2e:null 2f:["$","title",null,{"children":"AMD 그래픽카드를 위한 Tensorflow 설치가이드 | Humble Data Miner | Observable"}] 34:T8ca,--- ### Tensorflow 2.13 빌드 및 설치 * [AMD 공식 Tensorflow 설치 가이드](https://docs.amd.com/bundle/ROCm-Deep-Learning-Guide-v5.4.2/page/Frameworks_Installation.html) 참고 ${ !(arch=="gfx1031" || arch=="gfx1030") ? "":""} * ROCm 상에서 관리되는 tensorflow 소스코드 다운로드후 디렉토리 이동 ~~~bash git clone https://github.com/ROCmSoftwarePlatform/tensorflow-upstream.git cd tensorflow-upstream ~~~ ${arch=="gfx803" || arch=="gfx1010"? '* device_description.h 파일을 열어서 지원가능 목록에 '+arch+' 추가':""} ${arch=="gfx803"? '':''} ${arch=="gfx1010"? '':''} * build_rocm_python3를 실행해 빌드,패키징 및 설치를 순차적으로 진행 ~~~bash ${arch=="gfx1031"? "HCC_AMDGPU_TARGET=gfx1030 TF_ROCM_AMDGPU_TARGETS=gfx1030 ":""}./build_rocm_python3 /opt/rocm-5.4.2 ~~~ ${arch=="gfx1031"? " * gfx1031의 경우 추가 환경변수(HCC_AMDGPU_TARGET, TF_ROCM_AMDGPU_TARGETS)를 설정해 gfx1030 타겟빌드하도록 유도": ""} 35:[] 32:{"id":"47d20dc2be5b0489","slug":"amdgpu-tensorflow-install-guide-on-ubuntu","title":"AMD 그래픽카드를 위한 Tensorflow 설치가이드","thumbnail":"c6d9c52b02f7f344a60c100673943bc7582494c22011364b1aded488e0be0f32","access_level":"public","create_time":"$D2023-02-13T04:52:26.749Z","update_time":"$D2023-02-13T12:05:58.674Z","view_time":null,"first_public_time":"$D2023-02-13T04:52:26.749Z","has_unpublished_edits":null,"fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"50faa4da52a9ff9b","login":"humble-data-miner","name":"Humble Data Miner","avatar_url":"https://avatars-next.observableusercontent.com/3046b1bea86c5e825baea7c0c403d90ad9690b5396b11b5d02b3a0839fbcb946","tier":"starter_2024"},"version":301,"publish_version":null,"role":null,"authored":false,"authors":[{"type":"individual","id":"4e2fc41e06361cc3","login":"humble-data-miner","name":"Humble Data Miner","avatar_url":"https://avatars-next.observableusercontent.com/3046b1bea86c5e825baea7c0c403d90ad9690b5396b11b5d02b3a0839fbcb946","tier":null}],"members":[{"type":"individual","id":"4e2fc41e06361cc3","login":"humble-data-miner","name":"Humble Data Miner","avatar_url":"https://avatars-next.observableusercontent.com/3046b1bea86c5e825baea7c0c403d90ad9690b5396b11b5d02b3a0839fbcb946","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"title":"AMD 그래픽카드를 위한 Tensorflow 설치가이드","cells":[{"id":0,"value":"# AMD 그래픽카드를 위한 Tensorflow 설치가이드","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":10,"value":"---\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":73,"value":"### 전제\n* Ubuntu 22.04 기준으로 작성\n* Tensorflow 2.13 기준으로 작성\n* ROCm 5.4.2 기준으로 작성\n* 시스템에 직접 설치한다 가정\n* gfx1030외에는 공식지원이 아니기때문에 일부기능만 동작할수 있음\n * 설치 성공여부는 MNIST 학습 가능여부를 기준으로 판단\n * 웹 검색결과 gfx803은 FP16 미지원 이슈가 있는것으로 보임","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":91,"value":"### 설치할 그래픽카드의 아키텍쳐 선택","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":18,"value":"viewof arch = Inputs.radio([\"gfx803\",\"gfx1010\", \"gfx1030\", \"gfx1031\"], {value:\"gfx803\",label: \"아키텍쳐 선택\",format:x=>`${x} (${archprod_map[x]})`})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":38,"value":"### 선택한 아키텍쳐 **${arch}** 에 대한 테스트 결과\n\n* ${arch==\"gfx803\"? 'RX 580 으로 테스트 **통과**':\n arch==\"gfx1010\" || arch==\"gfx1012\"? '그래픽카드가 없어 테스트해보지 못함 (gfx803을 기준으로 작성)':\n arch==\"gfx1030\"? 'RX 6800 으로 테스트 **통과**':\n arch==\"gfx1031\"? 'RX 6700 XT 으로 테스트 **실패** (무한루프 이슈)': \"...\"}\n${arch==\"gfx1031\"? '* 무한루프 이슈에 대한 설명은 [호기심해결 실전편 참고](https://observablehq.com/@humble-data-miner/myth-buster-in-ml-tf-perf-btw-nvidia-and-amd)':\"\"}","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":131,"value":"---\n### ROCm 5.4.2 설치\n\n* [AMD 공식 ROCm 설치 가이드](https://docs.amd.com/bundle/ROCm-Installation-Guide-v5.4.2/page/How_to_Install_ROCm.html) 참고\n* AMDGPU installer package 설치\n```bash\nsudo apt-get update\nwget https://repo.radeon.com/amdgpu-install/5.4.2/ubuntu/jammy/amdgpu-install_5.4.50402-1_all.deb\nsudo apt-get install ./amdgpu-install_5.4.50402-1_all.deb\n```\n* AMDGPU installer를 통해 GPU driver, ROCm과 추가 개발툴 설치\n```bash\nsudo amdgpu-install --usecase=hiplibsdk,rocm\nsudo apt install miopen-hip\n```","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":148,"value":"$34","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":237,"value":"---\n### Tensorflow 테스트\n\n* 장치가 Tensorflow 사용가능한지 목록에서 확인\n~~~bash\npython3 -c 'from tensorflow.python.client import device_lib;print(device_lib.list_local_devices())'\n~~~\n\n\n* MNIST 테스트 스크립트 실행 후 OK 확인\n~~~bash\ncd ~\ngit clone https://github.com/tensorflow/models.git tensorflow.models\ncd tensorflow.models\n${arch==\"gfx1031\"? \"HSA_OVERRIDE_GFX_VERSION=10.3.0\":\"\"} python ./official/legacy/image_classification/mnist_test.py\n~~~\n${arch==\"gfx1031\"? \" * gfx1031일 경우 반드시 환경변수(HSA_OVERRIDE_GFX_VERSION)를 설정해 gfx1030으로 동작하도록 유도\":\"\"}","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":5,"value":"---\n## 부록","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":25,"value":"archprod_map = {return {\n \"gfx803\":\"RX470, RX480, RX570, RX580, ...\",\n \"gfx1010\":\"RX5600, RX5600XT, RX5700, RX5700XT\",\n \"gfx1030\":\"RX6800, RX6800XT, RX6900XT, RX6950XT\",\n \"gfx1031\":\"RX6700, RX6700XT\"};}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":7,"value":"html`\n\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W35","stdlib":"1","titling":"auto"}} 33:"$32:body"