ContributionsMost RecentMost LikesSolutionsRe: oneAPI compile errors pac_a10 Hello, I have not gotten the Sample designs to compile on any version yet, but I am working with the 2022.1 versions of the sample designs. Looking at the supported OS matrix, I'll try a RHEL 8 and Ubuntu 20.04 installation. I was using a guide on using OneAPI and the Intel FPGAs on Centos 7 provided by an Intel FPGA specialist, but will try one of the supported operating systems. Thanks, Joe oneAPI compile errors pac_a10 Hello, I've been trying to get our fpga systems to a state where our researchers can use them. After a firmware upgrade to the latest and moving to using oneAPI on Centos 7, I am still getting errors when I try to compile to the fpga. I'm working with the fgpa_compile example DPC++FPGA GettingStarted Tutorial from oneAPI source github examples 2022.1.0 (https://github.com/oneapi-src/oneAPI-samples/archive/refs/tags/2022.1.0.zip ) I was able to make and run the fpga_emu target, but when running make for the fpgu target I get the following error: Error (23035): Tcl error: while executing "qexec "quartus_fit --read_settings_files=on --write_settings_files=off dcp -c afu_$flow"" (file "compile_script.tcl" line 24) Error (23031): Evaluation of Tcl script compile_script.tcl unsuccessful ...... Error (23035): Tcl error: while executing "qexec "bash build/run.sh $revision_name"" ("default" arm line 9) invoked from within "switch $tcl_platform(platform) { windows { post_message -type error "Full compiles to generate hardware for the FPGA are available on supported ..." (file "build/entry.tcl" line 36) Error (23031): Evaluation of Tcl script build/entry.tcl unsuccessful I'm using gcc 7.4 installed from source and it looks like the oneAPI provided python 3.9.10. I've also tried the gcc 8.3 tool chain with the same results. I've attached the commands run, quartuserr.tmp, and quartus_sh_compile.log files for reference. Thanks, Joe Re: unable to install intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1 from yum repositories It may be something with the way CentOS 7.9 sees the yum package. I am trying on Fedora 35 and it is 38% at 7.0 Gb and still running. I'll update when that is done. Re: unable to install intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1 from yum repositories I get the same results when using --downloadonly flag for yum or the yumdownloader. I have tried downloading the package from multiple systems over two completely different internet connections and have had exactly the same results listed below. Thanks, Joe # cat /etc/yum.repos.d/intel.repo [intel] name=Intel® oneAPI repository baseurl=https://yum.repos.intel.com/oneapi enabled=1 gpgcheck=1 repo_gpgcheck=1 gpgkey=https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB # yumdownloader intel-oneapi-intelfpgadpcpp-a10gx-s10sx.x86_64 Loaded plugins: fastestmirror Loading mirror speeds from cached hostfile * base: mirrors.gigenet.com * epel: download-ib01.fedoraproject.org * extras: mirror.web-ster.com * updates: mirror.web-ster.com intel-oneapi-intelfpgadpcpp-a1 FAILED 109% [==========================================================================================] 1.6 MB/s | 2.1 GB --:--:-- ETA https://yum.repos.intel.com/oneapi/intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1.x86_64.rpm: [Errno 14] curl#63 - "Callback aborted"===========================================================] 1.6 MB/s | 2.1 GB --:--:-- ETA Trying other mirror. intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1.x86_64: [Errno 256] No more mirrors to try. unable to install intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1 from yum repositories I've been trying to install the oneAPI for Intel FPGA add-ons through