Version: Vitis 2024.1
- Target board: VCK190
- Last update: 06 Sep. 2024
3.0 Decimation FIR Filter DSP Design
Developing a DSP algorithm with the AI Engine technology is only one part of the design flow and not necessarily the most complicated one. At the end of the day, the DSP kernels implemented into the AI Engine array have to be connected with the Programmable Logic (PL) datamover kernels and with the ARM A72 CPU host Programmable Subsystem (PS): the PL kernels require AMD Vitis™ HLS as design flow, the PS application requires C/C++ programming with either XRT or OpenCL APIs to manage the kernels at runtime execution on the target board within the PetaLinux OS.
Building a system from scratch like this is not trivial and it is better to have a good starting point that could be adapted with few changes, a sort of "system design template" or "toolbox" to use some improper terminology.
This tutorial explains how to design a DSP algorithm and implement it into the AI Engine domain of the AMD Versal™ device, using the production VCK190 target board and the AMD Vitis 2024.1 Unified software platform design tools.
The DSP algorithm is a decimation FIR filter taken from the Vitis 2024.1 DSP Libraries (for AI Engine).
You will build a system running on the VCK190 board having:
- A DSP kernel implemented on AI Engine domain (the decimation FIR filter)
- Two PL datamover kernels, designed with HLS
- Software application for the ARM host processor using the XRT APIs
- An SD card with PetaLinux OS for booting
As you use this document, it is assumed that you have named this tutorial repository 02-AIE_DSP_with_Makefile_and_GUI
and placed it in a certain working directory ${WRK_DIR}
. For example, in this case:
export WRK_DIR=/media/danieleb/DATA/2024.1/Vitis-Tutorials/Developer_Contributed
.
The organization of the folders tree of this tutorial is illustrated here (with emphasis on the folder related to AI Engine; otherwise, the tree would explode):
/media/danieleb/DATA/2024.1
|
├── Vitis_Libraries
│ ├── dsp
│ │ ├── docs
│ │ ├── ext
│ │ ├── L1
│ │ └── L2
│ │ ├── benchmarks
│ │ ├── examples
│ │ ├── include
│ │ ├── meta
│ │ └── tests/aie
│ │
├── Vitis-Tutorials
│ ├── AI_Engine_Development
│ │ ├── Design_Tutorials
│ │ ├── Feature_Tutorials
│ │ └── Introduction
│ ├── Developer_Contributed # the ${WRK_DIR}
│ │ └── 01-Versal_Custom_Thin_Platform_Extensible_System
│ │ └── 03-HLS_Code_Optimization
│ │ └── 02-AIE_DSP_with_Makefile_and_GUI # this tutorial
│ │ | └── files # current directory
-
Everything shown in this project used an Ubuntu 22.04 Desktop with related Vitis 2024.1 release.
-
It is recommended to write the SD card that boots the VCK190 board with the Windows OS utility called Win32 Disk Imager 1.0.
-
Some figures are screenshots related to the earlier release of Vitis and there might be a few differences, although minimal, with the current one.
-
It is recommended that you set correctly the environment before running any script, as described in details in the next subsection 2.3.6. The
*_sample_env_setup.sh
scripts contain a template to be customized for your needs. -
The run_all.sh must always be launched only after all the variables of previous script have been set at least once, and only from the files folder, which has to be your current directory.
-
Read the entire README.md document carefully before launching any script or Makefile.
In case you get some unexpected errors during the execution of the scripts, once pre-process all the*.sh
, *.cpp
, *.h
files with the dos2unix utility.
In that case, run the following commands from your Ubuntu host PC:
#sudo apt-get install dos2unix
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files
source scripts/dos2unix_all.sh
You need the following archives from either AMD/Xilinx Download or GitHub websites, in particular:
-
From Vitis (SW Developer) Downloads area, take the 2024.1 Vitis Installer (it makes the Vitis install process easy) and the 2024.1 Vitis Update (if any).
-
From Vitis Embedded Platforms area, take the Common Images for Embedded Vitis Platforms 2024.1.
-
From PetaLinux area, take the 2024.1 PetaLinux Tools Installer.
-
Go to the bottom of PetaLinux 2024.1 - Product Update Release Notes and Known Issues and take 2024.1_PetaLinux_Package_List.xlsx file, which contains all the packages needed by PetaLinux into your Ubuntu OS computer (install all of them before installing PetaLinux).
