本文為英文版的機器翻譯版本，如內容有任何歧義或不一致之處，概以英文版為準。 # AWS 深度學習 AMI GPU PyTorch 2.5 (Amazon Linux 2023) **注意** 如需詳細的軟體套件資訊，請參閱下列連結的個別版本備註頁面 如需入門說明，請參閱 [DLAMI 入門](getting-started.md)。 #### AMI 名稱格式 + 深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.\$1\$1PATCH\$1VERSION\$1 (Amazon Linux 2023) \$1\$1YYYY-MM-DD\$1 #### SSM 參數查詢 **請使用下列公有 SSM 參數查詢來尋找最新的 DLAMI： ** ``` export SSM_PARAMETER=oss-nvidia-driver-gpu-pytorch-2.5-amazon-linux-2023/latest/ami-id && \ aws ssm get-parameter --region us-east-1 \ --name /aws/service/deeplearning/ami/x86_64/$SSM_PARAMETER \ --query "Parameter.Value" \ --output text ``` #### AWSCLI 查詢 **請使用下列 AWS 查詢來尋找最新的 DLAMI： ** ``` aws ec2 describe-images --region us-east-1 --owners amazon --filters 'Name=name,Values=Deep Learning OSS Nvidia Driver AMI GPU PyTorch 2.5.? (Amazon Linux 2023) ????????' 'Name=state,Values=available' --query 'reverse(sort_by(Images, &CreationDate))[:1].ImageId' --output text ``` #### 最新版本 **Topics** + [深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20251101](aws-deep-learning-ami-gpupt25oss-al2023-2025-11-10.md) + [深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20251007](aws-deep-learning-ami-gpupt25oss-al2023-2025-10-14.md) + [深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20250927](aws-deep-learning-ami-gpupt25oss-al2023-2025-09-30.md) # 深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20251101 如需入門說明，請參閱 [DLAMI 入門](getting-started.md)。 ##### 主要套件版本 | 套件名稱 | 版本 | | --- | --- | | framework\$1version | 2.5.1 | | gdr\$1copy | 2.4.4 | | supported\$1ec2\$1instances | G4dn, G5, G6, Gr6, G6e, P4, P4de, P5, P5e, P5en | | efa\$1version | 1.43.3 | | ebs\$1volume\$1type | gp3 | | nvidia\$1driver | 580.95.05 | | python\$1location | /opt/conda/envs/pytorch/bin/python | | nvidia\$1cuda\$1stack | /usr/local/cuda-12.4 | | ssm\$1agent\$1version | 3.3.3050.0 | | kernel\$1version | 6.1.156-177.286.amzn2023.x86\$164 | | nvidia\$1container\$1toolkit\$1version | 1.18.0 | | dcgm\$1version | /bin/bash: line 1: dcgmi: command not found | | ofi\$1nccl\$1version | 1.16.3 | | operating\$1system | Amazon Linux 2023.9.20251027 | | default\$1cuda | /usr/local/cuda-12.4/ | | compute\$1architecture | x86\$164 | ##### 先前版本的套件變更 **注意** 所有 DLAMI 都會依照 AWS 最佳實務掃描漏洞。雖然我們持續升級至最新的套件版本，但某些相依性可能需要特定的套件降級來維持相容性。這些調整不會影響我們的安全狀態。如需 AWS 安全措施的相關資訊，請參閱我們的[AWS 安全文件 ](https://docs.aws.amazon.com/dlami/latest/devguide/security.html) ##### 新增的套件 | | | --- | | apr-util-lmdb-1.6.3-1.amzn2023.0.2 | | kernel-livepatch-6.1.156-177.286-1.0-0.amzn2023 | ##### 更新套件 | 套件名稱 | 先前版本 | 新版本 | | --- | --- | --- | | Brotli | 1.1.0 | 1.2.0 | | amazon-linux-repo-s3 | 2023.9.20251020-0.amzn2023 | 2023.9.20251027-0.amzn2023 | | apr-util | 1.6.3-1.amzn2023.0.1 | 1.6.3-1.amzn2023.0.2 | | apr-util-openssl | 