add benchmark for maxout/put_along_axis/take_along_axis #1326

m3ngyang · 2022-03-01T09:20:35Z

add benchmark for maxout/put_along_axis/take_along_axis
测试结果已添加。

paddle-bot-old · 2022-03-01T09:20:55Z

✅ This PR's description meets the template requirements!
Please wait for other CI results.

paddle-bot-old · 2022-03-01T09:21:02Z

Thanks for your contribution! Please wait for the result of CI firstly.

CLAassistant · 2022-03-01T09:37:05Z

All committers have signed the CLA.

m3ngyang · 2022-03-04T04:12:39Z

本地测试，cpu 环境

(paddle) root@94866e7b8d4a:/home/benchmark/api/dynamic_tests_v2# bash run.sh maxout 0 speed
run.sh: line 6: [: !=: unary operator expected
-- Current directory: /home/benchmark/api/dynamic_tests_v2
-- Entering /home/benchmark/api/common
-- Current directory: /home/benchmark/api/common
run command: git rev-parse HEAD
run command: git show -s --format=%ad
-- Current directory: /home/benchmark/api/dynamic_tests_v2
===========================================================================
-- paddle version             : 0.0.0
-- paddle commit              : eaacf8bfee5c9583f7ebf0deff20b90db9d73478
-- benchmark commit           : 425703cc536122f96fb375c150699c85c47d1f9f
-- benchmark last update time : Tue Mar 1 17:20:00 2022 +0800
===========================================================================
run command: /usr/local/share/miniconda3/envs/paddle/bin/python /home/benchmark/api/dynamic_tests_v2/maxout.py --task speed --framework paddle --testing_mode dynamic --json_file /home/benchmark/api/tests_v2/configs/maxout.json --config_id 0 --profiler none --backward True --use_gpu False --repeat 1000 --allow_adaptive_repeat False --log_level 0
Cannot import pytorch, maybe paddle is not installed.
Cannot import pytorch, maybe pytorch is not installed.
Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=0, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=0, json_file='/home/benchmark/api/tests_v2/configs/maxout.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=False)
---- Initialize APIConfig from /home/benchmark/api/tests_v2/configs/maxout.json, config_id = 0.

Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=0, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=0, json_file='/home/benchmark/api/tests_v2/configs/maxout.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=False)
[paddle][maxout] maxout {
  run_tf: True
  run_torch: False
  repeat: 5000
  groups: 2
  axis: -1
  x_shape: [32, 12, 128, 128]
  x_dtype: float32
  atol: 1e-06
}
{"framework": "paddle", "version": "0.0.0", "name": "maxout", "device": "CPU", "backward": true, "speed": {"repeat": 1000, "begin": 10, "end": 990, "total": 67.3420314886132, "wall_time": 0, "total_include_wall_time": 67.3420314886132, "gpu_time": 0.0}, "parameters": "x (Variable) - dtype: float32, shape: [32, 12, 128, 128]\naxis (int): -1\ngroups (int): 2\n"}

(paddle) root@94866e7b8d4a:/home/benchmark/api/dynamic_tests_v2# bash run.sh take_along_axis 0 speed
run.sh: line 6: [: !=: unary operator expected
-- Current directory: /home/benchmark/api/dynamic_tests_v2
-- Entering /home/benchmark/api/common
-- Current directory: /home/benchmark/api/common
run command: git rev-parse HEAD
run command: git show -s --format=%ad
-- Current directory: /home/benchmark/api/dynamic_tests_v2
===========================================================================
-- paddle version             : 0.0.0
-- paddle commit              : eaacf8bfee5c9583f7ebf0deff20b90db9d73478
-- benchmark commit           : 6085eff3f1646448e431d037dbee1a1e0a231f9e
-- benchmark last update time : Tue Mar 1 17:20:00 2022 +0800
===========================================================================
run command: /usr/local/share/miniconda3/envs/paddle/bin/python /home/benchmark/api/dynamic_tests_v2/take_along_axis.py --task speed --framework paddle --testing_mode dynamic --json_file /home/benchmark/api/tests_v2/configs/take_along_axis.json --config_id 0 --profiler none --backward True --use_gpu False --repeat 1000 --allow_adaptive_repeat False --log_level 0
Cannot import pytorch, maybe paddle is not installed.
Cannot import pytorch, maybe pytorch is not installed.
Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=0, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=0, json_file='/home/benchmark/api/tests_v2/configs/take_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=False)
---- Initialize APIConfig from /home/benchmark/api/tests_v2/configs/take_along_axis.json, config_id = 0.

Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=0, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=0, json_file='/home/benchmark/api/tests_v2/configs/take_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=False)
[paddle][take_along_axis] take_along_axis {
  run_tf: True
  run_torch: False
  axis: 0
  arr_shape: [200, 300]
  arr_dtype: float32
  indices_shape: [1, 1]
  indices_dtype: int
  atol: 1e-06
}
{"framework": "paddle", "version": "0.0.0", "name": "take_along_axis", "device": "CPU", "backward": true, "speed": {"repeat": 1000, "begin": 10, "end": 990, "total": 0.23803297354250538, "wall_time": 0, "total_include_wall_time": 0.23803297354250538, "gpu_time": 0.0}, "parameters": "arr (Variable) - dtype: float32, shape: [200, 300]\nindices (Variable) - dtype: int, shape: [1, 1]\naxis (int): 0\n"}

(paddle) root@94866e7b8d4a:/home/benchmark/api/dynamic_tests_v2# bash run.sh put_along_axis 0 speed
run.sh: line 6: [: !=: unary operator expected
-- Current directory: /home/benchmark/api/dynamic_tests_v2
-- Entering /home/benchmark/api/common
-- Current directory: /home/benchmark/api/common
run command: git rev-parse HEAD
run command: git show -s --format=%ad
-- Current directory: /home/benchmark/api/dynamic_tests_v2
===========================================================================
-- paddle version             : 0.0.0
-- paddle commit              : eaacf8bfee5c9583f7ebf0deff20b90db9d73478
-- benchmark commit           : 6085eff3f1646448e431d037dbee1a1e0a231f9e
-- benchmark last update time : Tue Mar 1 17:20:00 2022 +0800
===========================================================================
run command: /usr/local/share/miniconda3/envs/paddle/bin/python /home/benchmark/api/dynamic_tests_v2/put_along_axis.py --task speed --framework paddle --testing_mode dynamic --json_file /home/benchmark/api/tests_v2/configs/put_along_axis.json --config_id 0 --profiler none --backward True --use_gpu False --repeat 1000 --allow_adaptive_repeat False --log_level 0
Cannot import pytorch, maybe paddle is not installed.
Cannot import pytorch, maybe pytorch is not installed.
Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=0, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=0, json_file='/home/benchmark/api/tests_v2/configs/put_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=False)
---- Initialize APIConfig from /home/benchmark/api/tests_v2/configs/put_along_axis.json, config_id = 0.

Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=0, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=0, json_file='/home/benchmark/api/tests_v2/configs/put_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=False)
[paddle][put_along_axis] put_along_axis {
  run_tf: True
  run_torch: False
  values: 99.0
  axis: 0
  arr_shape: [200, 300]
  arr_dtype: float32
  indices_shape: [1, 1]
  indices_dtype: int
  atol: 1e-06
}
{"framework": "paddle", "version": "0.0.0", "name": "put_along_axis", "device": "CPU", "backward": true, "speed": {"repeat": 1000, "begin": 10, "end": 990, "total": 0.3393506517215651, "wall_time": 0, "total_include_wall_time": 0.3393506517215651, "gpu_time": 0.0}, "parameters": "arr (Variable) - dtype: float32, shape: [200, 300]\nindices (Variable) - dtype: int, shape: [1, 1]\naxis (int): 0\nvalues (float32): 99.0\n"}

