embeded/odroid2025. 9. 22. 22:36

odroid c2 / 다이소 axxen sd card

다이소에서 싼 맛에 산

5천원 / 32GB / Axxen sd 카드가

odroid c2 에서 armbian 으로 쓰는데 허구한날 먼가 하다가 날아간다.

2개 가서 두개가 번갈아 가면서 몇번 발생하거면

지지리도 운이 없거냐

axxen이 내구도가 안 좋거나

armbian이 문제거나..

 

에러메시지는 맨날 btrfs 에서 멈춰서 혹시나 해서 봤는데 ext 인디..

$ mount
sysfs on /sys type sysfs (rw,nosuid,nodev,noexec,relatime)
proc on /proc type proc (rw,nosuid,nodev,noexec,relatime)
udev on /dev type devtmpfs (rw,nosuid,relatime,size=835580k,nr_inodes=208895,mode=755)
devpts on /dev/pts type devpts (rw,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000)
tmpfs on /run type tmpfs (rw,nosuid,nodev,noexec,relatime,size=196088k,mode=755)
/dev/mmcblk0p1 on / type ext4 (rw,relatime,errors=remount-ro,commit=120)
securityfs on /sys/kernel/security type securityfs (rw,nosuid,nodev,noexec,relatime)
tmpfs on /dev/shm type tmpfs (rw,nosuid,nodev)
tmpfs on /run/lock type tmpfs (rw,nosuid,nodev,noexec,relatime,size=5120k)
cgroup2 on /sys/fs/cgroup type cgroup2 (rw,nosuid,nodev,noexec,relatime,nsdelegate,memory_recursiveprot)
pstore on /sys/fs/pstore type pstore (rw,nosuid,nodev,noexec,relatime)
bpf on /sys/fs/bpf type bpf (rw,nosuid,nodev,noexec,relatime,mode=700)
systemd-1 on /proc/sys/fs/binfmt_misc type autofs (rw,relatime,fd=32,pgrp=1,timeout=0,minproto=5,maxproto=5,direct,pipe_ino=3927)
hugetlbfs on /dev/hugepages type hugetlbfs (rw,nosuid,nodev,relatime,pagesize=2M)
debugfs on /sys/kernel/debug type debugfs (rw,nosuid,nodev,noexec,relatime)
mqueue on /dev/mqueue type mqueue (rw,nosuid,nodev,noexec,relatime)
configfs on /sys/kernel/config type configfs (rw,nosuid,nodev,noexec,relatime)
fusectl on /sys/fs/fuse/connections type fusectl (rw,nosuid,nodev,noexec,relatime)
tmpfs on /tmp type tmpfs (rw,nosuid,relatime)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw,nosuid,nodev,noexec,relatime)
/dev/mmcblk0p1 on /var/log.hdd type ext4 (rw,relatime,errors=remount-ro,commit=120)
/dev/zram1 on /var/log type ext4 (rw,nosuid,nodev,noexec,relatime,discard)
sunrpc on /run/rpc_pipefs type rpc_pipefs (rw,relatime)
tmpfs on /run/user/1000 type tmpfs (rw,nosuid,nodev,relatime,size=196084k,nr_inodes=49021,mode=700,uid=1000,gid=1000)
portal on /run/user/1000/doc type fuse.portal (rw,nosuid,nodev,relatime,user_id=1000,group_id=1000)

 

commit=120은 2분에 한번 강제 저장인데

그렇다고 해서 무조건 또 저장이 되는건 아니라고

그런데. relatime 해놓고 commit 하면 sd 카드 수명 팍팍 갉아 먹는거 아닌가?

