hello world launch hang - intel a10 FPGA

ramgowda · ‎02-02-2019

Hello,

I am trying to setup, compile run an hello world example. Below are the instructions I have followed, until launch kernel which just hangs there.

Please let me know if the setup instructions looks right any pointers to debug the kernel hang.

SETUP

-----

source /local/ramachandra.utah/inteldevstack/init_env.sh

export INTELFPGAOCLSDKROOT=/local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld/

export ALTERAOCLSDKROOT=/local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld/

source /local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld/init_opencl.sh

source /local/ramachandra.utah/inteldevstack/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/linux64/libexec/setup_permissions.sh

export AOCL_BOARD_PACKAGE_ROOT=$INTELFPGAOCLSDKROOT/board/a10_ref

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld/board/a10_ref/linux64/lib

DRIVER install

------------

aocl install

Do you want to install /local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld//board/a10_ref? [y/n] y

LD [M] /tmp/opencl_driver_f2SkG2/aclpci_a10_ref_drv.ko

....

[root@fpga002 ~]# lsmod | grep a10

aclpci_a10_ref_drv 41362 0

[root@fpga002 ~]# lsmod | grep fpga

intel_fpga_fme 52380 0

intel_fpga_afu 31735 0

fpga_mgr_mod 14693 1 intel_fpga_fme

intel_fpga_pci 26519 2 intel_fpga_afu,intel_fpga_fme

OPENCL

-------

aocl diagnose

aocl program acl0 ./inteldevstack/a10_gx_pac_ias_1_1_pv/opencl/vector_add.aocx

aocl program acl1 ./inteldevstack/a10_gx_pac_ias_1_1_pv/opencl/vector_add.aocx

aocl diagnose // success

[root@fpga001 hello_world]# aoc -list-boards

Board list:

a10gx

Board Package: /local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld/board/a10_ref

a10gx_hostch

Board Package: /local/ramachandra.utah/inteldevstack/intelFPGA_pro/hld/board/a10_ref

Channels: host_to_dev, dev_to_host

pac_a10

Board Package: /local/ramachandra.utah/inteldevstack/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp

EMULATOR MODE (runs fine)

--------------------

aoc -march=emulator -v -board=pac_a10 device/hello_world.cl -o bin/hello_world.aocx

module load intel/compiler/2019

make // compile host

env CL_CONTEXT_EMULATOR_DEVICE_INTELFPGA=1 bin/host // runs fine

BOARD COMPILATION

-------------------------

aoc -v -board=pac_a10 device/hello_world.cl -o bin/hello_world.aocx

BOARD RUN (hang)

--------------

[root@fpga001 hello_world]# ./bin/host

Querying platform for info:

==========================

CL_PLATFORM_NAME = Intel(R) FPGA SDK for OpenCL(TM)

CL_PLATFORM_VENDOR = Intel(R) Corporation

CL_PLATFORM_VERSION = OpenCL 1.0 Intel(R) FPGA SDK for OpenCL(TM), Version 17.1.1

Querying device for info:

========================

CL_DEVICE_NAME = pac_a10 : PAC Arria 10 Platform (pac_a10_eb00001)

CL_DEVICE_VENDOR = Intel Corp

CL_DEVICE_VENDOR_ID = 4466

CL_DEVICE_VERSION = OpenCL 1.0 Intel(R) FPGA SDK for OpenCL(TM), Version 17.1.1

CL_DRIVER_VERSION = 17.1

CL_DEVICE_ADDRESS_BITS = 64

CL_DEVICE_AVAILABLE = true

CL_DEVICE_ENDIAN_LITTLE = true

CL_DEVICE_GLOBAL_MEM_CACHE_SIZE = 32768

CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE = 0

CL_DEVICE_GLOBAL_MEM_SIZE = 8589934592

CL_DEVICE_IMAGE_SUPPORT = true

CL_DEVICE_LOCAL_MEM_SIZE = 16384

CL_DEVICE_MAX_CLOCK_FREQUENCY = 1000

CL_DEVICE_MAX_COMPUTE_UNITS = 1

CL_DEVICE_MAX_CONSTANT_ARGS = 8

CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE = 2147483648

CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS = 3

CL_DEVICE_MEM_BASE_ADDR_ALIGN = 8192

CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE = 1024

CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR = 4

CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT = 2

CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT = 1

CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG = 1

CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT = 1

CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE = 0

Command queue out of order? = false

Command queue profiling enabled? = true

Using AOCX: hello_world.aocx

Reprogramming device [0] with handle 1

Kernel initialization is complete.