-
Go to the Xilinx GitHub page and zip both the Vitis Libraries and Vitis Tutorials with branch related to this 2024.1 release.
-
Go to the AMD Alveo™ Packages area, select release 2024.1 and Ubuntu 22.04 OS, and then take the Xilinx Runtime (XRT) archive.
At the end, you should have the following files:
Xilinx_Unified_2024.1_*_Lin64.bin
Xilinx_Vivado_Vitis_Update_2024.1_*.tar.gz
xilinx-versal-common-v2024.1_*.tar.gz
Vitis_Libraries-main.zip
Vitis-Tutorials-2024.1.zip
xrt_202410.*-amd64-xrt.deb
First, install the basic version of Vitis 2024.1 via the installer Xilinx_Unified_2024.1_*_Lin64.bin
. Then add the update Xilinx_Vivado_Vitis_Update_2024.1*.tar.gz
. Everything is placed in the folder /tools/Xilinx/
.
You might need sudo
privilege to install tools on these folders, primarily /tools
and /opt
, unless you change the ownership and group of folders accordingly.
If you created the /tools
directory as a super-user (or root
, or with sudo
), whatever you want to write or install there can only be done by the root
super-user.
This is a basic Linux OS behavior. However, you can change the group and owner so that you can do what you like as a normal user, the way you do it in your $HOME
directory.
So, if you run the following commands (and you need sudo
):
sudo su
mkdir /tools
mkdir /opt
# -R stays for recursively on each subfolder
chown -R you_user_name /tools
chgrp -R you_user_name /tools
chown -R you_user_name /opt
chgrp -R you_user_name /opt
exit
Then you can install the above tools and archives without sudo
privilege, just as a normal user.
As a last cross-check, if you installed the tools as a normal user, you should see the hidden folder .Xilinx
inside your $HOME
directory. If you installed them as a super-user, you should see
the hidden folder .Xilinx
inside your /root
directory.
In case of installation done with the sudo
privilege, there seems to be only a small issue with one line in the script installLibs.sh
:
su - $user -c "touch $logFile; chmod 777 $logFile"
The touch
and chmod
commands can not be executed because the logfile is located below the /root
directory and the shell was switched to a normal user without root privileges.
This means that the script assumes that the installation was done without root privileges. That line in the script could be replaced by the following commands:
sudo touch $logFile
sudo chmod 777 $logFile
In conclusion, either the installation had root privileges (sudo
) and the installLibs.sh
needs to be changed, or the installation was done as a normal user.
You have to install this archive xilinx-versal-common-v2024.1_*.tar.gz
in the folder /opt/xilinx/common/
folder (which has to be into an ext4
file system, not on an NTFS
one).
Then, you have to execute the following commands as a normal user, according to what is discussed in the previous subsection:
cd /opt/xilinx/common/xilinx-versal-common-v2024.1/
chmod 777 ./sdk.sh # if needed
./sdk.sh -p -y -d .
Check if you need to decompress the file rootfs.ext4.gz
directly.
You can then remove the sdk.sh
file to save storage space.
The easiest way to download XRT was from the ALVEO site. Select the ALVEO U200 card and then the Ubuntu 22.04 OS to arrive at this link.
Then install the archive via the following command:
sudo apt install --reinstall ./xrt_<version>.deb
To set xrt
into your terminal, you need the following command:
source /opt/xilinx/xrt/settings.sh
which is also in the *_sample_env_setup.sh
scripts.
WARNING: You cannot install PetaLinux into an NFS driver; otherwise, the install process ends with a non-predictable error message.
Before installing petalinux
, check in the Excel foil 2024.1_PetaLinux_Package_List.xlsx
what all packages petalinux
requires. Once done, install the missing ones.
#create the destination folder
sudo mkdir /petalinux_2024.1
#change permissions
$ chmod 777 ~/Downloads/petalinux-v2024.1-*-installer.run
$ ~/Downloads/petalinux-v2024.1-*-installer.run -d /petalinux_2024.1
. . .
INFO: Installing PetaLinux...
INFO: Checking PetaLinux installer integrity...
INFO: Installing PetaLinux to "/petalinux_2024.1/."
INFO: Installing buildtools in /petalinux_2024.1/./components/yocto/buildtools
INFO: Installing buildtools-extended in /petalinux_2024.1/./components/yocto/buildtools_extended
INFO: PetaLinux has been installed to /petalinux_2024.1/.