저작자표시 (새창열림)

'embeded > odroid' 카테고리의 다른 글

odroid c2 armbian / hdmi audio out 실패  (0) 2025.09.22
iperf3 설정별 속도  (0) 2025.08.21
빌드하다가 시스템이 날아감 -_-!  (0) 2025.08.13
odroid c2 opencv haar cascade / face_recognition  (0) 2025.08.12
armbian X11-forwarding 활성화 하기  (0) 2025.08.11
Posted by 구차니
embeded/odroid2025. 9. 22. 22:10

odroid c2 armbian / hdmi audio out 실패

aplay 로 목록을 보면나온다는데 안나오고

$ aplay -L 
null
    Discard all samples (playback) or generate zero samples (capture)
default
    Playback/recording through the PulseAudio sound server
lavrate
    Rate Converter Plugin Using Libav/FFmpeg Library
samplerate
    Rate Converter Plugin Using Samplerate Library
speexrate
    Rate Converter Plugin Using Speex Resampler
jack
    JACK Audio Connection Kit
oss
    Open Sound System
pulse
    PulseAudio Sound Server
speex
    Plugin using Speex DSP (resample, agc, denoise, echo, dereverb)
upmix
    Plugin for channel upmix (4,6,8)
vdownmix
    Plugin for channel downmix (stereo) with a simple spacialization
hw:CARD=ODROIDC2,DEV=0
    ODROID-C2, 
    Direct hardware device without any conversions
plughw:CARD=ODROIDC2,DEV=0
    ODROID-C2, 
    Hardware device with all software conversions
sysdefault:CARD=ODROIDC2
    ODROID-C2, 
    Default Audio Device
dmix:CARD=ODROIDC2,DEV=0
    ODROID-C2, 
    Direct sample mixing device
usbstream:CARD=ODROIDC2
    ODROID-C2
    USB Stream Output

[링크 : https://forums.libretro.com/t/odroid-c2-hdmi-audio-output-has-no-sound/10559]

 

커널 메시지 상으로도 이렇다할 무언가가 보이지 않는 느낌..

$ sudo dmesg | grep -i hdmi
[    0.069016] /soc/vpu@d0100000: Fixed dependency cycle(s) with /soc/hdmi-tx@c883a000
[    0.069037] /soc/hdmi-tx@c883a000: Fixed dependency cycle(s) with /soc/vpu@d0100000
[    0.073417] /soc/vpu@d0100000: Fixed dependency cycle(s) with /soc/hdmi-tx@c883a000
[    0.073562] /soc/vpu@d0100000: Fixed dependency cycle(s) with /soc/hdmi-tx@c883a000
[    0.073616] /soc/hdmi-tx@c883a000: Fixed dependency cycle(s) with /soc/vpu@d0100000
[    0.075830] /soc/hdmi-tx@c883a000: Fixed dependency cycle(s) with /hdmi-connector
[    0.075893] /hdmi-connector: Fixed dependency cycle(s) with /soc/hdmi-tx@c883a000
[    1.682512] meson-dw-hdmi c883a000.hdmi-tx: Detected HDMI TX controller v2.01a with HDCP (meson_dw_hdmi_phy)
[    1.683044] meson-dw-hdmi c883a000.hdmi-tx: registered DesignWare HDMI I2C bus driver
[    1.683310] meson-drm d0100000.vpu: bound c883a000.hdmi-tx (ops 0xffff800081233b38)

 

정식 배포된 마지막 버전인 ubuntu mate 20.04로 한번 해보고 나오나봐야겠다.

저작자표시 (새창열림)

'embeded > odroid' 카테고리의 다른 글

odroid c2 / 다이소 axxen sd card  (0) 2025.09.22
iperf3 설정별 속도  (0) 2025.08.21
빌드하다가 시스템이 날아감 -_-!  (0) 2025.08.13
odroid c2 opencv haar cascade / face_recognition  (0) 2025.08.12
armbian X11-forwarding 활성화 하기  (0) 2025.08.11
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 16. 12:11

eiq 모델 정리

오디오와 비디오로 크게 나뉘고

비디오에서는 classification / obejct detection / segmentation 정도가 현재 관심사

sementic segmentation과 instance-segmentation 차이는 멀까?

List of domains
Audio
anomaly detection
command recognition
speech recognition
Vision
classification
face recognition
object detection
pose estimation
semantic segmentation
super resolution
instance-segmentation
low-light enhancement

[링크 : https://github.com/NXP/eiq-model-zoo]

 

텍스트로 넣으니 css 때문에 깨져서 이미지로 복사 -_-

selfie-segmenter는 proprietary dataset이었군..

