Personal tools

Difference between revisions of "Adding Linux Support to our DMA engine (1-2S/B)"

From iis-projects

Jump to: navigation, search
 
Line 11: Line 11:
 
== Introduction ==
 
== Introduction ==
  
 +
At IIS we are developing a modular and extensible high-performance direct memory access (DMA) engine. So far we have integrated it in our cluster-based systems. On those systems are programmed using bare-metal code to achieve the highest energy efficiency.
 +
 +
Next to the cluster-based systems, we are working on Ariane (CVA6), a 64 bit application-grade linux-capable core.
 +
 +
We are now planning to integrate our DMA engine in an Ariane-based SoC and evaluate its benefits in a Linux environment.
  
 
== Project ==
 
== Project ==
  
 +
We propose to divide the project in these steps:
  
 +
* '''Research''': Investigate how Linux interacts with a system-level DMA and collect a set of minimal requirements for the hardware.
 +
* '''RTL design''': Create a hardware interface for the DMA engine that features this set of minimal requirements Linux opposes on the hardware.
 +
* '''Validation and driver development''':  Create a driver for Linux allowing it to interface with the DMA; validate your hardware using simple transfers
 +
* '''Evaluation''': Run complex applications to evaluate the performance of the DMA in this context.
  
 
== Requirements ==
 
== Requirements ==
  
 +
* Experience with digital design in SystemVerilog as taught in VLSI I
  
 
== Composition ==
 
== Composition ==
  
 
+
* 30% studying the DMA interface imposed by Linux
 +
* 20% RTL design
 +
* 30% driver development & validation
 +
* 10% evaluation
  
 
== Project Supervisors ==
 
== Project Supervisors ==

Revision as of 16:23, 19 November 2021


Introduction

At IIS we are developing a modular and extensible high-performance direct memory access (DMA) engine. So far we have integrated it in our cluster-based systems. On those systems are programmed using bare-metal code to achieve the highest energy efficiency.

Next to the cluster-based systems, we are working on Ariane (CVA6), a 64 bit application-grade linux-capable core.

We are now planning to integrate our DMA engine in an Ariane-based SoC and evaluate its benefits in a Linux environment.

Project

We propose to divide the project in these steps:

  • Research: Investigate how Linux interacts with a system-level DMA and collect a set of minimal requirements for the hardware.
  • RTL design: Create a hardware interface for the DMA engine that features this set of minimal requirements Linux opposes on the hardware.
  • Validation and driver development: Create a driver for Linux allowing it to interface with the DMA; validate your hardware using simple transfers
  • Evaluation: Run complex applications to evaluate the performance of the DMA in this context.

Requirements

  • Experience with digital design in SystemVerilog as taught in VLSI I

Composition

  • 30% studying the DMA interface imposed by Linux
  • 20% RTL design
  • 30% driver development & validation
  • 10% evaluation

Project Supervisors