m3ngyang · 2022-03-04T06:58:10Z

gpu 测试：

maxout

(base) λ 48d12ee09b84 /home/paddle/benchmark/api/dynamic_tests_v2 {add_ops} bash run.sh maxout 1 speed
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
-- Current directory: /home/paddle/benchmark/api/dynamic_tests_v2
-- Entering /home/paddle/benchmark/api/common
-- Current directory: /home/paddle/benchmark/api/common
run command: git rev-parse HEAD
run command: git show -s --format=%ad
-- Current directory: /home/paddle/benchmark/api/dynamic_tests_v2
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
===========================================================================
-- paddle version             : 0.0.0
-- paddle commit              : eaacf8bfee5c9583f7ebf0deff20b90db9d73478
-- benchmark commit           : 69b086c4ed78b3425955e9f3d683453e94a65598
-- benchmark last update time : Tue Mar 1 17:20:00 2022 +0800
===========================================================================
run command: nvidia-smi -L
run command: nvprof --profile-from-start off /root/miniconda3/bin/python /home/paddle/benchmark/api/dynamic_tests_v2/maxout.py --task speed --framework paddle --testing_mode dynamic --json_file /home/paddle/benchmark/api/tests_v2/configs/maxout.json --config_id 1 --profiler nvprof --backward True --use_gpu True --repeat 1000 --allow_adaptive_repeat False --log_level 0
            Type  Time(%)      Time     Calls       Avg       Min       Max  Name
 GPU activities:   60.71%  15.7347s      1000  15.735ms  15.561ms  15.750ms  void paddle::operators::math::KernelMaxoutGrad<float>(int, float const *, float const , float const , paddle::operators::math::KernelMaxoutGrad<float>*, int, int, int, int, int)
                   22.63%  5.86613s      1000  5.8661ms  5.8611ms  5.8945ms  void paddle::operators::math::KernelMaxOut<float>(int, float const *, int, int, int, int, int, paddle::operators::math::KernelMaxOut<float>*)
                   16.66%  4.31877s      1000  4.3188ms  4.2655ms  4.5649ms  void Eigen::internal::EigenMetaKernel<Eigen::TensorEvaluator<Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<float, int=1, int=1, long>, int=0, Eigen::MakePointer>, Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<float>, Eigen::TensorMap<Eigen::Tensor<float, int=1, int=1, long>, int=0, Eigen::MakePointer> const > const > const , Eigen::GpuDevice>, long>(float, int=1)

total gpu_time: 25917.8060 ms

run command: /root/miniconda3/bin/python /home/paddle/benchmark/api/dynamic_tests_v2/maxout.py --task speed --framework paddle --testing_mode dynamic --json_file /home/paddle/benchmark/api/tests_v2/configs/maxout.json --config_id 1 --profiler none --backward True --use_gpu True --repeat 1000 --allow_adaptive_repeat False --log_level 0  --gpu_time  25917.805962773844
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
W0304 08:43:11.716053   619 gpu_context.cc:240] Please NOTE: device: 0, GPU Compute Capability: 7.0, Driver API Version: 10.1, Runtime API Version: 10.1
W0304 08:43:11.721065   619 gpu_context.cc:268] device: 0, cuDNN Version: 7.6.
Cannot import pytorch, maybe paddle is not installed.
Cannot import pytorch, maybe pytorch is not installed.
Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=1, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=25917.805962773844, json_file='/home/paddle/benchmark/api/tests_v2/configs/maxout.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=True)
---- Initialize APIConfig from /home/paddle/benchmark/api/tests_v2/configs/maxout.json, config_id = 1.

Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=1, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=25917.805962773844, json_file='/home/paddle/benchmark/api/tests_v2/configs/maxout.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=True)
[paddle][maxout] maxout {
  run_tf: True
  run_torch: False
  repeat: 5000
  groups: 2
  axis: -1
  x_shape: [32, 8, 1024, 3072]
  x_dtype: float32
  atol: 1e-06
}
{"framework": "paddle", "version": "0.0.0", "name": "maxout", "device": "GPU", "backward": true, "speed": {"repeat": 1000, "begin": 10, "end": 990, "total": 25.951521980519196, "wall_time": 0, "total_include_wall_time": 25.951521980519196, "gpu_time": 25.917805962773844}, "parameters": "x (Variable) - dtype: float32, shape: [32, 8, 1024, 3072]\naxis (int): -1\ngroups (int): 2\n"}

take_along_axis

(base) λ 48d12ee09b84 /home/paddle/benchmark/api/dynamic_tests_v2 {add_ops} bash run.sh take_along_axis 1 speed
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
-- Current directory: /home/paddle/benchmark/api/dynamic_tests_v2
-- Entering /home/paddle/benchmark/api/common
-- Current directory: /home/paddle/benchmark/api/common
run command: git rev-parse HEAD
run command: git show -s --format=%ad
-- Current directory: /home/paddle/benchmark/api/dynamic_tests_v2
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
===========================================================================
-- paddle version             : 0.0.0
-- paddle commit              : eaacf8bfee5c9583f7ebf0deff20b90db9d73478
-- benchmark commit           : 69b086c4ed78b3425955e9f3d683453e94a65598
-- benchmark last update time : Tue Mar 1 17:20:00 2022 +0800
===========================================================================
run command: nvidia-smi -L
run command: nvprof --profile-from-start off /root/miniconda3/bin/python /home/paddle/benchmark/api/dynamic_tests_v2/take_along_axis.py --task speed --framework paddle --testing_mode dynamic --json_file /home/paddle/benchmark/api/tests_v2/configs/take_along_axis.json --config_id 1 --profiler nvprof --backward True --use_gpu True --repeat 1000 --allow_adaptive_repeat False --log_level 0
            Type  Time(%)      Time     Calls       Avg       Min       Max  Name
 GPU activities:   23.40%  34.194ms      1000  34.193us  33.472us  35.712us  [CUDA memcpy DtoD]
                   22.49%  32.875ms      1000  32.874us  32.160us  35.104us  void Eigen::internal::EigenMetaKernel<Eigen::TensorEvaluator<Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<float, int=2, int=1, int>, int=16, Eigen::MakePointer>, Eigen::TensorBroadcastingOp<Eigen::DSizes<long, int=2> const , Eigen::TensorMap<Eigen::Tensor<float const , int=2, int=1, int>, int=16, Eigen::MakePointer> const > const > const , Eigen::GpuDevice>, int>(float, int=2)
                   21.11%  30.856ms      1000  30.856us  29.120us  32.320us  void paddle::operators::GatherScatterGPUKernel<float, long, paddle::operators::TensorAssign, bool=0>(float*, int, long const *, float, long, int, int, long, long, paddle::operators::TensorAssign const &)
                   21.10%  30.843ms      1000  30.843us  29.120us  31.712us  void paddle::operators::GatherScatterGPUKernel<float, long, paddle::operators::ReduceAdd, bool=1>(float*, int, long const *, float, long, int, int, long, long, paddle::operators::ReduceAdd const &)
                   10.28%  15.029ms      1000  15.028us  14.688us  15.552us  void Eigen::internal::EigenMetaKernel<Eigen::TensorEvaluator<Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<float, int=1, int=1, long>, int=0, Eigen::MakePointer>, Eigen::TensorCwiseNullaryOp<Eigen::internal::scalar_constant_op<float>, Eigen::TensorMap<Eigen::Tensor<float, int=1, int=1, long>, int=0, Eigen::MakePointer> const > const > const , Eigen::GpuDevice>, long>(float, int=1)
                    1.61%  2.3474ms      1000  2.3470us  2.1760us  2.8800us  void Eigen::internal::EigenMetaKernel<Eigen::TensorEvaluator<Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<long, int=2, int=1, int>, int=16, Eigen::MakePointer>, Eigen::TensorBroadcastingOp<Eigen::DSizes<long, int=2> const , Eigen::TensorMap<Eigen::Tensor<long const , int=2, int=1, int>, int=16, Eigen::MakePointer> const > const > const , Eigen::GpuDevice>, int>(long, int=2)