[링크 : https://github.com/NXP/eiq-model-zoo/tree/main/products]

저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

eiq 데이터 구조  (0) 2025.09.05
ubuntu 22.04 + cuda + cudnn 설치  (0) 2025.09.04
import tensorflow illegal instruction  (0) 2025.09.04
eiq on windows with nvidia  (0) 2025.09.03
vainfo  (0) 2025.09.03
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 5. 10:39

eiq 데이터 구조

크게 세 가지가 가능한데

VOC dataset은 detection을 위한 것이고

structured folder는 classification

tfds는 classification과 detection 모두를 위한 것인데...

 

3.3.1 VOC dataset
VOC datasets are introduced in Pascal Visual Object Classes Challenge. This format is also used by the ImageNet
datasets. Images and labels in this dataset can be used to train detection models.

 

3.3.2 Structured folders dataset
Using the Import dataset screen, you can also import datasets that have a specific folder structure. These datasets do
not have to be compressed into TAR files as with VOC datasets. The only requirement is that the images are saved in
specific folders. The datasets created this way can be used to train models for classification tasks.

 

3.3.3 TFDS datasets
TFDS datasets allow you to import some of the datasets provided by tensorflow in their catalog here. The tool allows
you to download datasets for image classification and object detection.

 

그러면 segmentation은 누구냐..

문서 상에서는 cli를 통해서 직접해야 하는건가..

7.1 Image segmentation
The following models target the image segmentation use case:
• deeplab_v3
To run the “deeplab” model, the workspace\models\deeplab_v3\runner_demo.py script runs the inference using the
converted model on the target device and decodes the output. You may also inspect the Python code. The script sends
a simple HTTP request to the device and the rest of the script decodes the output. There are also a few images available
for testing in the imgs folder.
저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

eiq 모델 정리  (0) 2025.09.16
ubuntu 22.04 + cuda + cudnn 설치  (0) 2025.09.04
import tensorflow illegal instruction  (0) 2025.09.04
eiq on windows with nvidia  (0) 2025.09.03
vainfo  (0) 2025.09.03
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 4. 12:36

ubuntu 22.04 + cuda + cudnn 설치

 

압도적 감사!

그런데 비교해보니 환경변수가 필수였나.. 싶네

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

[링크 : https://starlane.tistory.com/1]

[링크 : https://bo-10000.tistory.com/209]

 

 

+

하드웨어

i7-10700F / 3070 8GB

 

명령어. 나중에 nouveau 쪽 수행하지 말고 바로 nvidia-driver 설치해봐야 겠다. 그러면 제법 시간이 단축될 듯

sudo apt install build-essential gcc ubuntu-drivers-common dkms vim nvidia-modprobe
sudo apt-get update; sudo apt-get upgrade -y
sudo bash -c "echo blacklist nouveau > /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
sudo bash -c "echo options nouveau modeset=0 >> /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
cat /etc/modprobe.d/blacklist-nvidia-nouveau.conf 
sudo update-initramfs -u

# after reboot
lsmod | grep nouveau
ubuntu-drivers devices
sudo apt install nvidia-driver-575

# add environment variables
echo export PATH=/usr/local/cuda/bin:$PATH > ~/.bashrc
echo export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH > ~/.bashrc

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-ubuntu2204.pin
sudo mv cuda-ubuntu2204.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/13.0.0/local_installers/cuda-repo-ubuntu2204-13-0-local_13.0.0-580.65.06-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu2204-13-0-local_13.0.0-580.65.06-1_amd64.deb
sudo cp /var/cuda-repo-ubuntu2204-13-0-local/cuda-BE1CC667-keyring.gpg /usr/share/keyrings/

sudo apt-get update
sudo apt-get -y install cuda-toolkit-13-0
sudo apt-get install -y cuda-drivers

# for test
nvcc -V


wget wget https://developer.download.nvidia.com/compute/cudnn/9.12.0/local_installers/cudnn-local-repo-ubuntu2204-9.12.0_1.0-1_amd64.deb
sudo dpkg -i cudnn-local-repo-ubuntu2204-9.12.0_1.0-1_amd64.deb
sudo cp /var/cudnn-local-repo-ubuntu2204-9.12.0/cudnn-*-keyring.gpg /usr/share/keyrings/

 

eIQ 설치를 위한 준비

sudo apt-get install libffi7

 

저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

eiq 모델 정리  (0) 2025.09.16
eiq 데이터 구조  (0) 2025.09.05
import tensorflow illegal instruction  (0) 2025.09.04
eiq on windows with nvidia  (0) 2025.09.03
vainfo  (0) 2025.09.03
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 4. 10:32

import tensorflow illegal instruction

신형을 쓰면 모를수도 있지만..

