A bug has already been found in the new ECU

You mean the NASCAR button?

 

from the early testing, where some teams were apparently able to do good launches, and some were hampered, it was a matter of time before the cat was entirely outta the bag... so it really does appear that when certain drivers and teams 'noticed' peculiarities during testing, they weren't just imagining the whole thing...

me, i'm not surprised. i would be surprised however, if the ECU's weren't exploited even further throughout the season. i had an inkling way back when, that somehow the teams would be able to 'tweak' it beyond the 'spec' unit...

 

I can't believe anyone who knows anything about computer programming or ECU's would give this rumour any credence whatsoever.

• The FIA certified the source code.
• All teams have access to the source code, and teams of programmers who would know what to look for.
• It might be possible to hide an access point to a hidden routine (though I doubt it), but that routine couldn't do much because that would be easy to spot if it were more than a few lines long.
• Traction Control and Launch Control algorithms are long, complicated and require tuning to each different engine/vehicle, there's no way you could hide one in the code and expect it to work.
• Launch Control requires a less agressive map, not a more agressive one. More agressive = less traction not more.

 

WHAT???:: Microsoft doing Open Source ????!!!!

That sounds as unlikely as turning on launch control by ctl-alt-del.

 

Sources for the first three points? My understanding is that an employee of MCL Systems is assigned to each team to manage the communication between the team engineers and the MCL guys who have access to the code. Points 4 and 5 are valid, but even if it didn't work, you could at least notice that the car behaves as if it had a poorly designed traction control (which would actually be the case).

 

For the first two my source is the FIA tender document for the ECU.

Also for reference

Obviously doing so would be in breach of the contract and I would guess would involve some pretty hefty penalties. Despite what people might think McLaren Electronic Syetems is not the same company as McLaren F1, they do share the same parent company but operate independently.

The source for the third point is my 30 years of systems programming experience, including about 10 years working with embedded systems. Plus about 15 years of playing around and tuning with different ECUs for my own amusement. Also the specifcations and statement "The algorithms are designed to provide the necessary control with the minimum of complexity." from the tender document would leave little room to hide code without the possibilty of detection.

 

I may be wrong, but I have read that the standard ECU has 12 maps that the teams can program themselves (within reason). The easiest way of building in TC, is to configure one of the maps with retarded ignition & reduced fueling (TC map).

Then build into the other maps a plot of the maximum acceleration of the engine that represents the limit of traction in each gear. This can be done from testing. When the engine accelerates faster than this (ie wheel spin) the ECU swaps map to the TC map. This is a crude sort of TC that does not rely on any outside input ie, sensors on the wheels.

The driver can only do this via the wheel switches .

Mike Gascoigne has hinted that some teams may be doing this.

 

They do have the ability to change fueling, but can only select from two base ignition maps, and even the only when the car has been stationery for 3 seconds.

So while they could reduce power by changing fuel maps, it wouldn't be a very good TC or LC alternative. You really need ignition cut for effective TC and that would be very obvious. The driver has a much better system for both TC & LC... his right foot

 

That's what we'd call a "request for proposals". In the final negotiations, several of those requirements could have been negotiated. Given MS's relationship to "open-source", it'd be interesting to see if that provision held on the actual contract.

Heck, they don't even give you useful documentation with most software.

Besides: a federal court ordered the break-up of Microsoft, due to anti-trust actions, and you see how much influence that had.
How much would MS worry about FIA's lawyers?

 

That's really clutching at straws. IME RFP's might be subject to negotiation before signing a contract but key points like the "open source" requirement, which is needed for the teams to be able to use the system to it's fullest extent, are generally non-negotible. I'm sure the teams would be screaming loudly if they didn't have access to the source code.

You should remember that it's Microsoft in name only. The software might be based on Microsoft systems, but I doubt the Microsoft has any control over the software development, and it's not Microsoft that would bear the brunt of any FIA action, it would be McLaren Electronic Systems (TAG?).