total gpu_time: 146.1282 ms

run command: /root/miniconda3/bin/python /home/paddle/benchmark/api/dynamic_tests_v2/take_along_axis.py --task speed --framework paddle --testing_mode dynamic --json_file /home/paddle/benchmark/api/tests_v2/configs/take_along_axis.json --config_id 1 --profiler none --backward True --use_gpu True --repeat 1000 --allow_adaptive_repeat False --log_level 0  --gpu_time  146.12820512820514
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
W0304 08:45:14.841684   706 gpu_context.cc:240] Please NOTE: device: 0, GPU Compute Capability: 7.0, Driver API Version: 10.1, Runtime API Version: 10.1
W0304 08:45:14.846143   706 gpu_context.cc:268] device: 0, cuDNN Version: 7.6.
Cannot import pytorch, maybe paddle is not installed.
Cannot import pytorch, maybe pytorch is not installed.
Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=1, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=146.12820512820514, json_file='/home/paddle/benchmark/api/tests_v2/configs/take_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=True)
---- Initialize APIConfig from /home/paddle/benchmark/api/tests_v2/configs/take_along_axis.json, config_id = 1.

Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=1, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=146.12820512820514, json_file='/home/paddle/benchmark/api/tests_v2/configs/take_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=True)
[paddle][take_along_axis] take_along_axis {
  run_tf: True
  run_torch: False
  axis: 0
  arr_shape: [1024, 3072]
  arr_dtype: float32
  indices_shape: [1, 1]
  indices_dtype: int
  atol: 1e-06
}
{"framework": "paddle", "version": "0.0.0", "name": "take_along_axis", "device": "GPU", "backward": true, "speed": {"repeat": 1000, "begin": 10, "end": 990, "total": 0.2995177191130969, "wall_time": 0, "total_include_wall_time": 0.2995177191130969, "gpu_time": 0.14612820512820515}, "parameters": "arr (Variable) - dtype: float32, shape: [1024, 3072]\nindices (Variable) - dtype: int, shape: [1, 1]\naxis (int): 0\n"}

put_along_axis

(base) λ 48d12ee09b84 /home/paddle/benchmark/api/dynamic_tests_v2 {add_ops} bash run.sh put_along_axis 1 speed
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
-- Current directory: /home/paddle/benchmark/api/dynamic_tests_v2
-- Entering /home/paddle/benchmark/api/common
-- Current directory: /home/paddle/benchmark/api/common
run command: git rev-parse HEAD
run command: git show -s --format=%ad
-- Current directory: /home/paddle/benchmark/api/dynamic_tests_v2
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
===========================================================================
-- paddle version             : 0.0.0
-- paddle commit              : eaacf8bfee5c9583f7ebf0deff20b90db9d73478
-- benchmark commit           : 69b086c4ed78b3425955e9f3d683453e94a65598
-- benchmark last update time : Tue Mar 1 17:20:00 2022 +0800
===========================================================================
run command: nvidia-smi -L
run command: nvprof --profile-from-start off /root/miniconda3/bin/python /home/paddle/benchmark/api/dynamic_tests_v2/put_along_axis.py --task speed --framework paddle --testing_mode dynamic --json_file /home/paddle/benchmark/api/tests_v2/configs/put_along_axis.json --config_id 1 --profiler nvprof --backward True --use_gpu True --repeat 1000 --allow_adaptive_repeat False --log_level 0
            Type  Time(%)      Time     Calls       Avg       Min       Max  Name
 GPU activities:   49.11%  68.423ms      2000  34.211us  33.504us  35.680us  [CUDA memcpy DtoD]
                   22.89%  31.888ms      1000  31.888us  31.615us  32.127us  void paddle::operators::ScatterInputGradGPUKernel<float, long>(float*, int, long const *, long, int, int, long, long)
                   22.85%  31.835ms      1000  31.834us  31.616us  32.064us  void paddle::operators::GatherScatterGPUKernel<float, long, paddle::operators::TensorAssign, bool=1>(float*, int, long const *, float, long, int, int, long, long, paddle::operators::TensorAssign const &)
                    2.15%  2.9977ms      1000  2.9970us  2.9440us  5.3440us  void Eigen::internal::EigenMetaKernel<Eigen::TensorEvaluator<Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<float, int=2, int=1, int>, int=16, Eigen::MakePointer>, Eigen::TensorBroadcastingOp<Eigen::DSizes<long, int=2> const , Eigen::TensorMap<Eigen::Tensor<float const , int=2, int=1, int>, int=16, Eigen::MakePointer> const > const > const , Eigen::GpuDevice>, int>(float, int=2)
                    1.70%  2.3701ms      1000  2.3700us  2.2720us  2.8480us  void Eigen::internal::EigenMetaKernel<Eigen::TensorEvaluator<Eigen::TensorAssignOp<Eigen::TensorMap<Eigen::Tensor<long, int=2, int=1, int>, int=16, Eigen::MakePointer>, Eigen::TensorBroadcastingOp<Eigen::DSizes<long, int=2> const , Eigen::TensorMap<Eigen::Tensor<long const , int=2, int=1, int>, int=16, Eigen::MakePointer> const > const > const , Eigen::GpuDevice>, int>(long, int=2)
                    1.30%  1.8160ms      1000  1.8160us  1.7600us  2.4640us  [CUDA memcpy HtoD]