아무생각 없이 구형에 펜티엄 cpu를 쓰다보니 이 사단이 나버리는 구먼.. -_-

조사를 해보니 숫자가 비슷하긴 한데.. 세대가 많이 차이가 났구먼..

 

G3460 / 하스웰 펜티엄 / SSE4.2

[링크 : https://www.intel.co.kr/content/www/kr/ko/products/sku/83428/intel-pentium-processor-g3460-3m-cache-3-50-ghz/specifications.html]

 

G4400T / 스카이레이크 펜티엄 / SSE4.2

[링크 : https://www.intel.co.kr/content/www/kr/ko/products/sku/90614/intel-pentium-processor-g4400t-3m-cache-2-90-ghz/specifications.html]

 

치사(?)하게 셀러론과 펜티엄은 뺐고, 타이거레이크(11세대) 이후 부터 펜티엄과 셀러론에도 넣어주었다고

CPUs with AVX
Intel
Sandy Bridge processors (Q1 2011) and newer, except models branded as Celeron and Pentium.[9]
Pentium and Celeron branded processors starting with Tiger Lake (Q3 2020) and newer.[10]
AMD:
Bulldozer processors (Q4 2011) and newer.[11]
Issues regarding compatibility between future Intel and AMD processors are discussed under XOP instruction set.

VIA:
Nano QuadCore
Eden X4
Zhaoxin:
WuDaoKou-based processors (KX-5000 and KH-20000)

[링크 : https://en.wikipedia.org/wiki/Advanced_Vector_Extensions]

 

텐서플로우 릴리즈 노트 찾아보니 1.5.0 rc 부터 AVX 넣어놔서 안될수도 있다고 한다.

TensorFlow 1.5.1
Release 1.5.1
Bug Fixes and Other Changes
Fixes a potential security vulnerability where on-the-fly changes to the dtype of a tensor reference may lead to undefined behavior.
Linux binaries on pypi are built with AVX. Please use GCS binaries for non-AVX machines.

TensorFlow 1.5.0
Release 1.5.0
Breaking Changes
Prebuilt binaries are now built against CUDA 9 and cuDNN 7.
Starting from 1.6 release, our prebuilt binaries will use AVX instructions.
This may break TF on older CPUs.

[링크 : https://github.com/tensorflow/tensorflow/releases?q=avx&expanded=true]

[링크 : https://github.com/tensorflow/tensorflow/issues/17411]

[링크 : https://m.blog.naver.com/kimground/221552744535]

[링크 : https://hiseon.me/data-analytics/tensorflow/tensorflow-errors/]

저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

eiq 데이터 구조  (0) 2025.09.05
ubuntu 22.04 + cuda + cudnn 설치  (0) 2025.09.04
eiq on windows with nvidia  (0) 2025.09.03
vainfo  (0) 2025.09.03
ubuntu nvidia 드라이버 설치  (0) 2025.09.03
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 3. 17:41

eiq on windows with nvidia

G3460 / win 10 클린설치하고

nvidia 580.97 드라이버

cuda 13.0.0.0

cudnn 9.12.0 설치하고

eiq 1.16.0.106_b250703 을 설치하고 실행하는데 아래와 같이 뜨길래

해보니 결론은.. 

msvc++ redist 2015-2011 14.44.35211 설치하고 해결 -_-

 

그런데 왜 학습할 때 gpu 0% 뜨냐 -_-

 

