CmpE 441 High Level Synthesis 2015 Spring

Course: 

CmpE 441 High Level Synthesis

Instructor: 

Arda Yurdakul

Assistants: 

Görker Alp Malazgirt

Course Schedule: 

MMT 457 at Computer Engineering Building A2 A2 A3

External Course Page: 

Course Page

Course Program: 

1. Introduction: 

Some Basics: Timing, Area, Power, Energy.

What is an Accelerator? What is its difference from coding for other computing units?

2. Fundamentals of High-Level Synthesis

From C code to Accelerator: Input, Output, Dataflow, Control, Scheduling, Resource Allocation.

Design Space. Objectives. Constraints. Compilation. Debug. Functional testing.

3. Revisiting Data Types, Arrays, Pointers for HLS

Limitations and their reasons: uninitialized variables, unbounded loops, system calls, dynamic memory allocation.

4. Revisiting Loops for HLS

Pipelining, Parallelism, Unrolling, Bounds, Dependencies between iterations, Optimizing Loop Counter, Optimizing Loop Control, Nested Loops

5. I/O Scheduling and Memories

Conditional and unconditional I/O. Stream I/O. Array to memory mapping. Designing for throughput.

Textbook: 

High-Level Synthesis Blue Book, Michael Fingeroff, ISBN-10: 1450097243 (available at the BoUn library)

Reference Books: 

- Vivado HLS User Manual (http://www.xilinx.com/support/documentation/sw_manuals/xilinx2014_4/ug902-vivado-high-level-synthesis.pdf)

- Vivado HLS Tutorial

 

Additional Material: 

(http://www.xilinx.com/support/documentation/sw_manuals/xilinx2014_4/ug871-vivado-high-level-synthesis-tutorial.pdf)

Notes: 

Grading: (Tentative)

·        60%: weekly implementations at the lectures

·        40%: final project: GROUP projects

o   10%: an initial report for project specifications (details will be announced)

o   15%: final demo

o   15%: final report submission (details will be announced)

Today, computing paradigm is shifting towards utilizing different types of processing units in the same platform. We see that Intel and AMD will produce their new chips with application processing units. These units are neither GPUs nor custom multicore processors. Application processing units are simply accelerators that speed up the computation while lowering the energy consumption.

Accelerators can be designed from C or Java by using high level synthesis tools. In this course, we will introduce the HLS concepts which can be used for an efficient accelerator design. Hands-on exercises in the lecture will also assist the course.

Contact us

Department of Computer Engineering, Boğaziçi University,
34342 Bebek, Istanbul, Turkey

  • Phone: +90 212 359 45 23/24
  • Fax: +90 212 2872461
 

Connect with us

We're on Social Networks. Follow us & get in touch.

About BOUN CmpE