An open-source project dedicated to tracking and segmenting any objects in videos, either automatically or interactively. The primary algorithms utilized include the Segment Anything Model (SAM) for key-frame segmentation and Associating Objects with Transformers (AOT) for efficient tracking and propagation purposes.
GNU Affero General Public License v3.0
2.77k
stars
334
forks
source link
RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED #76
File "/home/xxx/文档/github/Segment-and-Track-Anything-main/sam/segment_anything/modeling/transformer.py", line 231, in forward
attn = q @ k.permute(0, 1, 3, 2) # B x N_heads x N_tokens x N_tokens
RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling cublasGemmStridedBatchedExFix( handle, opa, opb, m, n, k, (void*)(&falpha), a, CUDA_R_16F, lda, stridea, b, CUDA_R_16F, ldb, strideb, (void*)(&fbeta), c, CUDA_R_16F, ldc, stridec, num_batches, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
Environment:
1*RTX4090(24GB)
Ubuntu 22.04
CUDA 11.7
torch 1.13.0+cu117
Description:
First I ran the demo.ipynb and got the error:
File "/home/xxx/文档/github/Segment-and-Track-Anything-main/sam/segment_anything/utils/amg.py", line 53, in filter
RuntimeError: CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
Then I tried adding a line to debug:
os.environ['CUDA_LAUNCH_BLOCKING'] = '1'
At the beginning of demo.ipynb and got the error:
File "/home/xxx/文档/github/Segment-and-Track-Anything-main/sam/segment_anything/modeling/transformer.py", line 231, in forward
attn = q @ k.permute(0, 1, 3, 2) # B x N_heads x N_tokens x N_tokens
RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling cublasGemmStridedBatchedExFix( handle, opa, opb, m, n, k, (void*)(&falpha), a, CUDA_R_16F, lda, stridea, b, CUDA_R_16F, ldb, strideb, (void*)(&fbeta), c, CUDA_R_16F, ldc, stridec, num_batches, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
After I ran the program, I couldn't kill the process, and it seemed that the resource on GPU could not be release. Also my PC could not run as usual, and could not even restart normally.
I would appreciate it much if you could help me with the problem!
Error Info:
File "/home/xxx/文档/github/Segment-and-Track-Anything-main/sam/segment_anything/modeling/transformer.py", line 231, in forward attn = q @ k.permute(0, 1, 3, 2) # B x N_heads x N_tokens x N_tokens RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling
cublasGemmStridedBatchedExFix( handle, opa, opb, m, n, k, (void*)(&falpha), a, CUDA_R_16F, lda, stridea, b, CUDA_R_16F, ldb, strideb, (void*)(&fbeta), c, CUDA_R_16F, ldc, stridec, num_batches, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
Environment:
1*RTX4090(24GB) Ubuntu 22.04 CUDA 11.7 torch 1.13.0+cu117
Description:
First I ran the demo.ipynb and got the error: File "/home/xxx/文档/github/Segment-and-Track-Anything-main/sam/segment_anything/utils/amg.py", line 53, in filter RuntimeError: CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
Then I tried adding a line to debug:
At the beginning of demo.ipynb and got the error: File "/home/xxx/文档/github/Segment-and-Track-Anything-main/sam/segment_anything/modeling/transformer.py", line 231, in forward attn = q @ k.permute(0, 1, 3, 2) # B x N_heads x N_tokens x N_tokens RuntimeError: CUDA error: CUBLAS_STATUS_EXECUTION_FAILED when calling
cublasGemmStridedBatchedExFix( handle, opa, opb, m, n, k, (void*)(&falpha), a, CUDA_R_16F, lda, stridea, b, CUDA_R_16F, ldb, strideb, (void*)(&fbeta), c, CUDA_R_16F, ldc, stridec, num_batches, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
After I ran the program, I couldn't kill the process, and it seemed that the resource on GPU could not be release. Also my PC could not run as usual, and could not even restart normally.
I would appreciate it much if you could help me with the problem!