Adding load columns and RPM rows to maps

Reply Subscribe

Thread Tools

Search this Thread

Oct 5, 2009 | 04:07 PM

#211

Evo_Jay

Evolved Member

iTrader: (17)

Joined: Jun 2005

Posts: 3,419

Likes: 14

From: Chico, CA (NOR-CAL)

Quote:

Originally Posted by mfr122887

These are the stepper tables and decel fuel cut tables from 94170015 on the left and 96530706 on the right. They are extremely different, I am going to set them equal and go for a drive. Has anyone done this?

edit: just realized that you cant see the numbers but just by the colors they are drastically different. I'll get a better screenshot up in a little while.

Why are the "decel fuel cut delay" tables for 9417 vertical and not horizontal like the other ROM.

Also, like you said, a better screen shot would be nice.

edit- Also, can i get the info for the "decel fuel cut delay" table in 9417? I have them for other ROMs but not that one.

Oct 5, 2009 | 04:19 PM

#212

mfr122887

Evolving Member

iTrader: (14)

Joined: Feb 2006

Posts: 305

Likes: 0

From: Westchester, NY

Quote:

Originally Posted by Evo_Kid

They have the same axis but are displayed vertically. Not sure why, that's how the xml was when i found it.

Here they are for 94170015:

<table name="Decel Fuel Cut Delay - Neutral A/C Off" category="Fuel" address="42bc" type="2D" level="2" scaling="uint8">
<table name="Load" address="675e" type="Y Axis" elements="7" scaling="Load"/>
</table>

<table name="Decel Fuel Cut Delay - In Gear A/C Off" category="Fuel" address="42ca" type="2D" level="2" scaling="uint8">
<table name="Load" address="675e" type="Y Axis" elements="7" scaling="Load"/>
</table>

<table name="Decel Fuel Cut Delay - Neutral A/C On" category="Fuel" address="42d8" type="2D" level="2" scaling="uint8">
<table name="Load" address="675e" type="Y Axis" elements="7" scaling="Load"/>
</table>

<table name="Decel Fuel Cut Delay - In Gear A/C On" category="Fuel" address="42e6" type="2D" level="2" scaling="uint8">
<table name="Load" address="675e" type="Y Axis" elements="7" scaling="Load"/>
</table>

I am going to work on a better screenshot, it's coming off a netbook so I have to figure out a good way to do it.

Oct 5, 2009 | 04:21 PM

#213

Jack_of_Trades

Thread Starter

Evolved Member

iTrader: (17)

Joined: Jun 2007

Posts: 3,523

Likes: 2

From: Opelika,AL

Quote:

Originally Posted by Evo_Kid

Why are the "decel fuel cut delay" tables for 9417 vertical and not horizontal like the other ROM.

You can choose which axis is the X axis by right clicking the table and editing the Map Definition where it says "TYPE: Y AXIS".

Oct 5, 2009 | 04:31 PM

#214

Evo_Jay

Evolved Member

iTrader: (17)

Joined: Jun 2005

Posts: 3,419

Likes: 14

From: Chico, CA (NOR-CAL)

Quote:

Originally Posted by mfr122887

thanks

edit - oh you have them defined a lil differnt then my other ROMs do. Im my ROms where these tables are defined, they are just numbered #1-4, and dont say "in gear A/C on/off" and "Neutral A/C On/off.

Seems like know those names will help a lil bit in figuring out how to get rid of the hang.

Last edited by Evo_Jay; Oct 5, 2009 at 04:35 PM.

Oct 5, 2009 | 04:33 PM

#215

Evo_Jay

Evolved Member

iTrader: (17)

Joined: Jun 2005

Posts: 3,419

Likes: 14

From: Chico, CA (NOR-CAL)

Quote:

Originally Posted by Jack_of_Trades

You can choose which axis is the X axis by right clicking the table and editing the Map Definition where it says "TYPE: Y AXIS".

I understand it can be changed in the map def.

I was just wondering why his was vertical because I have the "decel fuel cut delay" tables defined in other ROMs and they are all horizontal.