C:\nxp\eIQ_Toolkit_v1.16.0>
]eIQ Portal version 2.16.0
------------------------------------------------
C:\nxp\eIQ_Toolkit_v1.16.0\resources\app.asar
------------------------------------------------
Launch -> C:\nxp\eIQ_Toolkit_v1.16.0
16:08:22.435 > Launching Application
Display size is 1920x1080
[2025-09-03T07:08:22.572Z] ExtensionHostController running as pid 7112

is-elevated: false
[2025-09-03T07:08:24.417Z] Socket.io logging server connected to socket 9rpMJdiuTUcknHa9AAAB
[2025-09-03T07:08:24.443Z] InstalledExtensionsReady barrier opened
(buffered) [2025-09-03T07:08:23.262Z] ExtensionHost running as pid 8976
(buffered) [2025-09-03T07:08:23.474Z] Connected to sqlite://C:\Users\user\AppData\Roaming\eiqportal-nodejs\Config\state.sqlite
(buffered) [2025-09-03T07:08:24.425Z] Preparing Python environment...
(buffered) [2025-09-03T07:08:24.426Z] Installing Flask@3.0.3
(buffered) [2025-09-03T07:08:24.427Z] Activated 5 extensions
- NXP.eiqtlk-ext-armvela version 1.3.1
   from C:\Users\user\.eiqportal\extensions\eiqtlk-ext-armvela
- nxp.explainability version 0.4.0
   from C:\Users\user\.eiqportal\extensions\eiqtlk-ext-explainability
- NXP.eiqtlk-ext-tao version 0.0.1
   from C:\Users\user\.eiqportal\extensions\eiqtlk-ext-tao
- NXP.eiqtlk-ext-vision-pipeline version 1.4.0-rc2
   from C:\Users\user\.eiqportal\extensions\eiqtlk-ext-vision-pipeline
- NXP.eiqtlk-ext-watermarking version 1.2.3
   from C:\Users\user\.eiqportal\extensions\eiqtlk-ext-watermarking
16:08:27.202 > [CONVERTER] failed to import eiq.modelserver
16:08:27.227 > [CONVERTER] Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
    from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: DLL 占십깍옙화 占쏙옙틴占쏙옙 占쏙옙占쏙옙占쏙옙 占쏙옙 占쏙옙占쏙옙占싹댐옙.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\eiq\modelserver\__init__.py", line 4, in <module>
    import tensorflow as tf
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\__init__.py", line 40, in <module>
    from tensorflow.python import pywrap_tensorflow as _pywrap_tensorflow  # pylint: disable=unused-import
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 85, in <module>
    raise ImportError(
ImportError: Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
    from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: DLL 占십깍옙화 占쏙옙틴占쏙옙 占쏙옙占쏙옙占쏙옙 占쏙옙 占쏙옙占쏙옙占싹댐옙.


Failed to load the native TensorFlow runtime.
See https://www.tensorflow.org/install/errors for some common causes and solutions.
If you need help, create an issue at https://github.com/tensorflow/tensorflow/issues and include the entire stack trace above this error message.

16:08:27.280 > [TRAINER] failed to import eiq.trainer.server.__main__Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
    from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: DLL 占십깍옙화 占쏙옙틴占쏙옙 占쏙옙占쏙옙占쏙옙 占쏙옙 占쏙옙占쏙옙占싹댐옙.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\eiq\trainer\server\__main__.py", line 3, in <module>

16:08:27.299 > [TRAINER] LOADING PYTHON -> C:\nxp\eIQ_Toolkit_v1.16.0\python\python310.dll
PYTHONPATH: C:\nxp\eIQ_Toolkit_v1.16.0\python;C:\nxp\eIQ_Toolkit_v1.16.0\python\DLLs;C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib;C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages

16:08:27.323 > eiq-converter terminated
16:08:27.352 > [TRAINER]     from eiq.trainer.utils.messenger import ColoredPrint
  File "eiq/trainer/utils/messenger.py", line 3, in init eiq.trainer.utils.messenger
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\__init__.py", line 40, in <module>
    from tensorflow.python import pywrap_tensorflow as _pywrap_tensorflow  # pylint: disable=unused-import
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 85, in <module>
    raise ImportError(
ImportError: Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
    from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: DLL 占십깍옙화 占쏙옙틴占쏙옙 占쏙옙占쏙옙占쏙옙 占쏙옙 占쏙옙占쏙옙占싹댐옙.