To set petalinux
into your terminal, you need the following command:
source /petalinux_2024.1/settings.sh
which was also placed in the script *_sample_env_setup.sh
.
The *_sample_env_setup.sh
scripts contain a template to be adapted by the user to set the tools environment for designing an embedded system on the VCK190 target board.
Once done, call the script with the following command:
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files
# set your environment
source files/scripts/<YOUR_OWN>_sample_env_setup.sh
In this section, you will build the decimation FIR filter DSP system using the Vitis GUI. The PS application will apply XRT APIs. The OS is PetaLinux.
Note that the classic Vitis IDE (Eclipse-based) has been deprecated and will be removed in a future release, being it replaced by the new Vitis Unified IDE. If you want to use the classic IDE you have to call it with vitis --classic
command as in the following:
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files
# set your environment
#source files/scripts/<YOUR_OWN>_sample_env_setup.sh
# call the Vitis GUI
cd gui-flow/fir-dsplib_prj
vitis --classic --workspace ./classic_xrt_wrk
Follow the following steps:
-
Select
File -> Application Project -> Next
from the GUI and then select thexilinx_vck190_base_202410_1
platform for your project. -
Click Next and give a name to your application project and your system project. For example,
dsplib_fir_AIE
anddsplib_fir_system
, respectively. -
Select
aie_engine
as the target processor and clickNext
twice. -
Select an
Empty application
template and clickFinish
. Following the above steps, you have created the folder for the AI Engine subsystem. -
Select
File -> New -> HW Kernel Project -> Next
from the GUI and then select thexilinx_vck190_base_202310_1
platform again for your project. ClickNext
. -
Give a name to your PL HW kernel project, for example
dsplib_fir_PL
. Ensure that you are still using thedsplib_fir_system
. ClickFinish
. With these steps 5 and 6, you have created the folder for the PL subsystem. -
Select
File -> New -> Application Project -> Next
from the GUI and then select thexilinx_vck190_base_202310_1
platform for your project. -
Click
Next
and givedsplib_fir_PS
as name to your application project and ensure that you are still using thedsplib_fir_system
. -
Select
cortexA72
as the target processor and clickNext
. Then select thexrt domain
. -
Set the path names for the
Sysroot, Root FS, Kernel Image
of the embedded OS. Then clickNext
. Here are the environment settings for your reference:- /opt/xilinx/common/xilinx-versal-common-v2023.2/sysroots/cortexa72-cortexa53-xilinx-linux #Sysroot path - /opt/xilinx/common/xilinx-versal-common-v2023.2/rootfs.ext4 #Root FS - /opt/xilinx/common/xilinx-versal-common-v2023.2/Image #Kernel Image
-
Select an
Empty application
template and clickFinish
. With steps 7 to 11, you have created the folder for the PS subsystem. Now it is time to import all the source files into the Vitis project (still empty at the moment). -
From Vitis GUI, right-click on the
dsplib_fir_AIE [aiengine]
folder, selectImport Sources
and take all the source files fromfiles/src/ip/aie/src
and put them intodsplib_fir_AIE/src
. Copy the entire subfolderfiles/src/ip/aie/src/graphs
intodsplib_fir_AIE/src/
. Repeat the step for all the data files fromfiles/src/ip/aie/data
todsplib_fir_AIE/data
. -
From Vitis GUI, right-click on the
dsplib_fir_PL
folder, selectImport Sources
and take all the source files from the two foldersfiles/src/ip/mm2s_aie/src
andfiles/src/ip/s2mm_aie/src
and put them intodsplib_fir_PL/src
. -
From Vitis GUI, right-click on the
dsplib_fir_PS [xrt]
folder, selectImport Sources
and take all the source files fromfiles/src/ps_apps/aie_test/src
and put them intodsplib_fir_PS/src
. -
From Vitis GUI, right-click on the
dsplib_fir_system_hw_link [pl]
, selectImport Sources
and take thefiles/src/vitis/src/system.cfg
file and put it intodsplib_fir_system_hw_link [pl]
itself. Now that all the source files have been imported, you should see something like in Figure 3.1-1. Now set up all the settings to finally build the entire project. -
Select the
dsplib_fir_AIE.prj
file and select theTop-Level File
file namedfir_graph.cpp
. -
From Vitis GUI, right-click on the
dsplib_fir_AIE [aiengine]
domain and selectC/C++ Build Settings
. Make sure to selectConfiguration [All Configurations]
. Add the following directories to the default ones inInput Spec
, then clickApply and Close
. You should see something like in Figure 3.2-2.
