search

SuperMedIntel / Medical-SAM2

Medical SAM 2: Segment Medical Images As Video Via Segment Anything Model 2
Apache License 2.0
488 stars 62 forks source link

version `GLIBC_2.32' #53

Open Meaw0415 opened 1 month ago

Meaw0415 commented 1 month ago

python train_3d.py -net sam2 -exp_name BTCV_MedSAM2 -sam_ckpt ./checkpoints/sam2_hiera_small.pt -sam_config sam2_hiera_s -image_size 1024 -val_freq 1 -prompt bbox -prompt_freq 2 -dataset btcv -data_path ./data/btcv

ImportError: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.32' not found (required by /mnt/iMVR/zhengf/Medical-SAM2/sam2_train/_C.so)

DISAPPEARED13 commented 1 month ago

I've met the same questions like you do. I check the closed issue and found https://github.com/MedicineToken/Medical-SAM2/issues/25, seems need to update our GLIBC version, but I have some questions, will update GLIBC cause some new bugs for incompatibilities? or could I just reduce the limitation of the GLIBC version? I just want to limit the environment in this project and do no changes to the whole environment.

DISAPPEARED13 commented 1 month ago

PS: when I run train_2d.py, this problem doesn't happened, but just for 3d. why? great thanks.

cam-qf commented 1 month ago

PS: when I run train_2d.py, this problem doesn't happened, but just for 3d. why? great thanks.

Same as your situation, there is no problem when running train_2d.py, but no version "Glibc_2.32" occurs at 3D. After updating the Glibc version

Runtimeerror: CUDA ERROR: No Kernel Image is available for execution on the device

Compile with torch_use_cuda_dsa to enable device-size assereles.

Do you solve it?

sahikabetul commented 3 weeks ago

PS: when I run train_2d.py, this problem doesn't happened, but just for 3d. why? great thanks.

Same as your situation, there is no problem when running train_2d.py, but no version "Glibc_2.32" occurs at 3D. After updating the Glibc version

Runtimeerror: CUDA ERROR: No Kernel Image is available for execution on the device

Compile with torch_use_cuda_dsa to enable device-size assereles.

Do you solve it?

I have exactly same problem. I set up a Docker image with pytorch 2.4, ubuntu 22.04, cuda 12.1 and GLIB problem solved. And I am getting second error, too.

cam-qf commented 3 weeks ago

PS: when I run train_2d.py, this problem doesn't happened, but just for 3d. why? great thanks.

Same as your situation, there is no problem when running train_2d.py, but no version "Glibc_2.32" occurs at 3D. After updating the Glibc version Runtimeerror: CUDA ERROR: No Kernel Image is available for execution on the device Compile with torch_use_cuda_dsa to enable device-size assereles. Do you solve it?

I have exactly same problem. I set up a Docker image with pytorch 2.4, ubuntu 22.04, cuda 12.1 and GLIB problem solved. And I am getting second error, too.

May I ask what graphics card you are using?

sahikabetul commented 3 weeks ago

Hi, My graphics card informations:

DISAPPEARED13 commented 3 weeks ago

Hi, My graphics card informations:

  • Model: NVIDIA A100-SXM4-80GB
  • Driver Version: 560.35.03
  • CUDA Version: 12.6
  • Memory: Each GPU has 80 GB of memory (81920 MiB)
  • Number of GPUs: 4

Congrates for solving the problem!! May I ask that your GLIBC version? That's quite wired, seems some system installation conflicts with the project.