the yum repository, but have been unsuccessful. It shows 1.9Gb in the yum info, but it exceeds that by .2Gb/9% and then errors out. I've posted the output below. I was unable to access the direct download zip file due to permission errors or limitations on the Intel site. Thanks, Joe $ sudo yum install intel-oneapi-intelfpgadpcpp-a10gx-s10sx Loaded plugins: fastestmirror, nvidia Loading mirror speeds from cached hostfile Resolving Dependencies --> Running transaction check ---> Package intel-oneapi-intelfpgadpcpp-a10gx-s10sx.x86_64 0:2022.1.0-1 will be installed --> Finished Dependency Resolution Dependencies Resolved ========================================================================= Package Arch Version Repository Size ========================================================================= Installing: intel-oneapi-intelfpgadpcpp-a10gx-s10sx x86_64 2022.1.0-1 oneAPI 1.9 G Transaction Summary ========================================================================= Install 1 Package Total download size: 1.9 G Installed size: 0 Is this ok [y/d/N]: y Downloading packages: intel-oneapi-intelfpgadpcpp-a1 FAILED 109% [======================================================================] 23 MB/s | 2.1 GB --:--:-- ETA https://yum.repos.intel.com/oneapi/intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1.x86_64.rpm: [Errno 14] curl#63 - "Callback aborted"===================] 23 MB/s | 2.1 GB --:--:-- ETA Trying other mirror. Error downloading packages: intel-oneapi-intelfpgadpcpp-a10gx-s10sx-2022.1.0-1.x86_64: [Errno 256] No more mirrors to try. Re: Developer Zone Premier but OneAPI FPGA add-on can't be download I am having this same problem with getting the intel-oneapi-intelfpgadpcpp-a10gx-s10sx package. I have also tried to install through yum, but it fails after downloading 109% (2.1Gb of the expected 1.9Gb log attached). I was able to install the intel-basekit using yum on the system. Thanks, Joe Ryan Re: permission error accessing fpga aria 10 We were working on using the OPAE stack and were able to finally get an aocl diagnose output. We have determined that FIM/BMC Firmware upgrades are needed to move to the latest OPAE/IntelDevStack. As a related question, the archive page for version 1.1 (https://www.intel.com/content/www/us/en/software-kit/665876/intel-pac-with-intel-arria-10-gx-fpga-acceleration-stack-version-1-1.html?) links to FIM/BMC Firmware tools download with (https://www.bittware.com/fpga/intel/boards/pac-arria-10/). However, this returns a 404 error from Bittware. I am in the process of registering there, but is that the appropriate place to obtain the different firmware versions to upgrade through the versions to 1.2.1? Joe from https://www.intel.com/content/www/us/en/software-kit/665876/intel-pac-with-intel-arria-10-gx-fpga-acceleration-stack-version-1-1.html? BMC Firmware and Tools Download: Register at Intel ® PAC Firmware and Tools and select Intel ® PAC permission error accessing fpga aria 10 I am getting errors running fpgabist when logged in as an unprivileged user, but it works when using the root account. Additionally, I see permission denied errors in an strace when attempting to use aocl for any purpose. The permissions look correct based on the changes made by AOCL_BOARD_PACKAGE_ROOT/linux64/libexec/setup_permissions.sh. I am unable to run aocl commands under root user until I determine all the necessary environment variables/path updates needed to successfully run them. This is installed in a shared environment, so unprivileged users would need access to use the cards. $ ll /dev/intel-fpga* crw-rw-rw- 1 root root 242, 0 Sep 14 12:51 /dev/intel-fpga-fme.0 crw-rw-rw- 1 root root 242, 1 Sep 14 12:51 /dev/intel-fpga-fme.1 crw-rw-rw- 1 root