 

I've worked tech contracts from both sides. You might be surprised how much gets changed between the RFP and the contract -- in both directions. Sometimes, things put in the proposal that weren't RFP requirements get added as contract requirements. And sometimes the bidder makes the case that an original "requirement" isn't really needed.

After all, the premise that the source code is needed "to use the system to it's fullest extent" is something that many proprietary code suppliers would fight tooth and nail. If you really believed that, would you be running Windoze on your home computer?

(This post written on Linux. )

 

I wouldn't be surprised how much some contracts get changed, after 30 years in software I have worked on both sides of RFPs many times. However I would be very surprised if this requirement changed. Can you imagine the stink created by the other teams about the advantage McLaren was getting because it had access to the source to understand the workings of the ECU. Let alone the innuendo about McLaren having access to hidden features. Also note the FIA's requirement to sign off on hardware and software design and implementation, or do you think they would’ve dropped that one too

OS and ECU software are worlds apart in both function, complexity and intended market. Besides which I've seen plenty of proprietary code developers supply the source code because it was a requirement of the purchaser. IIRC the teams have had to provide access to source code in the past, so it’s no as though it’s a new requirement.

Anyway I've provided proof that was what the FIA required of the supplier, and, until such time as someone can provide evidence to the contrary, then that's the best information available.

So have you got any proof that the FIA dropped the requirement, or are you desperately seeking justification for continuing the current McLaren witch hunt?

Regardless of that, given that the original rumour had other glaring inaccuracies, I can’t see any reason to give it any credibility at all

 

if there's one thing that is ABSOLUTELY CERTAIN, it is that every team 'cheats' (pushes boundries), or at least tries to, as far as they can get away with. i would not be surprised if there were a few things to 'work around', despite the assurances that they all have 'standard' units. no one here knows every single detail behind the whole deal.

 

Imagine the weight of that car

 

That was my point: the RFC is circumstantial evidence, not "proof". The contract would be direct evidence, but that doesn't seem to be public information. (The contract isn't "open source". )

Yes, without the contract text, the RFP is the best we can go by, unless we hear from the teams themselves.

But microsoft and open source .... Well, if there was a news report of Hillary calling for an escalation of military involvement in Iraq, there are people who would want to see video of the speach.

It doesn't matter whether I like or dislike McLaren. Having a company associated with one competitor supplying equipment to all of them is a conflict of interest -- regardless of which team. How many times had you heard announcers speculate that Sauber wouldn't hold up a Ferrari -- when Ferrari was their engine supplier? I wouldn't be any happier with Ferrari or Toyota involved in the "spec" ECUs.
(Mitsu, on the other hand, has that nice Renesas R32 processor ... )

Besides: if it comes to a witch hunt, it's not McLaren that I'm inclined to see as the wicked wiccan of the west in that partnership.
(But coworkers know me as a vocal advocate of *nix systems.)

And the irony of the "three button" comment is enough to make this report sound like a joke. If it were real, why is there no mention on Honda's website? I'm not taking it seriously. But me turn down a chance to make fun of MS code?

 

That would never work. Pit stops would take 3 hours while the mechanics try to sort out wheel/nut/spindle dependencies, the car wouldn't be able to run on most tracks and the drivers would always be late and perpetually underdeveloped.

This still would be better than a Vista based car, which would be heavier, slower, ridden throughout with bugs, constantly maligned by the press, and generally a very bad car. In short, it would be last year's Honda.

 

hmmmm.......Here's what I said a couple months ago on the subject....I even called the cheap they would slide in, becasue it's the one I would slide in


http://www.ferrarichat.com/forum/showthread.php?t=180892&highlight=ecu&page=3

 

If timing can't be altered to reduce power that next best thing is dump in fuel. You can easily drop 15% of your hp that way without fouling plugs, I'd guess 20-30% if you limit the time because the plugs will load up.

If you try and go lean you don't lose power as fast, cook the engine and cause missing that would be obvious.