1.6.3-1.amzn2023.0.1 | 1.6.3-1.amzn2023.0.2 | | audit | 3.1.5-1.amzn2023.0.1 | 3.1.5-1.amzn2023.0.2 | | audit-libs | 3.1.5-1.amzn2023.0.1 | 3.1.5-1.amzn2023.0.2 | | awscli | 2.31.22 | 2.31.27 | | boost-filesystem | 1.75.0-4.amzn2023.0.3 | 1.75.0-4.amzn2023.0.4 | | boost-system | 1.75.0-4.amzn2023.0.3 | 1.75.0-4.amzn2023.0.4 | | boost-thread | 1.75.0-4.amzn2023.0.3 | 1.75.0-4.amzn2023.0.4 | | boto3 | 1.40.59 | 1.40.64 | | botocore | 1.40.59 | 1.40.64 | | fastapi | 0.120.0 | 0.120.4 | | fsspec | 2025.9.0 | 2025.10.0 | | go-srpm-macros | 3.2.0-37.amzn2023 | 3.8.0-1.amzn2023.0.1 | | grub2-common | 2.06-61.amzn2023.0.19 | 2.06-61.amzn2023.0.20 | | grub2-efi-x64-ec2 | 2.06-61.amzn2023.0.19 | 2.06-61.amzn2023.0.20 | | grub2-pc-modules | 2.06-61.amzn2023.0.19 | 2.06-61.amzn2023.0.20 | | grub2-tools | 2.06-61.amzn2023.0.19 | 2.06-61.amzn2023.0.20 | | grub2-tools-minimal | 2.06-61.amzn2023.0.19 | 2.06-61.amzn2023.0.20 | | inspectorssmplugin | 1.0.402-1 | 1.0.404-1 | | ipykernel | 7.0.1 | 7.1.0 | | java-17-amazon-corretto-headless | 17.0.16\$18-1.amzn2023.1 | 17.0.17\$110-1.amzn2023.1 | | kernel | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | kernel-devel | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | kernel-headers | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | kernel-libbpf | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | kernel-livepatch-repo-s3 | 2023.9.20251020-0.amzn2023 | 2023.9.20251027-0.amzn2023 | | kernel-modules-extra | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | kernel-modules-extra-common | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | kernel-tools | 6.1.155-176.282.amzn2023 | 6.1.156-177.286.amzn2023 | | lark | 1.3.0 | 1.3.1 | | libdnf | 0.69.0-8.amzn2023.0.5 | 0.69.0-8.amzn2023.0.6 | | librepo | 1.14.5-2.amzn2023.0.1 | 1.14.5-2.amzn2023.0.2 | | libsoup3 | 3.6.5-50.amzn2023 | 3.6.5-52.amzn2023 | | libsss\$1certmap | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | libsss\$1idmap | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | libsss\$1nss\$1idmap | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | libsss\$1sudo | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | libxslt | 1.1.43-1.amzn2023.0.2 | 1.1.43-1.amzn2023.0.3 | | libxslt-devel | 1.1.43-1.amzn2023.0.2 | 1.1.43-1.amzn2023.0.3 | | marshmallow | 4.0.1 | 4.1.0 | | narwhals | 2.9.0 | 2.10.1 | | perl-Math-BigInt-FastCalc | 0.500.900-458.amzn2023.0.2 | 0.501.400-3.amzn2023.0.1 | | pillow | 11.3.0 | 12.0.0 | | platformdirs | 4.2.2 | 4.5.0 | | psutil | 7.1.1 | 7.1.2 | | python3 | 3.9.23-1.amzn2023.0.3 | 3.9.24-1.amzn2023.0.3 | | python3-audit | 3.1.5-1.amzn2023.0.1 | 3.1.5-1.amzn2023.0.2 | | python3-hawkey | 0.69.0-8.amzn2023.0.5 | 0.69.0-8.amzn2023.0.6 | | python3-libdnf | 