total gpu_time: 139.3260 ms

run command: /root/miniconda3/bin/python /home/paddle/benchmark/api/dynamic_tests_v2/put_along_axis.py --task speed --framework paddle --testing_mode dynamic --json_file /home/paddle/benchmark/api/tests_v2/configs/put_along_axis.json --config_id 1 --profiler none --backward True --use_gpu True --repeat 1000 --allow_adaptive_repeat False --log_level 0  --gpu_time  139.32600285074324
grep: warning: GREP_OPTIONS is deprecated; please use an alias or script
W0304 08:46:06.412171   793 gpu_context.cc:240] Please NOTE: device: 0, GPU Compute Capability: 7.0, Driver API Version: 10.1, Runtime API Version: 10.1
W0304 08:46:06.416616   793 gpu_context.cc:268] device: 0, cuDNN Version: 7.6.
Cannot import pytorch, maybe paddle is not installed.
Cannot import pytorch, maybe pytorch is not installed.
Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=1, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=139.32600285074324, json_file='/home/paddle/benchmark/api/tests_v2/configs/put_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=True)
---- Initialize APIConfig from /home/paddle/benchmark/api/tests_v2/configs/put_along_axis.json, config_id = 1.

Namespace(allow_adaptive_repeat=False, api_name=None, backward=True, config_id=1, convert_to_fp16=False, filename=None, framework='paddle', get_status_without_running=False, gpu_time=139.32600285074324, json_file='/home/paddle/benchmark/api/tests_v2/configs/put_along_axis.json', log_level=0, profiler='none', repeat=1000, scheduling_times='{}', sync_interval=80, task='speed', testing_mode='dynamic', unknown_dim=16, use_gpu=True)
[paddle][put_along_axis] put_along_axis {
  run_tf: True
  run_torch: False
  values: 1000.0
  axis: 0
  arr_shape: [1024, 3072]
  arr_dtype: float32
  indices_shape: [1, 1]
  indices_dtype: int
  atol: 1e-06
}
{"framework": "paddle", "version": "0.0.0", "name": "put_along_axis", "device": "GPU", "backward": true, "speed": {"repeat": 1000, "begin": 10, "end": 990, "total": 0.4084915530924894, "wall_time": 0, "total_include_wall_time": 0.4084915530924894, "gpu_time": 0.13932600285074323}, "parameters": "arr (Variable) - dtype: float32, shape: [1024, 3072]\nindices (Variable) - dtype: int, shape: [1, 1]\naxis (int): 0\nvalues (float32): 1000.0\n"}

ZzSean · 2022-03-04T07:26:24Z

api/dynamic_tests_v2/maxout.py

@@ -0,0 +1,39 @@
+#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.


copyright年份

ZzSean · 2022-03-04T07:27:51Z

api/dynamic_tests_v2/put_along_axis.py

@@ -0,0 +1,45 @@
+#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.