${env_var:XILINX_VITIS}/aietools/include
${env_var:DSPLIB_ROOT}/L1/src/aie
${env_var:DSPLIB_ROOT}/L1/include/aie
${env_var:DSPLIB_ROOT}/L2/include/aie
${env_var:XILINX_HLS}/include
${workspace_loc:/${ProjName}/src}
${workspace_loc:/${ProjName}/src/graphs}
${workspace_loc:/${ProjName}/data}
-
Click on
dsplib_fir_system_hw_link.prj
file, right-click onbinary_container_1
, selectEdit V++ Options settings
, and add--config=../system.cfg
as illustrated in Figure 3.2-3. -
The
system.cfg
file foraie_L2dsplib_system_hw_link [pl]
domain is set as in the following (see Figure 3.2-4):
[connectivity]
stream_connect=mm2s_1.s:ai_engine_0.DataIn1
stream_connect=ai_engine_0.DataOut1:s2mm_1.s
-
Click on the
dsplib_fir_PL.prj
file. Push theAdd the HW functions
button and select themm2s
ands2mm
functions, as illustrated in Figure 3.2.5. -
From Vitis GUI, right-click on the
dsplib_fir_PS [xrt]
domain and selectC/C++ Build Settings
. Make sure to selectConfiguration [All Configurations]
.
- As illustrated in Figure 3.2-6, for the GCC host compiler, the other dialect flag has to be set to
-std=c++17
Then click Apply and Close
.
- As illustrated in Figure 3.2-7, add the following library search paths (
-L
flag):
${env_var:XILINX_VITIS)/aietools/lib/aarch64.o
${SYSROOT}/usr/lib
- and libraries (
-l
flag):
xilinxopencl
xaiengine
adf_api_xrt
xrt_core
xrt_coreutil
- As illustrated in Figure 3.2.8, add the following directories to the default ones in
Include
, then clickApply and Close
.
${env_var:XILINX_VITIS}/aietools/include
${env_var:XILINX_VITIS}/include
${env_var:XILINX_HLS}/include
${env_var:DSPLIB_ROOT}/L1/src/aie
${env_var:DSPLIB_ROOT}/L1/include/aie
${env_var:DSPLIB_ROOT}/L2/include/aie
${workspace_loc:/dsplib_fir_AIE/src}
${workspace_loc:/dsplib_fir_AIE/src/graphs}
${SYSROOT}/usr/include/xrt
(Figure 3.2-1) Setting OS paths to compile the host software subsystem.
(Figure 3.2-2) Vitis GUI-based flow: C/C++ Build Settings for the AIE subsystem.
(Figure 3.2-3) Vitis GUI-based flow: assigning the system.cfg file to the binary container.
(Figure 3.2-4) Vitis GUI-based flow: system.cfg configuration file in details.
(Figure 3.2-5) Vitis GUI-based flow: project settings for the kernels of PL subsystem.
(Figure 3.2-6) Vitis GUI-based flow: settings the C++ dialect to compile the host software subsystem.
(Figure 3.2-7) Setting the libraries and related paths to compile the host software subsystem.
(Figure 3.2-8) Setting the include paths to compile the host software subsystem.
Alternatively, to the GUI-based flow, use Vitis from command line with the proper makefile by typing the following:
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files
# set your environment
#source files/scripts/<YOUR_OWN>_sample_env_setup.sh
# launch makefile-based flow
#source ./run_all.sh
cd make-flow
make all
Whatever flow you have selected based on either GUI or Makefile, once you have written the generated SD card, you can boot the VCK190 board.
At the prompt on UART terminal digit, insert the petalinux
username and choose a password to login, for example, root
.
You should see the following:
versal-rootfs-common-20231 login: petalinux
Password:
versal-rootfs-common-20231:~$ sudo su
Password:
versal-rootfs-common-20231:/home/petalinux# mount /dev/mmcblk0p1 /mnt
versal-rootfs-common-20231:/home/petalinux# cd /mnt
versal-rootfs-common-20231:/mnt# ls -l
total 33549
-rwxrwx--- 1 root disk 4002464 Jan 1 2015 BOOT.BIN
-rwxrwx--- 1 root disk 22268416 Jan 1 2015 Image
drwxrwx--- 2 root disk 512 Apr 14 2023 System Volume Information
-rwxrwx--- 1 root disk 6840118 Jan 1 2015 binary_container_1.xclbin
-rwxrwx--- 1 root disk 3015 Jan 1 2015 boot.scr
-rwxrwx--- 1 root disk 1238840 Jan 1 2015 dsp_fir.exe
versal-rootfs-common-20231:/mnt#
versal-rootfs-common-20231:/mnt#
You can now launch the following commands:
sudo su
mount /dev/mmcblk0p1 /mnt
cd /mnt
./dsplib_fir_PS binary_container_1.xclbin
You should see the following:
versal-rootfs-common-20231:/mnt# ./dsp_fir.exe ./binary_container_1.xclbin
. . .