Launching the kernel...

MuhammadAr_U_Intel · ‎02-07-2019

Hi, Can I know what version of OS are you using ? Looking at aocl diagnose, it seems you are using multiple devices. 1. Can you try aocl diagnose acl0 or acl1 and check if you see and READ/ WRITE transactions. 2. I notice you are missing the aocl flash command, please follow steps as mentioned in https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/hb/opencl-sdk/aocl_getting_started.pdf Page 36, Programming the Flash Memory of an FPGA Let me know if this is helpful to you. Thanks, Arslan

ramgowda · ‎02-09-2019

Hi Arslan,

Thanks a lot for the response. I re did the setup referring the document from scratch.

But facing problems and errors when I do aocl flash, although the diagnose is successful.

Please review the below detailed instructions and let me know if I have did something wrong.

Check the interface ID

---------------------

[root@fpga001 inteldevstack_1_1]# sudo fpgainfo fme

//****** FME ******//

Class Path : /sys/class/fpga/intel-fpga-dev.0/intel-fpga-fme.0

Device Path : /sys/devices/pci0000:85/0000:85:02.0/0000:86:00.0/fpga/intel-fpga-dev.0/intel-fpga-fme.0

Bus : 0x86

Device : 0x00

Function : 0x00

Device Id : 0x09C4

Fim Version : 1.1.3

Ports Num : 1

Socket Id : 0

Bitstream Id : 0x113000200000177

Bitstream Metadata : 0x1805304

Pr Interface Id : 9926ab6d-6c92-5a68-aabc-a7d84c545738

Object Id : 247463936

//****** FME ******//

Class Path : /sys/class/fpga/intel-fpga-dev.1/intel-fpga-fme.1

Device Path : /sys/devices/pci0000:d7/0000:d7:00.0/0000:d8:00.0/fpga/intel-fpga-dev.1/intel-fpga-fme.1

Bus : 0xD8

Device : 0x00

Function : 0x00

Device Id : 0x09C4

Fim Version : 1.1.3

Ports Num : 1

Socket Id : 0

Bitstream Id : 0x113000200000177

Bitstream Metadata : 0x1805304

Pr Interface Id : 9926ab6d-6c92-5a68-aabc-a7d84c545738

Object Id : 247463937

Download and INSTALL the version for the above interface Id.

----------------------------------------------------------

http://download.altera.com/akdlm/software/dcp/1.1/a10_gx_pac_ias_1_1_pv_dev_installer.tar.gz

./setup and install into /home/ramachandra.utah/inteldevstack_1_1/

Confirm if driver is loaded

--------------------------

[root@fpga001 inteldevstack_1_1]# sudo lsmod | grep fpga

intel_fpga_pac_hssi 18107 0

intel_fpga_fme 52380 0

intel_fpga_afu 31735 0

intel_fpga_pci 26519 2 intel_fpga_afu,intel_fpga_fme

fpga_mgr_mod 14693 1 intel_fpga_fme

Setup environment

-----------------

[root@fpga001 inteldevstack_1_1]# source /home/ramachandra.utah/inteldevstack_1_1/init_env.sh

export QUARTUS_HOME=/home/ramachandra.utah/inteldevstack_1_1/intelFPGA_pro/quartus

Adding $QUARTUS_HOME/bin to PATH

export OPAE_PLATFORM_ROOT=/home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv

Adding $OPAE_PLATFORM_ROOT/bin to PATH

[root@fpga001 inteldevstack_1_1]# source /home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/linux64/libexec/setup_permissions.sh

This script handles device permissions and huge page table setup.