ZzSean · 2022-03-04T07:28:56Z

api/dynamic_tests_v2/take_along_axis.py

@@ -0,0 +1,44 @@
+#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.


ZzSean · 2022-03-04T07:29:38Z

api/tests_v2/configs/maxout.json

@@ -0,0 +1,21 @@
+[


再加个一个config，稍微大一点的

…t_fix

ZzSean

LGTM

m3ngyang mentioned this pull request Mar 1, 2022

[Phi] move ops: maxout/take_along_axis/put_along_axis PaddlePaddle/Paddle#39959

Merged

m3ngyang force-pushed the add_ops branch from 425703c to 6085eff Compare March 4, 2022 04:09

m3ngyang force-pushed the add_ops branch from 6085eff to 69b086c Compare March 4, 2022 06:29

m3ngyang closed this Mar 4, 2022

m3ngyang reopened this Mar 4, 2022

ZzSean reviewed Mar 4, 2022

View reviewed changes

api/tests_v2/configs/maxout.json

@@ -0,0 +1,21 @@

[

Copy link

Contributor

ZzSean Mar 4, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

再加个一个config，稍微大一点的

add benchmark for maxout/put_along_axis/take_along_axis, test=documen…

0be5a46

…t_fix

m3ngyang force-pushed the add_ops branch from 69b086c to 0be5a46 Compare March 4, 2022 08:47

m3ngyang closed this Mar 4, 2022

m3ngyang reopened this Mar 4, 2022

m3ngyang requested a review from ZzSean March 4, 2022 08:53

ZzSean approved these changes Mar 4, 2022

View reviewed changes

ZzSean merged commit af074d0 into PaddlePaddle:master Mar 4, 2022

m3ngyang deleted the add_ops branch March 4, 2022 08:56

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add benchmark for maxout/put_along_axis/take_along_axis #1326

add benchmark for maxout/put_along_axis/take_along_axis #1326

m3ngyang commented Mar 1, 2022 •

edited

Loading

paddle-bot-old bot commented Mar 1, 2022 •

edited

Loading

paddle-bot-old bot commented Mar 1, 2022

CLAassistant commented Mar 1, 2022 •

edited

Loading

m3ngyang commented Mar 4, 2022

m3ngyang commented Mar 4, 2022 •

edited

Loading

ZzSean Mar 4, 2022

ZzSean Mar 4, 2022

ZzSean Mar 4, 2022

ZzSean Mar 4, 2022

ZzSean left a comment

		@@ -0,0 +1,39 @@
		# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.

		@@ -0,0 +1,45 @@
		# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.

		@@ -0,0 +1,44 @@
		# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.

add benchmark for maxout/put_along_axis/take_along_axis #1326

add benchmark for maxout/put_along_axis/take_along_axis #1326

Conversation

m3ngyang commented Mar 1, 2022 • edited Loading

paddle-bot-old bot commented Mar 1, 2022 • edited Loading

paddle-bot-old bot commented Mar 1, 2022

CLAassistant commented Mar 1, 2022 • edited Loading

m3ngyang commented Mar 4, 2022

m3ngyang commented Mar 4, 2022 • edited Loading

ZzSean Mar 4, 2022

Choose a reason for hiding this comment

ZzSean Mar 4, 2022

Choose a reason for hiding this comment

ZzSean Mar 4, 2022

Choose a reason for hiding this comment

ZzSean Mar 4, 2022

Choose a reason for hiding this comment

ZzSean left a comment

Choose a reason for hiding this comment

m3ngyang commented Mar 1, 2022 •

edited

Loading

paddle-bot-old bot commented Mar 1, 2022 •

edited

Loading

CLAassistant commented Mar 1, 2022 •

edited

Loading

m3ngyang commented Mar 4, 2022 •

edited

Loading