Failed to load the native TensorFlow runtime.
See https://www.tensorflow.org/install/errors for some common causes and solutions.
If you need help, create an issue at https://github.com/tensorflow/tensorflow/issues and include the entire stack trace above this error message.

16:08:27.402 > eiq-trainer terminated
16:08:27.599 > [IMPORTER] Traceback (most recent call last):

16:08:27.622 > [IMPORTER]   File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
    from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: DLL 占십깍옙화 占쏙옙틴占쏙옙 占쏙옙占쏙옙占쏙옙 占쏙옙 占쏙옙占쏙옙占싹댐옙.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\eiq\importer\server\__main__.py", line 11, in <module>

16:08:27.653 > [IMPORTER]     from eiq.importer.datasets.TFDS_importer import TFDSDataset
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\eiq\importer\datasets\TFDS_importer.py", line 1, in <module>
    import tensorflow as tf
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\__init__.py", line 40, in <module>
    from tensorflow.python import pywrap_tensorflow as _pywrap_tensorflow  # pylint: disable=unused-import
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 85, in <module>
    raise ImportError(
ImportError: Traceback (most recent call last):
  File "C:\nxp\eIQ_Toolkit_v1.16.0\python\Lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
    from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: DLL 占십깍옙화 占쏙옙틴占쏙옙 占쏙옙占쏙옙占쏙옙 占쏙옙 占쏙옙占쏙옙占싹댐옙.


Failed to load the native TensorFlow runtime.
See https://www.tensorflow.org/install/errors for some common causes and solutions.
If you need help, create an issue at https://github.com/tensorflow/tensorflow/issues and include the entire stack trace above this error message.

16:08:27.802 > eiq-importer terminated
[2025-09-03T07:08:29.389Z] Extension host terminating: eIQ Portal shutdown
[2025-09-03T07:08:29.392Z] Deactivated all extensions
[2025-09-03T07:08:29.392Z] Backing up to C:\Users\user\AppData\Roaming\eiqportal-nodejs\Config\state.sqlite.backup
[TRAINER] Invalid Path
[DATASTORE] Shutting down...
TRAINER STOP ERROR: FetchError: request to http://127.0.0.1:10821/validate/stop failed, reason: connect ECONNREFUSED 127.0.0.1:10821
Uninstalled console patch
16:08:29.415 > eiq-validator terminated
16:08:29.426 > modelrunner client terminated
16:08:29.436 > modelrunner terminated
16:08:29.444 > modeleditor terminated
[2025-09-03T07:08:29.523Z] Extension host with pid 8976 exiting with code 0
[2025-09-03T07:08:29.577Z] ExtensionHost terminated with code 0 and signal null.
[2025-09-03T07:08:29.578Z] Closed Socket.io logging server
16:08:30.395 > eiq-datastore terminated
eIQ Portal Shutting Down...

 

 

[링크 : https://blog.naver.com/daebarq/222109221303]

[링크 : https://stackoverflow.com/questions/67259900/tensorflow-installation-on-windows-dll-load-failed-while-importing-pywrap-tens]

[링크 : https://blog.naver.com/se0_y/221986584579]

[링크 : https://learn.microsoft.com/en-us/cpp/windows/latest-supported-vc-redist?view=msvc-170]

[링크 : https://github.com/tensorflow/tensorflow/issues/22512]

 

저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

ubuntu 22.04 + cuda + cudnn 설치  (0) 2025.09.04
import tensorflow illegal instruction  (0) 2025.09.04
vainfo  (0) 2025.09.03
ubuntu nvidia 드라이버 설치  (0) 2025.09.03
eiq 에러들  (0) 2025.08.26
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 3. 11:06

vainfo

eiq 에서 libva 에러가 났는데

wayland에서 x.org로 바꾸니 에러가 발생하지 않는다. 중요한(?) 에러는 아니었는 듯.