INFO: iterations = 0
INFO: words SizeIn = 8192
INFO: words SizeOut = 4096
INFO: bytes SizeIn = 32768
INFO: bytes SizeOut = 16384
. . .
PASSED: my_graph.reset()
PASSED: my_graph.run(0)
INFO: Waiting for kernels to end...
PASSED: in_0_run.wait()
PASSED: out_0_run.wait()
PASSED: out_0_bo.sync(XCL_BO_SYNC_BO_FROM_DEVICE)
INFO: Veryfing output data vs. golden ones
PASSED: ./dsp_fir.exe
versal-rootfs-common-20231:/mnt#
The Versal Custom Thin Platform Extensible System tutorial has the following aspects to be considered:
- It is always well-maintained and promptly updated along the various Vitis releases.
- By default, it generates a
sd_card.img
file withpetalinux
file system on it and the SD card has only one partition (FAT32). - It applies a hierarchy of
Makefiles
that should enable an "easier" customization of the system to be designed, without touching or modifying the directories structure and the Makefile themselves too.
Few notes to be mentioned besides what is already reported in its related README.md file:
-
Do not use the
sd_card.img
file generated in thepackage_output_hw
. On the contrary, format the SD card with a Windows OS utility like SD Card Formatter (5.0.2 developed by Tuxera Inc) in FAT32 format to a size of more than 2GB (at least). Then you can directly copy the files from thepackage_output_hw/sd_card
folder. The reason is that with thesd_card.img
file, the VCK190 sometimes does not boot for this particular design. -
At the prompt on UART terminal digit, insert the
petalinux
username and choose a password (for example,root
).
As already stated in the previous subsection, the Thin Platform has several advantages, being based on makefile flow, including ease of use, and fast compilation time. In this separate section, you will design the same project of section 3 Decimation FIR Filter DSP Design, that is, a decimation-by-2 FIR filter taken from the Vitis DSP Library for AI Engine. This time, you will implement such a design into the Thin Platform makefile based template as an alternative to using the two flows described in subsection 3.1 (GUI-based flow) and 3.2 (Makefile-based flow).
The prepare_thin-plat_files.sh script called from within run_all.sh script copies the Thin Platform folder and renames it into thin-plat
. It then removes folders (ip/counter
, ip/mm2s_vadd_s
, ip/s2mm_vadd_s
, ip/subtractor
) that are not required for this FIR filter design. It then changes other files, as explained in the next section.
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files #current dir
# set your environment
#source files/scripts/<YOUR_OWN>_sample_env_setup.sh
# source run_all.#
source ./scripts/prepare_thin-plat-files.sh
The original Thin Platform has the following subfolders:
01-Versal_Custom_Thin_Platform_Extensible_System/
├── documentation
├── linux
├── ps_apps
│ └── linux
│ ├── aie_dly_test
│ ├── vadd_mm_cpp
│ ├── vadd_mm_ocl
│ └── vadd_s
├── vitis
│ ├── ip
│ │ ├── aie
│ │ ├── counter
│ │ ├── mm2s_vadd_s
│ │ ├── s2mm_vadd_s
│ │ ├── subtractor
│ │ ├── vadd_mm
│ │ ├── vadd_s
│ ├── src
│ └── xpfm_export
└── vivado
while this project requires the following folders:
thin-plat/
├── linux
├── ps_apps
│ └── linux
│ └── aie_test
├── vitis
│ ├── ip
│ │ ├── aie
│ │ ├── mm2s_aie
│ │ ├── s2mm_aie
│ ├── src
│ └── xpfm_export
└── vivado
therefore you should have to do manually the following changes:
- Add your AI Engine graph and kernels code into the
vitis/ip/aie
folder and adapt the relatedvitis/ip/aie/Makefile
to your case. - Add your HLS kernels code into the
vitis/ip
folder, one HLS kernel in each subfolder, for example,vitis/ip/mm2s_aie
andvitis/ip/s2mm_aie
. Then check or adapt the relatedvitis/ip/mm2s_aie/Makefile
andvitis/ip/s2mm_aie/Makefile
. - Add your PS code into the
ps_apps/linux
folder. For example,ps_apps/linux/aie_test
. Then check or adapt the relatedps_apps/linux/aie_test/Makefile
. - Adapt the
vitis/src/system.cfg
configuration file for the Vitis Linker according to your connectivity (in this specific case, it is the same as the original Thin Platform repository, so no changes at all). - Do not modify the
vitis/ip/Makefiles
because it orchestrates all the Makefiles of the various subfolders. The remaining ones are managed "automatically" by the other Makefiles of the Thin Platform project.