Oct 5, 2009 | 04:33 PM

#216

phenem

Evolved Member

iTrader: (39)

Joined: Jul 2005

Posts: 811

Likes: 4

From: Central PA

Quote:

Originally Posted by Evo_Kid

I dont think that that table is deffined correctly or something.

In the begining of ECUflash, when I had my IX, I did a lot of messing with the ISC stepper table. I was able to find a good balance of gettin rid of the throttle hang and still having cold start, but I dont think that is the correct way to get rid of the throttle hang. Like I said above, I think it has more to do with the "Decel fuel cut delay" tables.

Also when i tried the ISC Stepper setting that I used in my car to get rid of throttle hang on another car, they experienced poor cold start and said throttle hang remained.

We are way off topic here so I decided to create another thread to talk about the throttle hang / idle stepper table.

Here is that thread: https://www.evolutionm.net/forums/ec...ttle-hang.html

Oct 5, 2009 | 08:27 PM

#217

Evo_Jay

Evolved Member

iTrader: (17)

Joined: Jun 2005

Posts: 3,419

Likes: 14

From: Chico, CA (NOR-CAL)

Quote:

Originally Posted by phenem

We are way off topic here so I decided to create another thread to talk about the throttle hang / idle stepper table.

Here is that thread: https://www.evolutionm.net/forums/ec...ttle-hang.html

I know, great idea on starting another thread.

Oct 5, 2009 | 10:35 PM

#218

codgi

Evolved Member

iTrader: (22)

Joined: Aug 2004

Posts: 2,493

Likes: 41

From: Seattle, WA

Quote:

Originally Posted by logic

Geeky info: most basic SH instructions are executed in a single cycle, with multiplicative instructions taking either 2 cycles (for 32-bit results) or 2-4 cycles (for 64-bit results). For a 7052 or 7055 running at 40MHz, that's 25ns per instruction (cycle).

It's really a rather impressive little CPU, given it's age.

codgi: the table lookup itself is linear, but each axis lookup is a loop over it's length (short-circuiting when a match is found).

That's a single pass loop though? Basically an iterative search which would imply its linear as well.

Quote:

Originally Posted by logic

That's what it looked like to me. It struck me as strange at first, since you'd think you'd want to conserve cycles at higher RPM, but then again, I don't think Mitsubishi engineers wrote any of the code we're looking at: I'm pretty sure a compiler did.

+1. The compiler doesn't know nor care...it just looked for the simplest way to express this in the lowest form it knew (assm/machine code). Of course if you did try to optimize for higher RPMs there's a constant perf hit from doing that check. At lower RPMs the perf hit from doing that is going to be higher than doing the iterative search today. At some point it would cross over, and where that point is would be key.

Anyway carry on

Oct 5, 2009 | 11:50 PM

#219

Ceddy

Evolving Member

Joined: Apr 2008

Posts: 265

Likes: 1

From: Reading, PA

Quote:

Originally Posted by logic

codgi: the table lookup itself is linear, but each axis lookup is a loop over it's length (short-circuiting when a match is found).

I don't think you guys will ever have to worry about the ecu performance problems from added code.

The 1G DSM and Evo1/2 use a 68HC11 style processor than ran at 2 Mhz, with 4-5 cycles per instruction average.

Besides the OBD2 and Emissions code, you are basically doing the same things, same map lookups etc.

I think if you could measure cpu idle time, you would be surprised how little of the cpu resources are being used.

Oct 6, 2009 | 05:50 AM

#220

logic

Evolved Member

iTrader: (2)

Joined: Apr 2003

Posts: 1,022

Likes: 7

From: Berkeley, CA

Quote:

Originally Posted by codgi

That's a single pass loop though? Basically an iterative search which would imply its linear as well..

Ugh, I should never post when I'm tired, I screw up things like "word definitions" and "sentence structure".

The table lookup itself is fixed-cost, ie. O(1), but each axis lookup is a search over the width or height of the table, so a maximum of O(n), but no worse. You had it right.