition overall, considering that companies willing to go for RISC-V are doing it for specific reasons beyond merely getting CPU cores.

Andes is serious competition, especially in China. They had their own NDS32 ISA with gcc and linux kernel support and mature extensions such as DSP and a lot of customers using it for many years, and they're switching everyone to their RISC-V cores.

The cores from the Switzerland/Italy university PULP project are getting significant commercial use too, including in MCUs from NXP.

It would be nice for SiFive if they had the market to themselves, but it's got to be good for customers and overall RISC-V adoption that there are multiple serious vendors who can be substituted for each other while keeping software compatibility.

[SiliconWizard]

Good thing for SiFive is that RISC-V is currently a fast growing "market", and there's still very little competition overall, considering that companies willing to go for RISC-V are doing it for specific reasons beyond merely getting CPU cores.

It would be nice for SiFive if they had the market to themselves, but it's got to be good for customers and overall RISC-V adoption that there are multiple serious vendors who can be substituted for each other while keeping software compatibility.

A couple companies, that's still "little competition" in my eyes given the growing and diverse demand. Not sure it would be a good thing if SiFive actually had the whole market to themselves. It would be very hard to handle, and could likely lead to SiFive's demise.

[srce]

According to https://www.eetimes.com/qualcomm-takes-stake-in-sifive/ sounds like SiFive are doing pretty well.

Some of the numbers in that are complete bullshit though 

Or just Open Silicon business dressed up.

[brucehoult]

According to https://www.eetimes.com/qualcomm-takes-stake-in-sifive/ sounds like SiFive are doing pretty well.

Some of the numbers in that are complete bullshit though 

Or just Open Silicon business dressed up.

As Naveed said, that $100m  revenue is currently 75% from silicon and 25% from IP licensing. Go and look up Open Silicon revenues before SiFive acquired them -- they were a small fraction of that. Acquiring Open Silicon is of course part of what SiFive's funding has gone to, and so silicon business revenue is entirely relevant to the original question of whether SiFive will recoup its funding.

[asmi]

As Naveed said, that $100m  revenue is currently 75% from silicon and 25% from IP licensing. Go and look up Open Silicon revenues before SiFive acquired them -- they were a small fraction of that. Acquiring Open Silicon is of course part of what SiFive's funding has gone to, and so silicon business revenue is entirely relevant to the original question of whether SiFive will recoup its funding.

As if you'd say it here straight if it were otherwise Sorry but company's employee is the last person I'd ever ask if I would want to figure out what that company's financial state is.

[donotdespisethesnake]

As if you'd say it here straight if it were otherwise Sorry but company's employee is the last person I'd ever ask if I would want to figure out what that company's financial state is.

The last guy I would ask is some random anonymous person on the internet. 

[SiliconWizard]

Anyway, I don't think raw revenue is what matters for SiFive at the stage they are. A minimum number of contracts/clients is required to show that the business model is viable. Even if the net result was still not quite positive (although it appears to be, but it's not 100% clear and doesn't really matter).

I think they are still kind of a startup (albeit a big one now). Not that I necessarily wish them that, but that they eventually merge with a bigger fish is not unlikely at all. Then business and strategy will change.

[srce]

Go and look up Open Silicon revenues before SiFive acquired them -- they were a small fraction of that.

Whereabouts? I can only see conjecture. It seems they sold a 75% stake for $190m in 2007 - so presumably were doing well, but must have had a somewhat more realistic valuation when SiFive acquired them.

Acquiring Open Silicon is of course part of what SiFive's funding has gone to, and so silicon business revenue is entirely relevant to the original question of whether SiFive will recoup its funding.

If they (you) do it by selling silicon, then sure, but not from just processor IP. It would be interesting to know just how small a chunk of the IP sales is from RISC V IP. The numbers don't add up when you look at other vendor's figures.

[brucehoult]

Acquiring Open Silicon is of course part of what SiFive's funding has gone to, and so silicon business revenue is entirely relevant to the original question of whether SiFive will recoup its funding.

If they (you) do it by selling silicon, then sure, but not from just processor IP. It would be interesting to know just how small a chunk of the IP sales is from RISC V IP. The numbers don't add up when you look at other vendor's figures.

No one has ever said SiFive exists to make money just by licensing RISC-V IP. *Everything* the company publishes, from https://www.sifive.com/why to keynotes at conferences e.g. has been clear for years now that the goal of SiFive is to make it cheaper, faster, and easier for people to make customized chips and processors. A large part of that is by incorporating standard and proven but customisable and parameterisable components into an automated design process.

The free and open RISC-V ISA is a key enabler for that, not the end goal.

RISC-V was created because Krste, Andrew, and Yunsup needed a control processor for the experimental high performance Vector engines they were working on and all the existing alternatives either didn't have 64 bit (e.g. ARM and OpenRISC), wanted too much money (even MIPS), didn't allow customization of the ISA (i.e. adding the vector instructions), didn't have enough free opcode space anyway (OpenRISC, even if they extended it to 64 bit themselves), or licensing would not allow distribution of the results, even to other academics.

Even in the first days in 2010, RISC-V was an aid to achieving other goals, not the goal itself.

If the needs of your SoC are best met by putting a standard ARM core on it, not a RISC-V one, then SiFive will do that for you. Open Silicon is and remains an ARM Center of Excellence http://embedded-computing.com/articles/reducing-risk-time-market-the-arm-center-excellence/

[brucehoult]

Go and look up Open Silicon revenues before SiFive acquired them -- they were a small fraction of that.

Whereabouts? I can only see conjecture.

Me too, on the web. As a privately held company there is no requirement to publish financials publicly, but potential investors see them (of course) and financial analysts hear about that on the grapevine and apply standard multiples to convert between revenues or profits and valuations and will generally not be very far off :-)

(my first career straight out of university in the mid 80s was being the computer guy in the financial research department of a major NZ brokerage that dealt in stocks, bonds, futures, FX, and some floats (IPOs) and mergers and acquisitions. That company did the finance for the Auckland Harbour Bridge (long before) and while I was there I was involved in preparing he offer documents for NZs then largest ever M&A deal of Carter Holt Holdings with Alex Harvey Industries.)