Otá*_*lva 2 python conda tensorflow mamba mambaforge
我在启用的 WSL2 Ubuntu 22.04 上使用 mambaforge systemd。我尝试使用以下命令安装启用了 CUDA 的 TensorFlow 2.10:
mamba install tensorflow\nRun Code Online (Sandbox Code Playgroud)\nWSL2 的命令nvidia-smi -q给出:
==============NVSMI LOG==============\n\nTimestamp : Sat Dec 17 23:22:43 2022\nDriver Version : 527.56\nCUDA Version : 12.0\n\nAttached GPUs : 1\nGPU 00000000:01:00.0\n Product Name : NVIDIA GeForce RTX 3070 Laptop GPU\n Product Brand : GeForce\n Product Architecture : Ampere\n Display Mode : Disabled\n Display Active : Disabled\n Persistence Mode : Enabled\n MIG Mode\n Current : N/A\n Pending : N/A\n Accounting Mode : Disabled\n Accounting Mode Buffer Size : 4000\n Driver Model\n Current : WDDM\n Pending : WDDM\n Serial Number : N/A\n GPU UUID : GPU-f03a575d-7930-47f3-4965-290b89514ae7\n Minor Number : N/A\n VBIOS Version : 94.04.3f.00.d7\n MultiGPU Board : No\n Board ID : 0x100\n Board Part Number : N/A\n GPU Part Number : 249D-750-A1\n Module ID : 1\n Inforom Version\n Image Version : G001.0000.03.03\n OEM Object : 2.0\n ECC Object : N/A\n Power Management Object : N/A\n GPU Operation Mode\n Current : N/A\n Pending : N/A\n GSP Firmware Version : N/A\n GPU Virtualization Mode\n Virtualization Mode : None\n Host VGPU Mode : N/A\n IBMNPU\n Relaxed Ordering Mode : N/A\n PCI\n Bus : 0x01\n Device : 0x00\n Domain : 0x0000\n Device Id : 0x249D10DE\n Bus Id : 00000000:01:00.0\n Sub System Id : 0x118C1043\n GPU Link Info\n PCIe Generation\n Max : 3\n Current : 3\n Device Current : 3\n Device Max : 4\n Host Max : 3\n Link Width\n Max : 16x\n Current : 8x\n Bridge Chip\n Type : N/A\n Firmware : N/A\n Replays Since Reset : 0\n Replay Number Rollovers : 0\n Tx Throughput : 0 KB/s\n Rx Throughput : 0 KB/s\n Atomic Caps Inbound : N/A\n Atomic Caps Outbound : N/A\n Fan Speed : N/A\n Performance State : P8\n Clocks Throttle Reasons\n Idle : Active\n Applications Clocks Setting : Not Active\n SW Power Cap : Not Active\n HW Slowdown : Not Active\n HW Thermal Slowdown : Not Active\n HW Power Brake Slowdown : Not Active\n Sync Boost : Not Active\n SW Thermal Slowdown : Not Active\n Display Clock Setting : Not Active\n FB Memory Usage\n Total : 8192 MiB\n Reserved : 159 MiB\n Used : 12 MiB\n Free : 8020 MiB\n BAR1 Memory Usage\n Total : 8192 MiB\n Used : 1 MiB\n Free : 8191 MiB\n Compute Mode : Default\n Utilization\n Gpu : 0 %\n Memory : 0 %\n Encoder : 0 %\n Decoder : 0 %\n Encoder Stats\n Active Sessions : 0\n Average FPS : 0\n Average Latency : 0\n FBC Stats\n Active Sessions : 0\n Average FPS : 0\n Average Latency : 0\n Ecc Mode\n Current : N/A\n Pending : N/A\n ECC Errors\n Volatile\n SRAM Correctable : N/A\n SRAM Uncorrectable : N/A\n DRAM Correctable : N/A\n DRAM Uncorrectable : N/A\n Aggregate\n