How big/important is embeded software in F1 nowadays ?

[zack!]

Hi all,

In 2014, role of PU control by OS sofware has been in the spotlight. And it appears SW is also used in other systems, like brake by wire for example.

I wondering how big is SW in design/dvpt of a F1 today ? So this is a "estimate" / order of magnitude game .

How many computers are in F1 (FPGA, ECU, other progammable specific chipset) ?

How many systems are contoled by SW (I know 1 ECU can host several SW) : MGU-H, MGU-K, Brake by wire, telemetry, cockpit monitoring/display, Energy store, Fuel monitoring, Wheel monitoring, startup procedure, refueling procedure, etc ...

How big are embedded software : number of lines, or number of model in case of model based engineering.

How big is SW devpt compare to the rest of F1 dvpt : size of SW team compare to total size of Design Office for example (in %) ?

Thank a lot for your insights

[mep]

How important is software in your life?
How did that chance compared to 20 years ago?

Its the same as in F1.

[zack!]

How important is software in your life?
How did that chance compared to 20 years ago?

Its the same as in F1.

Well, you canno't compare my budget with F1 team's one, my computers (number, power, content) are not regulated by FIA.
Of course I understand your point, just want to see "how much" this counts in F1 dvpt (just the embedded part) :
- number of SW to develop,
- number of critical system controled by SW,
- computation power of a F1 (or an ECU for example compare to a smartphone for example),
- size of code,
- budget of SW dvpt,
- number of SW engineers

[n_anirudh]

I was told by RedBull's chief engineer that they do not rely on others to develop software (tools) for them and they do it inhouse, including all their CFD codes etc. Of couse, they are in a race of their own to get accurate readings. I guess they can code to implement anything new that needs to be done.

S/W to run the cars would also evolve and would remain as important as it was when the first bit of electronic was introduced- VERY!

[Jersey Tom]

Don't forget about all the other software like everything built around the data acquisition, keeping track of setup data, probably weather, etc etc.

Immense amounts of software development.

I actually rather like mep's answer.

[Jef Patat]

In your questions I read you are interested in F1 car SW/HW. Also consider the amount "out of car SW". My answers are based on "in car SW/HW"

In 2014, role of PU control by OS sofware has been in the spotlight.

Did I miss something? Since when do we have insight in OS choice?

I wondering how big is SW in design/dvpt of a F1 today ?

Define big, define design

How many computers are in F1 (FPGA, ECU, other progammable specific chipset) ?

146.3 computers are in F1
FPGA is normally not "programmed",its hardware is configured. You shouldn't think of it as "normal" code. It is a completely different way of thinking. I have done it in the past when it was quite new. Xilinx Virtex II.
Don't think you will get a decent answer. Take for example the fuel flow sensor. It's an intelligent device, has filtering, has configuration, has CAN connection, it has non volatile memory. Nobody of us has an exact idea of how many sensors there are. In fact, nobody working in the team will be able to answer you that. Of course, they can look in the breakdown, mark each "computer" and then give you the answer.

How many systems are contoled by SW (I know 1 ECU can host several SW) : MGU-H, MGU-K, Brake by wire, telemetry, cockpit monitoring/display, Energy store, Fuel monitoring, Wheel monitoring, startup procedure, refueling procedure, etc ...

that would be 6835

How big are embedded software : number of lines, or number of model in case of model based engineering.

Nof lines is a meaningless factor. Do you consider autogenerated code form the models as well? Expect it to be millions, you can choose how many millions.

How big is SW devpt compare to the rest of F1 dvpt : size of SW team compare to total size of Design Office for example (in %) ?

Lots of SW will be outsourced. Lots of SW will be written by suppliers (again, think of the fuel flow sensor, think of MCL ECU provided to all teams). Do you consider model based design as SW design? (Note: you shouldn't) To give you an idea: the company I worked for has an engineering department of around 150 people for one type of machines. Around 30 of them in a group for "elektrical and electronic". Around 10 of them doing model based, 5 actual embedded code, 5 in testing and managing. Just wait for the teams HR to read this topic, they will be able to give you an exact anwer.

Believe me, I have a pretty good insight in automotive embedded SW development. I have been working in it for over 10 years. I have written ECU SW(as in electronic controller unit, and not in engine controler unit). I have written desktop tools to download engine controller firmware and maps. SW development, as in writing code for controllers has been reduced to an absolute minimum. Almost all 'software' is code that is autogenerated out of models.

As for HW. That depends on the requirements of the device. Connected I/O, memory, complexity of code, ... I can safely say that processors of complex controllers (engine for example) are along the lines of tablets. ARM Cortex top range.

[zack!]

Thank you for sharing your knowledge and experience, very appreciated.

OK, actually the "out of the car" seems to be the really "big" thing : CFD, simulator, strategy, telemetry and data analysis tool, etc. This is where we have the most classical "in house" SW dvpt (using programming language). Am I right ? Can we put a size factor here ? 10/1 ?

Concerning design, I do not considering SW to design the car (3D modeling, structural computation, etc), as in CASE tools used in Design Office (this part is not "inside" the car).

I consider SW as implementing functions of the car (part of systems). It could be generated code from models, or written directly in programming language.

If I understand correctly, many intelligent sensor, DSP, or FPGA, are in supplied system, so not really managed by F1 teams (except to tune/configure them by bus/connection). So let this low-level stuff out of scope for the moment.

So, we consider only really complex system controled with a true programmable code (auto-genrated by model or directly hand written). That means, they are executed by real versatlite and powerfull "computer". How many systems in that category ?

Then, would like to know how this "computer" are regulated by FIA.

Can F1 team use the "computer" they want, or is this kind of "computer" is regulated by FIA ? I know for example that FIA impose one "ECU" from MCL as you say. Is that ECU the only "computer" that they can use ? Can they use as much as they want of that "computer" ? Is there different "regulated" computers, you seem to imply that PU controler is different from ECU, that is right ? Is that "regulation" only for certain sensitive system for sporting (telemetry, and remote control of the car from pit lane, PU control, traction control, brake control), or all SW system (ie no SW oustide ECU or PU CU, all SW visible from FIA).

Thanks

[monsi]

Take a look at Hail22's post of a video from Lotus in the Lotus team thread.

[zack!]

Take a look at Hail22's post of a video from Lotus in the Lotus team thread.

[youtube]http://www.youtube.com/watch?v=g3b4xC0FIRU[/youtube]

This one ? It explains hte importance for the first category, the SW that is outside of the car, not for the second category, the SW inside the car.
But anyway, interesting context. Thanks.

[Jersey Tom]

I consider SW as implementing functions of the car (part of systems). It could be generated code from models, or written directly in programming language.

That's an awfully narrow view of things, don't you think? Or perhaps it would have been better to phrase this post as, "How important is firmware and embedded system code development in F1 nowadays?"

I don't have a great answer in any event. Certainly the development of everything that goes into the steering wheel is pretty significant! Not sure how involved the teams are in the guts of electronic differentials but that would also be a significant one I'd think.