Compile bug: nvcc fatal : Unsupported gpu architecture 'compute_'

[pockers21]

Git commit

dd8ba93

Operating systems

Linux

GGML backends

CUDA

Problem description & steps to reproduce

CUDA Build Failure on RTX 3090 on auto-dl with Latest Master Branch

Problem Description

CUDA build fails when running the CI script on the latest master branch with RTX 3090 GPU.

Environment

• GPU: NVIDIA RTX 3090
• Commit: dd8ba93416f492c168f7e20f0b1434c08ebeaeb5 (tag: b5533)
• Branch: master
• Build Command: GG_BUILD_CUDA=1 bash ./ci/run.sh ./tmp/results ./tmp/mnt

Steps to Reproduce

1. Clone the latest master branch
2. Checkout commit dd8ba93416f492c168f7e20f0b1434c08ebeaeb5
3. Run: GG_BUILD_CUDA=1 bash ./ci/run.sh ./tmp/results ./tmp/mnt
4. Build fails

Expected Behavior

The CUDA build should complete successfully without errors.

Actual Behavior

Build fails with errors .

System Information

nvidia-smi
Thu May 29 21:55:41 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.54.14              Driver Version: 550.54.14      CUDA Version: 12.4     |
|-----------------------------------------+------------------------+----------------------+
| 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 RTX 3090        On  |   00000000:02:00.0 Off |                  N/A |
|  0%   24C    P8             28W /  350W |       0MiB /  24576MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:33:58_PDT_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0

gcc --version
gcc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

cmake --version
cmake version 3.22.1

CMake suite maintained and supported by Kitware (kitware.com/cmake).


uname -a
Linux autodl-container-e179408ca5-c46e6369 5.15.0-101-generic #111-Ubuntu SMP Tue Mar 5 20:16:58 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux

First Bad Commit

No response

Compile command

GG_BUILD_CUDA=1 bash ./ci/run.sh ./tmp/results ./tmp/mnt

Relevant log output

CMake Error at /usr/share/cmake-3.22/Modules/CMakeDetermineCompilerId.cmake:726 (message):
  Compiling the CUDA compiler identification source file
  "CMakeCUDACompilerId.cu" failed.

  Compiler: /usr/local/cuda/bin/nvcc

  Build flags:

  Id flags: --keep;--keep-dir;tmp;-gencode=arch=compute_,code=sm_ -v



  The output was:

  1

  nvcc fatal : Unsupported gpu architecture 'compute_'

[pockers21]

While remove native in ci/run.sh and hard code like

if [ ! -z ${GG_BUILD_CUDA} ]; then
    CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=86"

problem fixed
perhaps this machine is just a virtual machine in auto-dl?
But l have committed pr in auto-dl before, ci works well.

[pockers21]

if [ ! -z ${GG_BUILD_CUDA} ]; then
if command -v nvidia-smi >/dev/null 2>&1; then
CUDA_ARCH=$(nvidia-smi --query-gpu=compute_cap --format=csv,noheader,nounits 2>/dev/null | head -1 | tr -d '.')
if [[ -n "$CUDA_ARCH" && "$CUDA_ARCH" =~ ^[0-9]+$ ]]; then
CMAKE_EXTRA="${CMAKE_EXTRA} -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=${CUDA_ARCH}"
fi
fi
fi perhaps solved this bug , l will check it tomorrow.

[LostRuins]

Does it look like this?

[pockers21]

Does it look like this?

no, program just prints nvcc fatal : Unsupported gpu architecture 'compute_' and exit