0.69.0-8.amzn2023.0.5 | 0.69.0-8.amzn2023.0.6 | | python3-libs | 3.9.23-1.amzn2023.0.3 | 3.9.24-1.amzn2023.0.3 | | runc | 1.3.1-1.amzn2023.0.1 | 1.3.2-2.amzn2023.0.1 | | sagemaker | 2.253.1 | 2.254.1 | | sagemaker-core | 1.0.59 | 1.0.61 | | scipy | 1.16.2 | 1.16.3 | | sssd-client | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | sssd-common | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | sssd-kcm | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | sssd-nfs-idmap | 2.9.4-1.amzn2023.0.2 | 2.9.4-1.amzn2023.0.3 | | starlette | 0.48.0 | 0.49.3 | | system-release | 2023.9.20251020-0.amzn2023 | 2023.9.20251027-0.amzn2023 | | tblib | 3.2.0 | 3.2.1 | | termcolor | 3.1.0 | 3.2.0 | | typing\$1extensions | 4.15.0 | 4.12.2 | | webcolors | 24.11.1 | 25.10.0 | | xyzservices | 2025.4.0 | 2025.10.0 | | zipp | 3.23.0 | 3.19.2 | ##### 移除的套件 | 套件名稱 | | --- | | kernel-livepatch-6.1.155-176.282 | # 深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20251007 如需入門說明，請參閱 [DLAMI 入門](getting-started.md)。 ##### 注意 + 此版本包含 Nvidia [10 月安全性佈告欄中找到受影響 Nvidia 驅動程式套件的 CVE ](https://nvidia.custhelp.com/app/answers/detail/a_id/5703)修補程式 ##### 主要套件版本 | 套件名稱 | 版本 | | --- | --- | | framework\$1version | 2.5.1 | | gdr\$1copy | 2.4.4 | | supported\$1ec2\$1instances | G4dn, G5, G6, Gr6, G6e, P4, P4de, P5, P5e, P5en | | efa\$1version | 1.43.3 | | ebs\$1volume\$1type | gp3 | | nvidia\$1driver | 580.95.05 | | python\$1location | /opt/conda/envs/pytorch/bin/python | | nvidia\$1cuda\$1stack | /usr/local/cuda-12.4 | | ssm\$1agent\$1version | 3.3.3050.0 | | kernel\$1version | 6.1.153-175.280.amzn2023.x86\$164 | | nvidia\$1container\$1toolkit\$1version | 1.17.8 | | ofi\$1nccl\$1version | 1.16.3 | | operating\$1system | Amazon Linux 2023.9.20250929 | | default\$1cuda | /usr/local/cuda-12.4/ | | compute\$1architecture | x86\$164 | ##### 先前版本的套件變更 **注意** 所有 DLAMI 都會依照 AWS 最佳實務掃描漏洞。雖然我們持續升級至最新的套件版本，但某些相依性可能需要特定的套件降級來維持相容性。這些調整不會影響我們的安全狀態。如需 AWS 安全措施的相關資訊，請參閱我們的[AWS 安全文件 ](https://docs.aws.amazon.com/dlami/latest/devguide/security.html) ##### 新增的套件 | | | --- | | kernel-livepatch-6.1.153-175.280-1.0-0.amzn2023 | | nvidia-fabricmanager-580.95.05-1 | ##### 更新套件 | 套件名稱 | 先前版本 | 新版本 | | --- | --- | --- | | Authlib | 1.6.4 | 1.6.5 | | MarkupSafe | 3.0.2 | 3.0.3 | | amazon-linux-repo-s3 | 2023.8.20250915-0.amzn2023 | 2023.9.20250929-0.amzn2023 | | amazon-rpm-config | 228-9.amzn2023.0.1 | 228-10.amzn2023.0.1 | | amazon-ssm-agent | 3.3.2299.0-1.amzn2023 | 3.3.3050.0-1.amzn2023 | | attrs | 25.3.0 | 25.4.0 | | awscli | 2.31.3 | 2.31.9 | | beautifulsoup4 | 4.13.5 | 4.14.2 | | binutils | 2.41-50.amzn2023.0.3 | 2.41-50.amzn2023.0.4 | | boto3 | 1.40.40 | 1.40.46 | | botocore | 1.40.40 | 1.40.46 | | certifi | 2025.8.3 | 2025.10.5 | | container-selinux | 