Please refer to DCP Quick Start User Guide for OPAE driver and SW installation

instructions.

/etc/security/limits.d/99-opae_memlock.conf is already setup.

setup huge pages. must be done after every reboot

sudo bash -c "echo 20 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages"

setup permissions for device. must be done after every reboot.

sudo chmod 666 /dev/intel-fpga-port.*

sudo chmod 666 /sys/class/fpga/intel-fpga-dev.*/intel-fpga-port.*/userclk_freqcmd

sudo chmod 666 /sys/class/fpga/intel-fpga-dev.*/intel-fpga-port.*/userclk_freqcntrcmd

sudo chmod 666 /sys/class/fpga/intel-fpga-dev.*/intel-fpga-port.*/errors/clear

sudo chmod 666 /dev/intel-fpga-fme.*

export AOCL_BOARD_PACKAGE_ROOT=/home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/

source /home/ramachandra.utah/inteldevstack_1_1/intelFPGA_pro/hld/init_opencl.sh

INTELFPGAOCLSDKROOT is set to /home/ramachandra.utah/intelFPGA_pro/18.1/hld. Using that.

Found a Quartus directory at /home/ramachandra.utah/intelFPGA_pro/18.1/quartus. Using that.

Adding /home/ramachandra.utah/intelFPGA_pro/18.1/hld/bin to PATH

Adding /home/ramachandra.utah/intelFPGA_pro/18.1/hld/host/linux64/lib to LD_LIBRARY_PATH

Adding /home/ramachandra.utah/intelFPGA_pro/18.1/hld/board/a10_ref/linux64/lib to LD_LIBRARY_PATH

Aocl install, diagnose and flash

-------------------------------

[root@fpga001 inteldevstack_1_1]# aocl install

Do you want to install /home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp? [y/n] y

aocl install: Adding the board package /home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp to the list of installed pckages

aocl install: Setting up FCD

aocl install: Running install from /home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/linux64/libexec

WARNING: install not implemented. Please refer to DCP Quick Start User Guide

for installation instructions.

[root@fpga001 inteldevstack_1_1]# aocl diagnose

/home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/linux64/libexec/diagnose: error while loading shared libraries: libintel_opae_mmd.so: cannot open shared object file: No such file or directory

[root@fpga001 inteldevstack_1_1]# export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/linux64/lib/

[root@fpga001 inteldevstack_1_1]# aocl diagnose

--------------------------------------------------------------------

Device Name:

acl0

BSP Install Location:

/home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp

Vendor: Intel Corp

Physical Dev Name Status Information

pac_a10_eb00001 Passed PAC Arria 10 Platform (pac_a10_eb00001)

PCIe 216:00.0

FPGA temperature = 44 degrees C.

DIAGNOSTIC_PASSED

--------------------------------------------------------------------

Device Name:

acl1

BSP Install Location:

/home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp

Vendor: Intel Corp

Physical Dev Name Status Information

pac_a10_eb00000 Passed PAC Arria 10 Platform (pac_a10_eb00000)

PCIe 134:00.0

FPGA temperature = 44 degrees C.

DIAGNOSTIC_PASSED

--------------------------------------------------------------------

aocl flash acl0 /home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/hello_world.aocx

aocl flash: Running flash from /home/ramachandra.utah/inteldevstack_1_1/a10_gx_pac_ias_1_1_pv/opencl/opencl_bsp/linux64/libexec

ERROR: flash not implemented. Please refer to DCP Quick Start User Guide for

installation instructions.

aocl flash: Program failed.

MuhammadAr_U_Intel · ‎06-07-2019

Hi, Are you still facing issue or you are able to proceed ? Please try latest Acceleration stack version 1.2 and follow the instructions in quick starter guide first.. https://www.intel.com/content/www/us/en/programmable/documentation/iyu1522005567196.html Thanks, Arslan