F35 Software and Numbers

Because the F35 is the most scrutinised defence programme ever, is surrounded by a clutch of vested interests and as many opinions as there arseholes in the world trying to make sense of its progress is almost impossible. 

Costs are going up, costs are going down, on any given day you can find evidence of both.

There has recently been much discussion about delays on the F35 due to software, if you listen to the more the shrill reporting you will learn that there are more lines of code than atoms in the universe!

What I find most interesting is how the software argument so often comes down to quantity.

Looking at the F35.Com website, the official LM information site;

From the program’s outset, the software team has focused on developing six key software releases known as Blocks:

Block 1A/1B – Block 1 comprises 76 percent of the more than 8 million source lines of code required for the F-35’s full warfighting capability. Block 1A was the ready for training configuration while Block 1B provided initial multi-level security.

Block 2A – Block 2A is currently released to the F-35 fleet. It provides enhanced training including functionality for off-board fusion, initial data links, electronic attack and mission debrief. Under Block 2A, nearly 86 percent of the required code for full warfighting capability is flying.

Block 2B – Block 2B provides initial warfighting capabilities, including but not limited to expanded data links, multi-ship fusion and initial live weapons. The U.S. Marines will declare IOC with Block 2B. Under Block 2B, more than 88 percent of the required code for full warfighting capability is flying.

Block 3i – Block 3i provides the same capabilities as Block 2B. The principal difference between 2B and 3i is the implementation of new hardware, specifically the updated Integrated Core Processor. The Air Force will declare IOC with Block 3i. With Block 3i, more than 89 percent of code required for full warfighting capability will fly.

Block 3F – Block 3F provides 100 percent of the software required for full warfighting capability, including but not limited to data link imagery, full weapons and embedded training. Requirements development for Block 3F was completed in June of 2013.

It also describes the current status of software development.

As of March 2014, more than 88 percent of the required F-35 software is currently flying. Approximately 97 percent of the required software has been coded. This equates to about 200,000 lines of code that remain to be written.

In 2013, the F-35 program flight tested the first four iterations of Block 2B, the software configuration the Marines will use to declare Initial Operating Capability. Pilots conducted sensor fusion operations, executed night landings on the USS Wasp and dropped both air-to-air and air-to-ground ordnance. The software is proving to be stable and performing well.

So 88% of the code is already flying and 97% has been coded.

That puts some perspective on the ranting but what it does not do is describe timescales for the remaining code and what functionality the same remaining code is related to.

If we say the remaining 12% is going to take 10 years and is needed for weapons release then Houston, as they say, we have a problem.

Those whose job it is know will no doubt know.

Advocates and critics will take their positions and nothing will change their minds, I tend to think the F35 is the product of the cream of the Western aerospace engineering profession and together, they are unlikely to deliver a pup.

Costs on the other hand, lets be honest, and this should come as no surprise, it is going to cost a great deal.

In a world of unlimited defence funds this would be taken with a shrug of the shoulders but we do not and whatever the industrial benefits of the F35 (and they are many) we should not let the industrial tail wag the military dog.

The defence budget has many competing demands and the UK armed forces have to avoid being hollowed out due to single capability areas.

With the UK is due to confirm its first production order for the F35B with final numbers being decided in SDSR 2015 the MoD should remember.

Balance is all.



Newest Most Voted
Inline Feedbacks
View all comments
Paul Robinson
Paul Robinson
April 5, 2014 8:54 pm

I just dunno what it is, but the more i see of close ups & details, it looks very frail, maybe more stealthy with small cross section. Just looks like it’s a remote control model that would break with the slightest clout. Ok it’s supposed to be filled with tech that’s ahead of it’s time, but weekly get at least one or 2 e mail from defence or military sites, in French, English, or translation to either from other languages, saying the tech is behind, & delayed delivery dates for the equipment to fly the damn thing, & make it an effective weapon. Sorry but severe doubts it will be really the next gen, by time it ever arrives in service. I see more conventional a/c getting modernised that can compete with it on equal terms or better. Pig in a poke if they don’t remove the digit. A pity the don’t give the honourable Order of the Irremovable Digit any more – Pilot officer Prune is alive & well, as the dropping of the camera in the MRTT Voyager confirms, among many other cock ups, sorry but especially in USAF & USN aviators’ cases lately. Time i changed my profile pic back to P.O.Prune or maybe his French equavalent, since i’m in Frogland now.

April 6, 2014 6:59 am

@ TD

“We should not allow the industrial tail to wag the military dog”

Not sure if I agree with this statement. Most countries with higher defence spending have significant military industrial bases. By shrinking the domestic industrial base in favour of foreign equipment the MOD risks eroding support for military spending even more.

The F35 is such an important industrial project for the UK that it would literally be worth the DTI buying our F35 quota even if the MOD will not use them.

That being said in my the program seems to be increasingly stabilising and costs seem to be falling. we will buy our minimum quota for the carriers then eventually some more in the 2030’s. The F35 is the right aircraft for the MOD and a fantastic deal for UK industry and the tax payer. Given that our work share is not that much smaller than our work-share of Typhoon 15% vs 30% we should stop looking a deals like the South Korean FX contract as failures as well.

April 6, 2014 9:01 am

From a March 22 2014 Bloomberg report
“The F-35 program is estimated to cost $391.2 billion” @ 15% that’s $58.7 billion for British industry , if we buy 150 F35B’s that equates to $390million each , less their cost of $150million we are up $240 million per aircraft (Martin has it bang on , the DTI should fund this if no one else as I would be sure the contracts specify we buy 150 or the next American contractor takes away an equivalent proportion of our share over the next 30 year production run if we buy less)
The said same article linked above discusses the software issues from a ‘leaked’ report from the U.S. Government Accountability Office (why are there no similar ‘leaked’ reports about the development of the Sukhoi PAK FA?;-)
It states “the Marines want the plane to be deemed ready for warfare in mid-2015, tests on some of its software might not be completed on time, and may be as much as 13 months late.” which is also the version we want ? Although the QE is not due to complete sea trials until 2016 surely part of that is aircraft handling , are we going to have sufficient aircraft?