To provide ease, all above steps are already done by the script prepare_thin-plat_files.sh and the Makefiles are already changed and available in this repository.
Run the following commands:
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files #cur dir
# set your environment
#source files/scripts/<YOUR_OWN>_sample_env_setup.sh
#prepare the Thin Platform folder
source ./scripts/prepare_thin-plat_files.sh
Now you can build the SD card with the following commands:
cd <WRK_DIR> #your working directory
cd 02-AIE_DSP_with_Makefile_and_GUI/files
# set your environment
#source files/scripts/<YOUR_OWN>_sample_env_setup.sh
cd thin-plat
make all TARGET=hw
As already mentioned in the previous subsection, do not use the sd_card.img
file generated in the package_output_hw
directly. On the contrary, just format the SD card with a Windows OS utility like SD Card Formatter (5.0.2 developed by Tuxera Inc) in FAT32 to a size of more than 2GB (at least) as illustrated in Figure 4.2.1-1. Then you can directly copy the files from the package_output_hw/sd_card
folder, as illustrated in Figure 4.2.1-2.
(Figure 4.2.1-1) FAT32 formatting to prepare the sd card for VCK190.
(Figure 4.2.1-2) Copy the files on sd card to boot the VCK190.
At the prompt on UART terminal digit, insert the petalinux
username and choose a password to login. For example, root
. You should see the following:
versal-rootfs-common-20231 login: petalinux
You are required to change your password immediately (administrator enforced).
New password:
Retype new password:
versal-rootfs-common-20231:~$ sudo su
We trust you have received the usual lecture from the local System
Administrator. It usually boils down to these three things:
#1) Respect the privacy of others.
#2) Think before you type.
#3) With great power comes great responsibility.
Password:
You can now launch the following commands:
sudo su
mount /dev/mmcblk0p1 /mnt
cd /mnt
./xrt_host.exe ./*.xclbin
You should see something similar to this:
versal-rootfs-common-20231:/mnt# ./xrt_host.exe ./a.xclbin
INFO: iterations = 0
INFO: words SizeIn = 8192
INFO: words SizeOut = 4096
INFO: bytes SizeIn = 32768
INFO: bytes SizeOut = 16384
PASSED: auto my_device = xrt::device(0)
PASSED: auto xclbin_uuid = my_device.load_xclbin(./a.xclbin)
PASSED: auto in_0 = xrt::kernel(my_device, xclbin_uuid, "mm2s:{mm2s_1}")
. . .
PASSED: my_graph.reset()
PASSED: my_graph.run(0)
INFO: Waiting for kernels to end...
PASSED: in_0_run.wait()
PASSED: out_0_run.wait()
PASSED: out_0_bo.sync(XCL_BO_SYNC_BO_FROM_DEVICE)
INFO: Veryfing output data vs. golden ones
PASSED: ./xrt_host.exe
versal-rootfs-common-20231:/mnt#
Contains sections on how to develop AI Engine graphs, how to use the AI Engine compiler and AI Engine simulation, and performance analysis.
Below are the links to the XRT information used by this tutorial:
-
XRT Documentation: Explains general XRT API calls used in the PS Host Application.
-
XRT Github Repo: Contains the XRT source code.
-
XRT AI Engine API: Documents the AI Engine XRT API calls.
Below are links to Vitis related information referenced in this tutorial:
GitHub issues will be used for tracking requests and bugs. For questions, go to support.xilinx.com.
The MIT License (MIT)
Copyright © 2023 Advanced Micro Devices, Inc. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
© Copyright 2023 Advanced Micro Devices, Inc.