 

$ cat /proc/cpuinfo 
model name : Intel(R) Core(TM) i7-10510U CPU @ 1.80GHz

$ vainfo
libva info: VA-API version 1.14.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_14
libva info: va_openDriver() returns 0
vainfo: VA-API version: 1.14 (libva 2.12.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 22.3.1 ()
vainfo: Supported profile and entrypoints
      VAProfileMPEG2Simple            : VAEntrypointVLD
      VAProfileMPEG2Main              : VAEntrypointVLD
      VAProfileH264Main               : VAEntrypointVLD
      VAProfileH264Main               : VAEntrypointEncSliceLP
      VAProfileH264High               : VAEntrypointVLD
      VAProfileH264High               : VAEntrypointEncSliceLP
      VAProfileJPEGBaseline           : VAEntrypointVLD
      VAProfileJPEGBaseline           : VAEntrypointEncPicture
      VAProfileH264ConstrainedBaseline: VAEntrypointVLD
      VAProfileH264ConstrainedBaseline: VAEntrypointEncSliceLP
      VAProfileVP8Version0_3          : VAEntrypointVLD
      VAProfileHEVCMain               : VAEntrypointVLD
      VAProfileHEVCMain10             : VAEntrypointVLD
      VAProfileVP9Profile0            : VAEntrypointVLD
      VAProfileVP9Profile2            : VAEntrypointVLD

 

G3460 + nvidia 1060

$ cat /proc/cpuinfo 
model name : Intel(R) Pentium(R) CPU G3460 @ 3.50GHz

$ cat log 
$ vainfo
libva info: VA-API version 1.14.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/nvidia_drv_video.so
libva info: va_openDriver() returns -1
vaInitialize failed with error code -1 (unknown libva error),exit

 

$ LIBVA_DRIVER_NAME=iHD vainfo 
libva info: VA-API version 1.14.0
libva info: User environment variable requested driver 'iHD'
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_14
DRM_IOCTL_I915_GEM_APERTURE failed: Invalid argument
Assuming 131072kB available aperture size.
May lead to reduced performance or incorrect rendering.
get chip id failed: -1 [22]
param: 4, val: 0
libva error: /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so init failed
libva info: va_openDriver() returns 18
vaInitialize failed with error code 18 (invalid parameter),exit

 

G3460 

$ vainfo
libva info: VA-API version 1.14.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: Found init function __vaDriverInit_1_14
libva error: /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so init failed
libva info: va_openDriver() returns 1
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: Found init function __vaDriverInit_1_10
libva info: va_openDriver() returns 0
vainfo: VA-API version: 1.14 (libva 2.12.0)
vainfo: Driver version: Intel i965 driver for Intel(R) Haswell Desktop - 2.4.1
vainfo: Supported profile and entrypoints
      VAProfileMPEG2Simple            : VAEntrypointVLD
      VAProfileMPEG2Simple            : VAEntrypointEncSlice
      VAProfileMPEG2Main              : VAEntrypointVLD
      VAProfileMPEG2Main              : VAEntrypointEncSlice
      VAProfileH264ConstrainedBaseline: VAEntrypointVLD
      VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
      VAProfileH264Main               : VAEntrypointVLD
      VAProfileH264Main               : VAEntrypointEncSlice
      VAProfileH264High               : VAEntrypointVLD
      VAProfileH264High               : VAEntrypointEncSlice
      VAProfileH264MultiviewHigh      : VAEntrypointVLD
      VAProfileH264MultiviewHigh      : VAEntrypointEncSlice
      VAProfileH264StereoHigh         : VAEntrypointVLD
      VAProfileH264StereoHigh         : VAEntrypointEncSlice
      VAProfileVC1Simple              : VAEntrypointVLD
      VAProfileVC1Main                : VAEntrypointVLD
      VAProfileVC1Advanced            : VAEntrypointVLD
      VAProfileNone                   : VAEntrypointVideoProc
      VAProfileJPEGBaseline           : VAEntrypointVLD

[링크 : https://github.com/elFarto/nvidia-vaapi-driver/issues/272]