April 6, 2014 9:15 am

Martin – agreed. There are nations that use their industrial sector as an asset to develop marketable products; in the defence arena the US, France, Israel and Singapore are evidently skilled with the application of support to their defence sector companies, and as a result they each have advanced equipment and solid exports.

It seems obvious to me that defence ministries that engage fund nurture and cooperate with the nation’s defence industrial sector encourage coordinate and channel innovation into proper leading edge desirable equipment best suited to the need of the time. It also seems obvious that not investing; leaving the businesses to invest their own R&D in time & effort necessary to create a world beating cutting edge production-ready product is naive in the extreme, if not la-la-land. Why does it surprise MOD every time they find UK industry has little to offer as modern as that available from the likes of the US or Israel etc – the UK companies are most likely to offer their off the shelf decades old product with minor changes or add-ins to scrape a degree of compliance because that’s all the available R&D could stretch to.

I recall the Centre for Defence Excellence (the MOD’s portal for exciting new innovations in military engineering) making a really big deal of buying in data-loggers to fit to army vehicles. It was for a time the pin-up project. One of my friends who races motorbikes bought a datalogger and its sensors off the web and over a weekend fitted the sensors and processor to the bike and made the wiring harness and got the system running and calibrated the software to suit his application. The MOD systems might have had a few more sensors and might have included more functionality, but really is this sort of ‘innovation’ an example of the best that UK defence investment can now manage?

April 6, 2014 5:31 pm

I wonder, how modular the software of the F-35 is.

It seems, 3i is a complete rewrite rather than an adaption of hardware-accessing-(aka driver-)side of things. If that would be the case, this would be against any development principle I can think of. It shouldn’t matter for the A2A-functions, what processor the underlying OS is running on. It shouldn’t matter, what sensor lingers on the other side of the driver-interface.

I hope, that is not the case.

The Bloomberg-article states, there are problems with the test cycle. This is a rather wide definition. Normally, customer requirements are grained into unit-tests, C++ supports that kind of thing, more or less.

If the problems arise on the systems integration test side, then we have a communications problem requiring major validation and rework, at best. In the worst case, subsystems or hardware is misbehaving. Then we truly have a problem.

If the problems are simply “it’s built, it’s working, but we don’t get the verification, because it doesn’t fly”, then it’s rather a theoretical problem.

Rocket Banana
April 6, 2014 6:12 pm

If Microsoft or Oracle can’t truly integrate their software portfolio whatever makes people think that Lockheed will?

Does anyone know what kernel the F35 systems are running on?

Not that I expect a response, but it’s not trivial writing an OS kernel that allows high-speed integration (required for sensor fusion) and the separation/encapsulation necessary to guarantee quality and software stability.

How may iterations did it take Microsoft to come up with DirectX… 11 and counting? That’s the kind of integration layer performance required for the data bandwidth of AESA fusion. Alternatively they could buy the IPR to Oracle’s cache-fusion which is similarly clever but will support shared-everything scalability.

Just like most things though I’d suggest the 80:20 rule applies. Spend 20% of the money to get 80% of the capability. That means they’re about 40% of the way through in terms of resource outlay at 88%.

The Other Chris
April 6, 2014 6:14 pm

Green Hills Software “Integrity” Real-Time OS.

Rocket Banana
April 6, 2014 6:35 pm


Any idea what language it’s all written in?

The Other Chris
April 6, 2014 6:57 pm

Yes, C++ using Green Hills Software own IDE for the majority development. In some cases, such as the Field Programmable Gate Arrays, there specific languages involved. Main hardware in the ICP (Integrated Core Processor) are Motorola PowerPC.

Main feature is the real-time nature with everything running being deterministic. It’s phenomenally fast compared to a consumer/business OS and not directly comparable.

As an aside (I’m using that phrase a lot lately, sorry): Java, Windows, .Net etc are not allowed to be used in aircraft navigation, life support, weapons systems, nuclear facilities. It’s even in their EULA’s.

The Other Chris
April 6, 2014 7:00 pm

Correction: I’m aware of “Windows for Warships”, I intended to indicate the generally available OS :)

Rocket Banana
April 6, 2014 9:23 pm

The Other Chris,

Interesting. So how can GIPC be a) deterministic, and b) so fast when it has to cross a segmented memory boundary (task-switch latency and all that)?

I like the idea of Integrity but think its very design (security) prohibits high-speed interchange of data between discrete processes – something I am admittedly assuming is being done in order to segregate the development and testing cycles.

Personally I’d have gone for BASIC on a 6502 ;-)

The Other Chris
April 6, 2014 10:07 pm

Semaphores and pipes are supplied for Inter Prices Communications. The rest is architecture. e.g. shared-nothing.

The Motorola’s are one half of the Integrated Core Processor. The other half are the Xilinx Virtex 7 Field Programmable Gate Arrays to handle the number crunching on the image processing and fusion.

No idea if they’re using the specific VHDL language or a C compiler for those. There’s a lot of supported ways to instruct the sets. They’ve been proud of the choice of C++ on the program to increase access to skilled resource and allow faster iteration, but that doesn’t mean they wouldn’t pick a more appropriate language for a task if needed.

Virtex is one of those pieces of hardware I wish I had a use for and the budget to play with!

As for BASIC? I’all always have a soft spot for my Acorn too :)