2.233.0-1.amzn2023 | 2.242.0-1.amzn2023 | | coreutils | 8.32-30.amzn2023.0.3 | 8.32-30.amzn2023.0.4 | | coreutils-common | 8.32-30.amzn2023.0.3 | 8.32-30.amzn2023.0.4 | | cpp14 | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | cryptography | 46.0.1 | 46.0.2 | | cups-filesystem | 2.4.11-8.amzn2023.0.1 | 2.4.14-1.amzn2023.0.1 | | cups-libs | 2.4.11-8.amzn2023.0.1 | 2.4.14-1.amzn2023.0.1 | | dnf-plugin-support-info | 1.7-1.amzn2023 | 1.8-1.amzn2023 | | efa | 2.17.2-1.amzn2023 | 2.17.3-1.amzn2023 | | expat | 2.6.3-1.amzn2023.0.2 | 2.6.3-1.amzn2023.0.3 | | fastapi | 0.117.1 | 0.118.0 | | fonttools | 4.60.0 | 4.60.1 | | gcc14 | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | inspectorssmplugin | 1.0.398-1 | 1.0.399-1 | | kernel | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | kernel-devel | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | kernel-headers | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | kernel-libbpf | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | kernel-livepatch-repo-s3 | 2023.8.20250915-0.amzn2023 | 2023.9.20250929-0.amzn2023 | | kernel-modules-extra | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | kernel-modules-extra-common | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | kernel-tools | 6.1.150-174.273.amzn2023 | 6.1.153-175.280.amzn2023 | | libatomic | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | libgcc | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | libgccjit | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | libgfortran | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | libgomp | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | libnccl-ofi | 1.16.2-1.amzn2023 | 1.16.3-1.amzn2023 | | libquadmath | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | libstdc\$1\$1 | 14.2.1-7.amzn2023.0.1 | 14.2.1-7.amzn2023.0.2 | | llvmlite | 0.45.0 | 0.45.1 | | microcode\$1ctl | 2.1-53.amzn2023.0.13 | 2.1-53.amzn2023.0.14 | | narwhals | 2.5.0 | 2.7.0 | | nltk | 3.9.1 | 3.9.2 | | numba | 0.62.0 | 0.62.1 | | nvlsm | 2025.06.5-1 | 2025.06.6-1 | | openjpeg2 | 2.4.0-11.amzn2023.0.6 | 2.5.2-5.amzn2023.0.1 | | pandas | 2.3.2 | 2.3.3 | | platformdirs | 4.4.0 | 4.2.2 | | pydantic | 2.11.9 | 2.11.10 | | python3-awscrt | 0.26.1-1.amzn2023.0.1 | 0.27.6-1.amzn2023.0.1 | | rust-toolset-srpm-macros | 1.89.0-1.amzn2023.0.3 | 1.90.0-1.amzn2023.0.1 | | sagemaker | 2.251.1 | 2.252.0 | | selinux-policy | 38.1.50-1.amzn2023.0.2 | 38.1.65-1.amzn2023.0.1 | | selinux-policy-targeted | 38.1.50-1.amzn2023.0.2 | 38.1.65-1.amzn2023.0.1 | | system-release | 2023.8.20250915-0.amzn2023 | 2023.9.20250929-0.amzn2023 | | tomli | 2.2.1 | 2.0.1 | | typing-inspection | 0.4.1 | 0.4.2 | | typing\$1extensions | 4.15.0 | 4.12.2 | | zipp | 3.19.2 | 3.23.0 | ##### 移除的套件 | 套件名稱 | | --- | | kernel-livepatch-6.1.150-174.273 | | nvidia-fabric-manager | # 深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20250927 如需入門說明，請參閱 [DLAMI 入門](getting-started.md)。 ##### 主要套件版本 | 套件名稱 | 版本 | | --- | --- | | supported\$1ec2\$1instances | G4dn, G5, G6, Gr6, G6e, P4, P4de, P5, P5e, P5en | | operating\$1system | Amazon Linux 2023.8.20250915 | | compute\$1architecture | x86\$164 | | kernel\$1version | 6.1.150-174.273.amzn2023.x86\$164 | | framework\$1version | 2.5.1 | | python\$1location | /opt/conda/envs/pytorch/bin/python | | nvidia\$1driver | 570.172.08 | | default\$1cuda | /usr/local/cuda-12.4/ | | nvidia\$1cuda\$1stack | /usr/local/cuda-12.4 | | gdr\$1copy | 2.4.4 | | nvidia\$1container\$1toolkit\$1version | 1.17.8 | | efa\$1version | 1.43.1 | | ofi\$1nccl\$1version | 1.16.2 | | ebs\$1volume\$1type | gp3 | | ssm\$1agent\$1version | 3.3.2299.0 | ##### 先前版本的套件變更 **注意** 所有 DLAMI 都會依照 AWS 最佳實務掃描漏洞。雖然我們在每個版本中持續升級至最新的套件版本，但某些相依性可能需要特定的套件降級來維持相容性。這些調整不會影響我們的安全狀態。如需 AWS 安全措施的完整資訊，請參閱我們的[AWS 安全文件 ](https://docs.aws.amazon.com/dlami/latest/devguide/security.html) ##### 新增的套件 | | | --- | | kernel-livepatch-6.1.150-174.273-1.0-0.amzn2023 | ##### 更新套件 | 套件名稱 | 先前版本 | 新版本 | | --- | --- | --- | | Authlib | 1.6.3 | 1.6.4 | | PyYAML | 6.0.2 | 6.0.3 | | amazon-linux-repo-s3 | 2023.8.20250908-0.amzn2023 | 2023.8.20250915-0.amzn2023 | | anyio | 4.10.0 | 4.11.0 | | awscli | 2.30.1 | 2.31.3 | | boto3 | 1.40.30 | 1.40.40 | | botocore | 1.40.30 | 1.40.40 | | cffi | 1.17.1 | 2.0.0 | | click | 8.2.1 | 8.3.0 | | containerd | 2.0.5-1.amzn2023.0.2 | 2.0.6-1.amzn2023.0.1 | | cryptography | 45.0.7 | 46.0.1 | | debugpy | 1.8.16 | 1.8.17 | | docker | 25.0.8-1.amzn2023.0.5 | 25.0.8-1.amzn2023.0.6 | | fastapi | 0.116.1 | 0.117.1 | | fonttools | 4.59.2 | 4.60.0 | | glibc | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | glibc-all-langpacks | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | glibc-common | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | glibc-devel | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | glibc-gconv-extra | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | glibc-headers-x86 | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | glibc-locale-source | 2.34-196.amzn2023.0.1 | 2.34-231.amzn2023.0.1 | | gstreamer1-plugins-base | 1.24.10-1.amzn2023.0.1 | 1.24.10-1.amzn2023.0.2 | | gymnasium | 1.2.0 | 1.2.1 | | inspectorssmplugin | 1.0.395-1 | 1.0.398-1 | | jupyterlab | 4.4.7 | 4.4.9 | | kernel | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | kernel-devel | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | kernel-headers | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | kernel-libbpf | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | kernel-livepatch-repo-s3 | 2023.8.20250908-0.amzn2023 | 2023.8.20250915-0.amzn2023 | | kernel-modules-extra | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | kernel-modules-extra-common | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | kernel-tools | 6.1.148-173.267.amzn2023 | 6.1.150-174.273.amzn2023 | | lark | 1.2.2 | 1.3.0 | | libtiff | 4.4.0-4.amzn2023.0.21 | 4.4.0-4.amzn2023.0.22 | | libunwind | 1.4.0-5.amzn2023.0.2 | 1.4.0-5.amzn2023.0.3 | | llvmlite | 0.44.0 | 0.45.0 | | lustre-client | 2.15.6-19.amzn2023 | 2.15.6-21.amzn2023 | | nbclassic | 1.3.2 | 1.3.3 | | numba | 0.61.2 | 0.62.0 | | prometheus\$1client | 0.22.1 | 0.23.1 | | psutil | 7.0.0 | 7.1.0 | | pydantic | 2.9.2 | 2.11.9 | | pydantic\$1core | 2.23.4 | 2.33.2 | | pyparsing | 3.2.4 | 3.2.5 | | regex | 2025.9.1 | 2025.9.18 | | rust-toolset-srpm-macros | 1.89.0-1.amzn2023.0.2 | 1.89.0-1.amzn2023.0.3 | | safety-schemas | 0.0.14 | 0.0.16 | | starlette | 0.47.3 | 0.48.0 | | system-release | 2023.8.20250908-0.amzn2023 | 2023.8.20250915-0.amzn2023 | | typer | 0.17.4 | 0.19.2 | | typing\$1extensions | 4.15.0 | 4.12.2 | | uvicorn | 0.35.0 | 0.37.0 | | wcwidth | 0.2.13 | 0.2.14 | | zipp | 3.23.0 | 3.19.2 | ##### 移除的套件 | 套件名稱 | | --- | | kernel-livepatch-6.1.148-173.267 | #### 存檔 **Topics** + [版本備註封存](archive-detail-gpu-pytorch-2.5-amazon-linux-2023.md) # 版本備註封存 ##### 發行日期：2025-02-17 **AMI 名稱：**深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20250216 ##### Updated + 將 NVIDIA Container Toolkit 從 1.17.3 版更新至 1.17.4 版 + 如需詳細資訊，請參閱此處的版本備註頁面：https：//[https://github.com/NVIDIA/nvidia-container-toolkit/releases/tag/v1.17.4](https://github.com/NVIDIA/nvidia-container-toolkit/releases/tag/v1.17.4) + 在 Container Toolkit 1.17.4 版中，CUDA Compat 程式庫的掛載現已停用。為了確保與容器工作流程上的多個 CUDA 版本相容，請確保您更新 LD\$1LIBRARY\$1PATH 以包含 CUDA 相容性程式庫，如[如果您使用 CUDA 相容性層](https://docs.aws.amazon.com/sagemaker/latest/dg/inference-gpu-drivers.html#collapsible-cuda-compat)教學中所示。 ##### 已移除 + 已移除 [NVIDIA CUDA 工具組](https://docs.nvidia.com/cuda/)提供的使用者空間程式庫 cuobj 和 nvdisasm，以解決 [2025 年 2 月 18 日 NVIDIA CUDA Toolkit 安全公告](https://nvidia.custhelp.com/app/answers/detail/a_id/5594)中存在的 CVEs ##### 發行日期：2025-01-08 **AMI 名稱：**深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20250107 ##### 已新增 + 新增對 [G4dn 執行個體](https://aws.amazon.com/ec2/instance-types/g4/)的支援。 ##### 版本日期：2024-11-21 **AMI 名稱：**深度學習 OSS Nvidia 驅動程式 AMI GPU PyTorch 2.5.1 (Amazon Linux 2023) 20241120 ##### 已新增 + 適用於 Amazon Linux 2023 的 Deep Learning OSS Nvidia Driver AMI GPU PyTorch 2.5 初始版本 ##### 已知問題 + 此 DLAMI 目前不支援 G4dn 和 G5 EC2 執行個體。 AWS 知道可能導致 CUDA 初始化失敗的不相容，當使用開放原始碼 NVIDIA 驅動程式搭配 Linux 核心 6.1 版或更新版本時，這會影響 G4dn 和 G5 執行個體系列。此問題會影響 Linux 發行版本，例如 Amazon Linux 2023、Ubuntu 22.04 或更新版本，或 SUSE Linux Enterprise Server 15 SP6 或更新版本等。