Menu

Deep Neural Networks on Edge Devices with Vitis AI

Webinar


The pervasive use of AI in imaging tasks is showing in all markets and many products need to provide similar functionality. Often the core technologies are not easily available in scalable implementations, and this is where AMD FPGAs or Adaptive SoC can provide excellent solutions.
We will show how convolutional neural networks (CNN) models can be accelerated on AMD Adaptive SoCs with the help of the Vitis AI toolchain. The focus of this talk is set on understanding the modeling framework topology and parameter database based on simple models to gain insight into what a good fidelity-trained floating point model would resemble. This performance needs to be validated in the translation process through Vitis AI, like checking and correlating accuracy before and after quantization. A translation into microcode for the DPU is done by the Vitis AI compiler. Demonstrating the execution when finally running a model on the Deep Learning Processing Unit (DPU) inside the devices, the performance gain will become evident. The session closes by discussing how to integrate the DPU accelerator into an actual design and provides insight into pre-processing and post-processing approaches for data preparation.

Portrait von Michael Schwarz

Contact

Michael Schwarz

+49 7664 91313-15

michael.schwarz@plc2.de

This might also interest you...

Course Format Category Location Duration Date

Compact Vitis AI

OL (Online Live)

Development
Tools & Methodology

Online

3 days

Nov 04, 2024

Info

Vitis AI AI Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Kernel c/c++ machine learning ML inference on the edge neuronal networks CNN Deep Learning Processing Unit DPU quantization embedded software

Compact Vitis AI

WO (Workshop)

Development
Tools & Methodology

Freiburg

3 days

Jun 10, 2024

Info

Vitis AI AI Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Kernel c/c++ machine learning ML inference on the edge neuronal networks CNN Deep Learning Processing Unit DPU quantization embedded software

Compact Vitis AI

WO (Workshop)

Development
Tools & Methodology

Stuttgart

3 days

Sep 23, 2024

Info

Vitis AI AI Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Kernel c/c++ machine learning ML inference on the edge neuronal networks CNN Deep Learning Processing Unit DPU quantization embedded software

Compact Vitis AI

WO (Workshop)

Development
Tools & Methodology

Berlin

3 days

Nov 04, 2024

Info

Vitis AI AI Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Kernel c/c++ machine learning ML inference on the edge neuronal networks CNN Deep Learning Processing Unit DPU quantization embedded software

Compact Vitis for Acceleration

OL (Online Live)

Development
Tools & Methodology

Online

3 days

Nov 20, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Kernel Boot Bootflow elf c/c++ Debugging

Compact Vitis for Acceleration

WO (Workshop)

Development
Tools & Methodology

Berlin

3 days

May 27, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Kernel Boot Bootflow elf c/c++ Debugging

Compact Vitis for Acceleration

WO (Workshop)

Development
Tools & Methodology

Freiburg

3 days

Aug 12, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Kernel Boot Bootflow elf c/c++ Debugging

Compact Vitis for Acceleration

WO (Workshop)

Development
Tools & Methodology

Frankfurt / Main

3 days

Nov 20, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Kernel Boot Bootflow elf c/c++ Debugging

Professional Vitis

OL (Online Live)

Development
Tools & Methodology

Online

5 days

Nov 04, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Boot Bootflow elf c/c++ Standalone Debugging Profiling Acceleration XRT

Professional Vitis

PW (Power Workshop)

Development
Tools & Methodology

Stuttgart

5 days

Jun 24, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Boot Bootflow elf c/c++ Standalone Debugging Profiling Acceleration XRT

Professional Vitis

PW (Power Workshop)

Development
Tools & Methodology

Freiburg

5 days

Aug 05, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Boot Bootflow elf c/c++ Standalone Debugging Profiling Acceleration XRT

Professional Vitis

PW (Power Workshop)

Development
Tools & Methodology

Berlin

5 days

Nov 04, 2024

Info

Vitis Embedded Software Tool Flow Blockdiagram SoC MPSoC Versal ACAP C C++ BSP Driver Board Support Package Profile Profiling Debugger System Debugger GCC GNU Eclipse Library Linux PetaLinux Boot Bootflow elf c/c++ Standalone Debugging Profiling Acceleration XRT

showing: 1 to 2 (2)

How can we help?

Contact

FAQ

01. Can you help me with reserving the room?

If you need a room for a face-to-face course, please contact us. We will support you in choosing suitable accommodation.

02. What times in the day do courses start and end?

Our face-to-face courses start daily at 9 a.m. (CET/CEST) and end at 5 p.m., with two coffee breaks and an extended lunch break. The total course time per day is 6 hours. For online courses, the start is also at 9 a.m. (CET/CEST), but they end at 2:30 p.m.. The total course time per day is 4 and a half hours plus breaks.

03. Are course materials provided?

All participants of paid courses will receive English training materials in electronic or paper form.

04. Will I receive a course certificate?

Each participant of a paid PLC2 course will receive a signed certificate of attendance at the end of the course.

05. Can I pay with Training Credits (TCs)?

Of course, you can also pay for your training using your available TCs (Training Credits). We will then invoice AMD directly. If you do not have sufficient credit for the booked training, the difference can be paid for with an additional payment. The additional charge will be invoiced directly to you.

06. What are the options for payment?

You can pay us via bank transfer or with TCs (Training Credits). Unfortunately, payment via debit, credit card, or PayPal is not possible.

07. What is the deadline for payment?

Payment is due 14 days after the invoice date, without deduction. Please transfer the amount within that time.

08. What are the options for cancelling?

You can cancel free of charge up to 14 days before the course starts. If you have to cancel at short notice due to illness or other reasons, you have the option of adding a substitute participant or of making up the course at a later date (within 6 months).

09. When will I receive a firm commitment as to whether the course will take place?

You will receive binding confirmation 7 days before the course begins as to whether it will go ahead.

10. How many participants are required for a course to be given/what is the minimum number of participants?

We generally offer courses for 5 or more people. That means that if you would like to book a course at your premises, you need 5 people to register or need to pay the amount for 5 participants, even if fewer people attend.

11. Can I book a course that is given in English?

If you need a course that is officially advertised as given in German to be given in English, please contact us. We will try to make that possible for you.

12. What language is the course in?

Our webinars and online courses are conducted in English. All our F2F training courses are generally conducted in German. Feel free to contact us if you need face-to-face training in English or online training in German.