OOM in evaluation

[Luo-Z13]

Environment ( I have tried pytorch 1.9.0 and pytorch 1.8.0 and their corresponding cudnn/ mmcv-full =1.14.0, both will cause memory increasing when training.)

sys.platform: linux
Python: 3.8.17 (default, Jul 5 2023, 21:04:15) [GCC 11.2.0]
CUDA available: True
GPU 0: NVIDIA TITAN RTX
CUDA_HOME: /usr/local/cuda-10.2
NVCC: Cuda compilation tools, release 10.2, V10.2.89
GCC: gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
PyTorch: 1.10.0
PyTorch compiling details: PyTorch built with:

• GCC 7.3
• C++ Version: 201402
• Intel(R) oneAPI Math Kernel Library Version 2023.1-Product Build 20230303 for Intel(R) 64 architecture applications
• Intel(R) MKL-DNN v2.2.3 (Git Hash 7336ca9f055cf1bfa13efb658fe15dc9b41f0740)
• OpenMP 201511 (a.k.a. OpenMP 4.5)
• LAPACK is enabled (usually provided by MKL)
• NNPACK is enabled
• CPU capability usage: AVX512
• CUDA Runtime 11.3
• NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_37,code=compute_37
• CuDNN 8.2
• Magma 2.5.2
• Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.3, CUDNN_VERSION=8.2.0, CXX_COMPILER=/opt/rh/devtoolset-7/root/usr/bin/c++, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-unused-local-typedefs -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.10.0, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON,

TorchVision: 0.11.1
OpenCV: 4.7.0
MMCV: 1.4.0
MMCV Compiler: GCC 7.3
MMCV CUDA Compiler: 11.3
MMDetection: 2.13.0+c820f32

2023-09-05 21:09:00,423 - mmdet - INFO - Distributed training: False
2023-09-05 21:09:01,221 - mmdet - INFO - Config:
checkpoint_config = dict(interval=1)
log_config = dict(interval=50, hooks=[dict(type='TextLoggerHook')])
custom_hooks = [dict(type='NumClassCheckHook')]
dist_params = dict(backend='nccl')
log_level = 'INFO'
load_from = None
resume_from = None
workflow = [('train', 1)]
norm_cfg = dict(type='GN', num_groups=32, requires_grad=True)

Bug Description

[>>> ] 426/3822, 16.0 task/s, elapsed: 27s, ETA: 213s
[>>> ] 427/3822, 15.9 task/s, elapsed: 27s, ETA: 214s
[>>> ] 428/3822, 15.9 task/s, elapsed: 27s, ETA: 213s
[>>> ] 429/3822, 15.9 task/s, elapsed: 27s, ETA: 213s
[>>> ] 430/3822, 15.9 task/s, elapsed: 27s, ETA: 213s
[>>> ] 431/3822, 15.9 task/s, elapsed: 27s, ETA: 214s
[>>> ] 432/3822, 15.9 task/s, elapsed: 27s, ETA: 213s
[>>> ] 433/3822, 15.9 task/s, elapsed: 27s, ETA: 213s
[>>> ] 434/3822, 15.9 task/s, elapsed: 27s, ETA: 213sTraceback (most recent call last):
File "tools/train.py", line 192, in
main()
File "tools/train.py", line 181, in main
train_detector(
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/apis/train.py", line 172, in train_detector
runner.run(data_loaders, cfg.workflow)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 127, in run
epoch_runner(data_loaders[i], **kwargs)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 54, in train
self.call_hook('after_train_epoch')
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/runner/base_runner.py", line 307, in call_hook
getattr(hook, fn_name)(self)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/runner/hooks/evaluation.py", line 267, in after_train_epoch
self._do_evaluate(runner)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/core/evaluation/eval_hooks.py", line 44, in _do_evaluate
results = single_gpu_test(runner.model, self.dataloader, show=False)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/apis/test.py", line 27, in single_gpu_test
result = model(return_loss=False, rescale=True, **data)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/parallel/data_parallel.py", line 50, in forward
return super().forward(*inputs, **kwargs)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/torch/nn/parallel/data_parallel.py", line 166, in forward
return self.module(*inputs[0], **kwargs[0])
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/runner/fp16_utils.py", line 98, in new_func
return old_func(*args, **kwargs)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/detectors/base.py", line 177, in forward
return self.forward_test(img, img_metas, **kwargs)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/detectors/base.py", line 150, in forward_test
return self.simple_test(imgs[0], img_metas[0], **kwargs)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/detectors/P2BNet.py", line 375, in simple_test
test_result, pseudo_boxes = self.roi_head.simple_test(stage,
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/roi_heads/P2B_head.py", line 446, in simple_test
det_bboxes, det_labels, pseudo_bboxes = self.simple_test_bboxes(
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/roi_heads/P2B_head.py", line 474, in simple_test_bboxes
bbox_results = self._bbox_forward(x, rois, gt_bboxes, stage)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/roi_heads/P2B_head.py", line 210, in _bbox_forward
bbox_feats = self.bbox_roi_extractor(
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in call_impl
return forward_call(*input, **kwargs)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/runner/fp16_utils.py", line 186, in new_func
return old_func(*args, **kwargs)
File "/media/dell/data1/ljw/code/test1/CPR/P2BNet/TOV_mmdetection/mmdet/models/roi_heads/roi_extractors/single_level_roi_extractor.py", line 102, in forward
roi_feats_t = self.roi_layers[i](feats[i], rois)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/ops/roi_align.py", line 212, in forward
return roi_align(input, rois, self.output_size, self.spatial_scale,
File "/media/dell/data1/miniconda3/envs/mmdetp2b/lib/python3.8/site-packages/mmcv/ops/roi_align.py", line 84, in forward
output = input.new_zeros(output_shape)
RuntimeError: CUDA out of memory. Tried to allocate 2.28 GiB (GPU 0; 23.65 GiB total capacity; 14.99 GiB already allocated; 1.28 GiB free; 21.04 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

[Luo-Z13]

@Darren-pfchen Could you give some advice about controlling the number of rois please?