SRAM Correctable : N/A\n SRAM Uncorrectable : N/A\n DRAM Correctable : N/A\n DRAM Uncorrectable : N/A\n Retired Pages\n Single Bit ECC : N/A\n Double Bit ECC : N/A\n Pending Page Blacklist : N/A\n Remapped Rows : N/A\n Temperature\n GPU Current Temp : 46 C\n GPU Shutdown Temp : 101 C\n GPU Slowdown Temp : 98 C\n GPU Max Operating Temp : 87 C\n GPU Target Temperature : N/A\n Memory Current Temp : N/A\n Memory Max Operating Temp : N/A\n Power Readings\n Power Management : Supported\n Power Draw : 12.08 W\n Power Limit : 4294967.50 W\n Default Power Limit : 80.00 W\n Enforced Power Limit : 100.00 W\n Min Power Limit : 1.00 W\n Max Power Limit : 100.00 W\n Clocks\n Graphics : 210 MHz\n SM : 210 MHz\n Memory : 405 MHz\n Video : 555 MHz\n Applications Clocks\n Graphics : N/A\n Memory : N/A\n Default Applications Clocks\n Graphics : N/A\n Memory : N/A\n Deferred Clocks\n Memory : N/A\n Max Clocks\n Graphics : 2100 MHz\n SM : 2100 MHz\n Memory : 6001 MHz\n Video : 1950 MHz\n Max Customer Boost Clocks\n Graphics : N/A\n Clock Policy\n Auto Boost : N/A\n Auto Boost Default : N/A\n Voltage\n Graphics : 637.500 mV\n Fabric\n State : N/A\n Status : N/A\n Processes\n GPU instance ID : N/A\n Compute instance ID : N/A\n Process ID : 24\n Type : G\n Name : /Xwayland\n Used GPU Memory : Not available in WDDM driver model\n\nRun Code Online (Sandbox Code Playgroud)\n我的其他环境按预期工作:
\n\xe2\xac\xa2 [Systemd] \xe2\x9d\xaf mamba activate tf\n\n~ via tf via 774MiB/19GiB | 0B/5GiB\n\xe2\xac\xa2 [Systemd] \xe2\x9d\xaf python\nPython 3.9.15 | packaged by conda-forge | (main, Nov 22 2022, 08:45:29)\n[GCC 10.4.0] on linux\nType "help", "copyright", "credits" or "license" for more information.\n>>> import tensorflow as tf\n2022-12-17 23:25:13.867166: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: SSE4.1 SSE4.2 AVX AVX2 FMA\nTo enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\nRun Code Online (Sandbox Code Playgroud)\n然后,它尝试安装软件包版本cuda112py39h9333c2f_1,winch使用Python 3.9,但我想要Python 3.10。每当我尝试安装 3.10 版本时,都会显示错误:
Could not solve for environment specs\nEncountered problems while solving:\n - nothing provides __cuda needed by tensorflow-2.10.0-cuda112py310he87a039_0\n\nThe environment can't be solved, aborting the operation\nRun Code Online (Sandbox Code Playgroud)\n为什么会出现此错误以及如何解决它?
\n我今天遇到了这个问题,找到了一个有效的解决方案(在看到你的 GitHub 帖子之后)。长话短说,您需要按照conda-forge 博客文章CONDA_OVERRIDE_CUDA中的描述来完成这项工作。
例如,对于 CUDA 11.8 和 mamba,请使用:
CONDA_OVERRIDE_CUDA="11.8" mamba install tensorflow -c conda-forge
对于 CUDA 11.8 和 conda,它将是:
CONDA_OVERRIDE_CUDA="11.8" conda install tensorflow -c conda-forge
根据您的设置,您可能还想安装 cudatoolkit,例如,
CONDA_OVERRIDE_CUDA="11.8" mamba install tensorflow cudatoolkit -c conda-forge
编辑:根据有用的评论修复了命令!
| 归档时间: |
|
| 查看次数: |
1877 次 |
| 最近记录: |