저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

import tensorflow illegal instruction  (0) 2025.09.04
eiq on windows with nvidia  (0) 2025.09.03
ubuntu nvidia 드라이버 설치  (0) 2025.09.03
eiq 에러들  (0) 2025.08.26
eiq 학습 시도  (0) 2025.08.26
Posted by 구차니
embeded/i.mx 8m plus2025. 9. 3. 10:41

ubuntu nvidia 드라이버 설치

 

sudo add-apt-repository ppa:graphics-drivers/ppa
ubuntu-drivers devices
sudo ubuntu-drivers autoinstall

 

$ ubuntu-drivers devices
== /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0 ==
modalias : pci:v000010DEd00001C03sv000010DEsd00000000bc03sc00i00
vendor   : NVIDIA Corporation
model    : GP106 [GeForce GTX 1060 6GB]
driver   : nvidia-driver-550 - distro non-free
driver   : nvidia-driver-470 - distro non-free
driver   : nvidia-driver-450-server - distro non-free
driver   : nvidia-driver-565 - third-party non-free
driver   : nvidia-driver-575 - distro non-free recommended
driver   : nvidia-driver-580 - third-party non-free
driver   : nvidia-driver-470-server - distro non-free
driver   : nvidia-driver-575-server - distro non-free
driver   : nvidia-driver-570 - third-party non-free
driver   : nvidia-driver-545 - distro non-free
driver   : nvidia-driver-535 - distro non-free
driver   : nvidia-driver-418-server - distro non-free
driver   : nvidia-driver-390 - distro non-free
driver   : nvidia-driver-570-server - distro non-free
driver   : nvidia-driver-535-server - distro non-free
driver   : xserver-xorg-video-nouveau - distro free builtin

 

sudo apt-get install nvidia-driver-580

 

리부팅 이후

$ nvidia-smi
Wed Sep  3 10:53:56 2025       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 580.76.05              Driver Version: 580.76.05      CUDA Version: 13.0     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce GTX 1060 6GB    Off |   00000000:01:00.0  On |                  N/A |
| 42%   38C    P8              6W /  120W |     172MiB /   6144MiB |      1%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|    0   N/A  N/A             976      G   /usr/lib/xorg/Xorg                       73MiB |
|    0   N/A  N/A            1271      G   /usr/bin/gnome-shell                     86MiB |
+-----------------------------------------------------------------------------------------+

[링크 : https://2dudwns.tistory.com/20]

 

cuda 설치도 패키지로..!

sudo apt-get install nvidia-cuda-toolkit
저작자표시 (새창열림)

'embeded > i.mx 8m plus' 카테고리의 다른 글

eiq on windows with nvidia  (0) 2025.09.03
vainfo  (0) 2025.09.03
eiq 에러들  (0) 2025.08.26
eiq 학습 시도  (0) 2025.08.26
nvidia tao toolkit  (0) 2025.08.22
Posted by 구차니
embeded/ARM2025. 8. 28. 14:18

SVE(Scalable Vector Extension)

armv7 까지는 NEON 이고 armv8 부터는 SVE 라고 이름이 달라지는 듯

 

Scalable Vector Extension (SVE) is a vector extension the A64 instruction set of the Armv8-A architecture. Armv9-A builds on SVE with the SVE2 extension. Unlike other SIMD architectures, SVE and SVE2 do not define the size of the vector registers, but constrains it to a range of possible values, from a minimum of 128 bits up to a maximum of 2048 in 128-bit wide units. Therefore, any CPU vendor can implement the extension by choosing the vector register size that better suits the workloads the CPU is targeting. The design of SVE and SVE2 guarantees that the same program can run on different implementations of the instruction set architecture without the need to recompile the code.

[링크 : https://developer.arm.com/Architectures/Scalable%20Vector%20Extensions]

[링크 : https://developer.arm.com/Architectures/SVE]

저작자표시 (새창열림)

'embeded > ARM' 카테고리의 다른 글

emmc 파티션 정렬  (0) 2024.02.07
arm asm rev  (0) 2023.09.14
cortex-a53  (0) 2023.08.31
aarch64 vector register  (0) 2023.08.23
arm vsub operator  (0) 2023.08.09
Posted by 구차니

티스토리툴바