Download presentation
Presentation is loading. Please wait.
Published byJonathan Cooper Modified over 9 years ago
1
Alon Horn and Oren Ierushalmi Supervised by Mony Orbach Winter 2010 Characterization Presentation Implementation of an Engine Control Unit over Many-Core System
2
Project Objectives Implement an Engine Control Unit (ECU) over a Many Core System. Understand Plurality's HAL Architecture. Write Parallel Code using Task Oriented Programming. Simulate the Code Using Plurality's cycle accurate HAL simulator. Achieve Fast, Accurate, Robust and Efficient system.
3
Engine Control An ECU reads data from various engine sensors, processes them to calculate fuel injection pulse and spark timing, then sends back 3 control signals: Fuel pulse width. Spark timing. Stepper IAC – Idle air controller. ECU Engine
4
1) Engine Main Sensors M.A.P sensor – Manifold Absolute Pressure. C.P.S – Crankshaft Position Sensor. Sync Signal – synchronization signal from the distributor. M.A.T – Manifold Absolute Temperature. T.P.S – Throttle Position Sensor. E.G.O – Exhaust Gas Oxygen feedback.
5
Project Top Level Project's Main Blocks: HAL Simulator Engine Sensors Outputs Checker Golden Model 1 2 4 5 TOP C code 3
6
2) HAL HAL - Plurality's HyperCore Architecture Line Main Features: 16 to 256 32-bit RISC cores supports a task- oriented programing. Shared memory architecture. Hardware-Based scheduler. HAL simulator is a GDB based cycle-accurate simulator for Plurality's Many Core architecture.
7
2) HAL
8
3) TOP C Code TOP – Task Oriented Programming. Task is a code section that runs sequentially. Different or duplicable Tasks can runs simultaneously. A Task Map defines the relations and conditions between tasks.
9
3) TOP C Code – Main Blocks Inputs - sensors Tasks – outputs (Realtime) Tasks - sensor calculations Outputs - Control signals Tasks – sync
10
3) TOP C Code – ECU Sensor and Outputs level granularity. Independent calculations will be done in separate tasks. Each output has separate task. Syncronization / scheduler tasks as needed. Simple and logical design. Readable code.
11
4) Golden Model An independent model that calculates the outputs of the simulator. Should be simpler to code and debug. Will be implemented by an off-line Matlab script or an on-line/off-line C code.
12
5) Outputs Checker The Checker will compare the outputs from the HAL simulator and the Golden model and will write the result to a text file. It will be done by collecting the outputs timing and comparing on-line with the Golden Model outputs.
13
6) Time Plan http://diglab.technion.ac.il/Projects/Folders/D01109/project-A.html
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.