Puffwagon Posted June 30, 2020 Share Posted June 30, 2020 Here's a clip that I thought would interest a lot of people on this forum. Quote Link to comment Share on other sites More sharing options...
Roland@pcmtec Posted June 30, 2020 Share Posted June 30, 2020 Yeah so as he says WinOLS is really beyond 95% (or more) of tuners. Generating definitions from scratch is extremely time consuming. For example there are 165 operating systems from 2002 to 2016 in the Ford Falcon. Each of these has to be defined separately. We have written our own decompiler that pattern matches several calibration that we have defined by hand against all of the others, this makes the process much faster, but you still have to write the decompiler. There is probably 2 years full time work that went into just that one part of the software. Basically this job gets harder each year as PCMs get more complex, different security (and checksums) and they add more and more advanced algorithms and tables. The current 2018 Mustang for example has over 33,000 scalars and tables. This is a compilation of 100s of engineers work over 20+ years. It is not really possible for one (or even several) people to reverse engineer and understand all of this. Sure you can reverse engineer and define the tables, but how they all interact and work is incredibly complex. Even calibrators who work for large OEMs who have access to source code and calibration guides struggle to understand how a lot of it works at times. This is simply because of the vast amount of code that needs to be understood. Often you'll have a full team of engineers, each working on calibrating only one part of the PCM, they often don't understand what the other ones are doing as they don't have the time to learn everything. 2 Quote Link to comment Share on other sites More sharing options...
Puffwagon Posted June 30, 2020 Author Share Posted June 30, 2020 Yeah it's pretty crazy how complex it has become. After watching this clip it gave me a new appreciation for what you guys have achieved and also for being able to have the version of PCMTec that I have. 2 Quote Link to comment Share on other sites More sharing options...
Puffwagon Posted July 11, 2020 Author Share Posted July 11, 2020 Testing the browser time. Quote Link to comment Share on other sites More sharing options...
Roland@pcmtec Posted July 12, 2020 Share Posted July 12, 2020 3 hours ago, Puffwagon said: Testing the browser time. ? Quote Link to comment Share on other sites More sharing options...
Puffwagon Posted July 12, 2020 Author Share Posted July 12, 2020 8 minutes ago, Roland@pcmtec said: ? My browser was being a dickhead and posting the wrong time stamps on another forum. I just posted here to check if it was happening here. I think it would have taken 24 hours before the actual time came up on the post anyway. Quote Link to comment Share on other sites More sharing options...
Roland@pcmtec Posted July 12, 2020 Share Posted July 12, 2020 Timestamp will come from the server not your machine. Your machine will then translate it back to local time, so if it's wrong it is usually your time locale or actual clock is wrong. Quote Link to comment Share on other sites More sharing options...
Puffwagon Posted July 12, 2020 Author Share Posted July 12, 2020 I think it was because resistfingerprinting was true in the browser. Quote Link to comment Share on other sites More sharing options...
jakka351 Posted November 27, 2021 Share Posted November 27, 2021 On 6/30/2020 at 4:51 PM, Roland@pcmtec said: Often you'll have a full team of engineers, each working on calibrating only one part of the PCM, they often don't understand what the other ones are doing as they don't have the time to learn everything. My (non) understanding of it is that they designed basically as flow charts, which in turn each part of the flow chart is a reference to another flow chart, and so on and so on and so on until you reach flowchinifinity. It's quite impressive what PCMTec has done, it's one thing to figure out how to download and upload the data, but to then go on and define it and make use of it was likely not an easy process. Quote Link to comment Share on other sites More sharing options...
Roland@pcmtec Posted November 28, 2021 Share Posted November 28, 2021 17 hours ago, jakka351 said: My (non) understanding of it is that they designed basically as flow charts, which in turn each part of the flow chart is a reference to another flow chart, and so on and so on and so on until you reach flowchinifinity. It's quite impressive what PCMTec has done, it's one thing to figure out how to download and upload the data, but to then go on and define it and make use of it was likely not an easy process. That's why you often see 6 copies of the same variable, depends how many departments wrote blocks of code that need the same thing. Some of the code in the modern mustangs is shared all the way back to the 90s and uses the same struct definitions and even table values. Then you have 5 different tables types, Bosch tables (knock and low level stuff), old 90s ford float tables, modern ford uint16 tables with built in equation coefficients to turn them back into low effectively half word floats to save memory. Then you also have code and tables that are clearly simulink generated and absolutely impossible to follow the asm due to the incredibly inefficient code that it outputs. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.