SDAccel Design Contest: How to use SDAccel on AWS
0 likes199 views
This document outlines a seminar on using Xilinx SDAccel on AWS. The agenda includes creating an AWS account, launching EC2 instances, deploying SDAccel with the HDK, developing kernels on AWS, creating AMIs, and executing kernels on AWS FPGAs. It provides step-by-step instructions on setting up accounts, instances, and environments for FPGA development and execution on AWS.
SDAccel Design Contest: How to use SDAccel on AWS
- 1. Courses @ NECST Lorenzo Di Tucci <[email protected]> Emanuele Del Sozzo <[email protected]> Marco Rabozzi <[email protected]> Marco D. Santambrogio <[email protected]> How to use Xilinx SDAccel on AWS 22/02/2018 DEIB Seminar Room
- 2. 2 Agenda - Create an account on AWS - Launch an EC2 compute instance - Manage running Instances - Connect to the AWS instance using SSH/GUI - Deploy Xilinx SDAccel on AWS with the HDK - Develop your kernel on AWS - Create an AMI and execute your kernel on AWS
- 3. 3 AWS: how to register -
- 5. 5 AWS: how to register
- 7. 7 AWS: Credit Request • Request credits on https://aws.amazon.com/education/awseducate/ and then click Apply for AWS Educate for Educators link under the Educators section • Select the appropriate role (Educators or Students)
- 8. 8 AWS: Credit Request • Fill out the form and enter the AWS account number that you created in the previous step
- 10. 10 AWS: Services
- 13. 13 AWS: Amazon Machine Image
- 14. 14 AWS: FPGA Developer AMI
- 15. 15 AWS: FPGA Developer AMI
- 17. 17 AWS: Instance Selection ● f1.2xlarge: AWS instance with a single FPGA for design deployment ● f1.16xlarge: AWS instance with 8 FPGAs (really expensive!) ● c4.4xlarge ○ Ideal instance for design development ○ much cheaper than a f1.2xlarge instance
- 19. 19 AWS: Storage
- 20. 20 AWS: Security Group ● Add a rule to allow RDP connection ○ Once RDP connection is allowed to our instance we can use remmina/remote desktop to connect to it
- 21. 21 AWS: Security Group ● Add a rule to allow RDP connection ○ Once RDP connection is allowed to our instance we can use remmina/remote desktop to connect to it
- 23. 23 AWS: generate key pair
- 24. 24 AWS: generate key pair
- 28. 28 AWS: Instance Access ● Connect to the instance with ssh -i <pem_file_location> centos@<public_ip> ● The PEM file must have specific permissions
- 29. 29 AWS: Instance Access (Linux) ● Connect to the instance with ssh -i <pem_file> centos@<public_ip> ● The PEM file must have specific permissions: ○ chmod 400 <pem_file>
- 30. 30 AWS: Connect with PuTTy (Windows) ● Select SSH Connection Type ● Specify centos@<address> ● Port Number : 22 ● Under Connection, select SSH and then Auth ● Click on Browse and select the Private Key ● Now you can connect to the instance!
- 31. 31 AWS: GUI Setup ● in src/GUI_README.md is possible to find instructions on how to setup a GUI for an instance
- 33. 33 AWS: GUI Setup ● Remember to setup a password for user “centos” to connect via RDP: ○ sudo passwd centos
- 36. 36 AWS: RDP Access ● Connect using user “centos” and the password specified before
- 37. 37 AWS: GitHub
- 38. 38 AWS: GitHub
- 40. 40 AWS: how to register
- 42. 42 AWS: SDAccel Setup Source this script to setup all the environment for SDAccel on your instance
- 45. 45 AWS: Makefile
- 47. 47 AWS: Build
- 50. 50 AWS: Amazon FPGA Image ● It is required to create an AFI to execute FPGA binaries on AWS F1
- 51. 51 AWS: Amazon FPGA Image
- 52. 52 AWS: Amazon FPGA Image First, create your credential from the aws website
- 63. 63 AWS: AFI Registration Status
- 64. 64 AWS: AFI Registration Status
- 65. 65 AWS: Save Host and AWSxclbin
- 66. 66 AWS: Save Host and AWSxclbin
- 67. 67 AWS: F1 Instance Creation
- 69. 69 AWS: F1 Instance ● Request a new F1 instance as we did before for the c4.2xlarge
- 70. 70 AWS: F1 Instance Setup
- 71. 71 AWS: Launching F1 Instance
- 72. 72 AWS: Launching F1 Instance
- 73. 73 AWS: Copy Host and AWSxclbin
- 77. 77 AWS: Running on FPGA
- 86. 86 Remember to terminate your instance
- 87. 87 Remember to terminate your instance
- 88. 88 Feedbacks • We are working at improving this course, would you share your feedback for this lesson? https://goo.gl/forms/29nHmTG0oP7WDKrb2
- 89. Thank You for the Attention! 89 Lorenzo Di Tucci [email protected] Emanuele Del Sozzo [email protected] Marco Rabozzi [email protected] Marco D. Santambrogio [email protected]