root 242, 2 Sep 14 12:51 /dev/intel-fpga-fme.2 crw-rw-rw- 1 root root 241, 0 Sep 14 12:51 /dev/intel-fpga-port.0 crw-rw-rw- 1 root root 241, 1 Sep 14 12:51 /dev/intel-fpga-port.1 crw-rw-rw- 1 root root 241, 2 Sep 14 12:51 /dev/intel-fpga-port.2 $ fpgabist /opt/intel/inteldevstack/hw/samples/nlb_mode_3/bin/nlb_mode_3.gbs -B 3b Beginning FPGA Built-In Self-Test ========================================================== Device: bus = 3b, device = 00, func = 0 //****** FME ******// Object Id : 0xF200000 PCIe s:b:d:f : 0000:3B:00:0 Device Id : 0x09C4 Socket Id : 0x00 Ports Num : 01 Bitstream Id : 0x123456789ABCDEF Bitstream Version : 1.2.3 Pr Interface Id : ce489693-98f0-5f33-946d-560708be108a //****** PORT ******// Object Id : 0xF100000 PCIe s:b:d:f : 0000:3B:00:0 Device Id : 0x09C4 Socket Id : 0x00 Ports Num : 01 Bitstream Id : 0x123456789ABCDEF Bitstream Version : 1.2.3 Pr Interface Id : ce489693-98f0-5f33-946d-560708be108a Accelerator Id : f7df405c-bd7a-cf72-22f1-44b0b93acd18 Accelerator GUID : f7df405c-bd7a-cf72-22f1-44b0b93acd18 //****** TEMP ******// Object Id : 0xF200000 PCIe s:b:d:f : 0000:3B:00:0 Device Id : 0x09C4 Socket Id : 0x00 Ports Num : 01 Bitstream Id : 0x123456789ABCDEF Bitstream Version : 1.2.3 Pr Interface Id : ce489693-98f0-5f33-946d-560708be108a Package Temperature : 38 Centigrade metrics_utils.c:633:enum_bmc_metrics_info() **ERROR** : Failed to load BMC SDR. metrics.c:80:xfpga_fpgaGetNumMetrics() **ERROR** : Failed to Discover Metrics Error getting number of metrics: not found Error reading metrics from BMC: not found //****** POWER ******// .... Writing bitstream Done Running fpgadiag read test... sysfs.c:1481:sync_object() **ERROR** : Error opening /sys/class/fpga/intel-fpga-dev.0/intel-fpga-fme.0/dperf/fabric/freeze: Permission denied terminate called after throwing an instance of 'opae::fpga::types::exception' what(): failed with return code FPGA_EXCEPTION at: sysobject.cpp:get():85 Failed Test: read Command 'fpgadiag -B 0x3b --mode=read --read-vc=vh0 --write-vc=vh0 --multi-cl=4 --begin=1024, --end=1024 --timeout-sec=5 --cont' returned non-zero exit status 250 Running fpgadiag write test... .... $ sudo fpgabist /opt/intel/inteldevstack/hw/samples/nlb_mode_3/bin/nlb_mode_3.gbs -B 3b .... Done Running fpgadiag read test... Cachelines Read_Count Write_Count Cache_Rd_Hit Cache_Wr_Hit Cache_Rd_Miss Cache_Wr_Miss Eviction 'Clocks(@400 MHz)' Rd_Bandwidth Wr_Bandwidth 1024 540436620 0 0 0 0 0 0 2000964806 6.914 GB/s 0.000 GB/s VH0_Rd_Count VH0_Wr_Count VH1_Rd_Count VH1_Wr_Count VL0_Rd_Count VL0_Wr_Count 540436620 1 0 0 0 0 Running fpgadiag write test... Cachelines Read_Count Write_Count Cache_Rd_Hit Cache_Wr_Hit Cache_Rd_Miss Cache_Wr_Miss Eviction 'Clocks(@400 MHz)' Rd_Bandwidth Wr_Bandwidth 1024 0 543165164 0 0 0 0 0 2000611414 0.000 GB/s 6.950 GB/s VH0_Rd_Count VH0_Wr_Count VH1_Rd_Count VH1_Wr_Count VL0_Rd_Count VL0_Wr_Count 0 543165165 0 0 0 0 Running fpgadiag trput test... Cachelines Read_Count Write_Count Cache_Rd_Hit Cache_Wr_Hit Cache_Rd_Miss Cache_Wr_Miss Eviction 'Clocks(@400 MHz)' Rd_Bandwidth Wr_Bandwidth 1024 487225740 488951172 0 0 0 0 0 2001626166 6.231 GB/s 6.253 GB/s VH0_Rd_Count VH0_Wr_Count VH1_Rd_Count VH1_Wr_Count VL0_Rd_Count VL0_Wr_Count 487225740 488951173 0 0 0 0 Finished Executing NLB (FPGA DIAG)Tests Built-in Self-Test Completed. $ aocl diagnose aocl diagnose: Running diagnose from /opt/intel/inteldevstack/opencl/opencl_bsp/linux64/libexec aocl diagnose: failed 3 times. First error below: Vendor: Intel Corp handle is: -2 Phys Dev Name Status Information pac_a10_f100002Uninitialized Not configured with OpenCL BSP. Found no active device installed on the host machine. Please consult documentation for troubleshooting steps DIAGNOSTIC_FAILED