Skip to content

Commit 74a6e5a

Browse files
JanuszLJanuszL
committed
Review fix
Signed-off-by: Janusz Lisiecki <jlisiecki@nvidia.com>
1 parent 7637a09 commit 74a6e5a

File tree

1 file changed

+146
-146
lines changed

1 file changed

+146
-146
lines changed

‎docs/examples/custom_operations/custom_operator/create_a_custom_operator.ipynb

Lines changed: 146 additions & 146 deletions
Original file line numberDiff line numberDiff line change
@@ -43,47 +43,47 @@
4343
"name": "stdout",
4444
"output_type": "stream",
4545
"text": [
46-
"#ifndef EXAMPLE_DUMMY_H_\r\n",
47-
"#define EXAMPLE_DUMMY_H_\r\n",
48-
"\r\n",
49-
"#include <vector>\r\n",
50-
"\r\n",
51-
"#include \"dali/pipeline/operator/operator.h\"\r\n",
52-
"\r\n",
53-
"namespace other_ns {\r\n",
54-
"\r\n",
55-
"template <typename Backend>\r\n",
56-
"class Dummy : public ::dali::Operator<Backend> {\r\n",
57-
" public:\r\n",
58-
" inline explicit Dummy(const ::dali::OpSpec &spec) :\r\n",
59-
" ::dali::Operator<Backend>(spec) {}\r\n",
60-
"\r\n",
61-
" virtual inline ~Dummy() = default;\r\n",
62-
"\r\n",
63-
" Dummy(const Dummy&) = delete;\r\n",
64-
" Dummy& operator=(const Dummy&) = delete;\r\n",
65-
" Dummy(Dummy&&) = delete;\r\n",
66-
" Dummy& operator=(Dummy&&) = delete;\r\n",
67-
"\r\n",
68-
" protected:\r\n",
69-
" bool CanInferOutputs() const override {\r\n",
70-
" return true;\r\n",
71-
" }\r\n",
72-
"\r\n",
73-
" bool SetupImpl(std::vector<::dali::OutputDesc> &output_desc,\r\n",
74-
" const ::dali::Workspace &ws) override {\r\n",
75-
" const auto &input = ws.Input<Backend>(0);\r\n",
76-
" output_desc.resize(1);\r\n",
77-
" output_desc[0] = {input.shape(), input.type()};\r\n",
78-
" return true;\r\n",
79-
" }\r\n",
80-
"\r\n",
81-
" void RunImpl(::dali::Workspace &ws) override;\r\n",
82-
"};\r\n",
83-
"\r\n",
84-
"} // namespace other_ns\r\n",
85-
"\r\n",
86-
"#endif // EXAMPLE_DUMMY_H_\r\n"
46+
"#ifndef EXAMPLE_DUMMY_H_\n",
47+
"#define EXAMPLE_DUMMY_H_\n",
48+
"\n",
49+
"#include <vector>\n",
50+
"\n",
51+
"#include \"dali/pipeline/operator/operator.h\"\n",
52+
"\n",
53+
"namespace other_ns {\n",
54+
"\n",
55+
"template <typename Backend>\n",
56+
"class Dummy : public ::dali::Operator<Backend> {\n",
57+
" public:\n",
58+
" inline explicit Dummy(const ::dali::OpSpec &spec) :\n",
59+
" ::dali::Operator<Backend>(spec) {}\n",
60+
"\n",
61+
" virtual inline ~Dummy() = default;\n",
62+
"\n",
63+
" Dummy(const Dummy&) = delete;\n",
64+
" Dummy& operator=(const Dummy&) = delete;\n",
65+
" Dummy(Dummy&&) = delete;\n",
66+
" Dummy& operator=(Dummy&&) = delete;\n",
67+
"\n",
68+
" protected:\n",
69+
" bool CanInferOutputs() const override {\n",
70+
" return true;\n",
71+
" }\n",
72+
"\n",
73+
" bool SetupImpl(std::vector<::dali::OutputDesc> &output_desc,\n",
74+
" const ::dali::Workspace &ws) override {\n",
75+
" const auto &input = ws.Input<Backend>(0);\n",
76+
" output_desc.resize(1);\n",
77+
" output_desc[0] = {input.shape(), input.type()};\n",
78+
" return true;\n",
79+
" }\n",
80+
"\n",
81+
" void RunImpl(::dali::Workspace &ws) override;\n",
82+
"};\n",
83+
"\n",
84+
"} // namespace other_ns\n",
85+
"\n",
86+
"#endif // EXAMPLE_DUMMY_H_\n"
8787
]
8888
}
8989
],
@@ -115,40 +115,40 @@
115115
"name": "stdout",
116116
"output_type": "stream",
117117
"text": [
118-
"#include \"dummy.h\"\r\n",
119-
"\r\n",
120-
"namespace other_ns {\r\n",
121-
"\r\n",
122-
"template <>\r\n",
123-
"void Dummy<::dali::CPUBackend>::RunImpl(::dali::Workspace &ws) {\r\n",
124-
" const auto &input = ws.Input<::dali::CPUBackend>(0);\r\n",
125-
" auto &output = ws.Output<::dali::CPUBackend>(0);\r\n",
126-
"\r\n",
127-
" ::dali::TypeInfo type = input.type_info();\r\n",
128-
" auto &tp = ws.GetThreadPool();\r\n",
129-
" const auto &in_shape = input.shape();\r\n",
130-
" for (int sample_id = 0; sample_id < in_shape.num_samples(); sample_id++) {\r\n",
131-
" tp.AddWork(\r\n",
132-
" [&, sample_id](int thread_id) {\r\n",
133-
" type.Copy<::dali::CPUBackend, ::dali::CPUBackend>(\r\n",
134-
" output.raw_mutable_tensor(sample_id),\r\n",
135-
" input.raw_tensor(sample_id),\r\n",
136-
" in_shape.tensor_size(sample_id), 0);\r\n",
137-
" },\r\n",
138-
" in_shape.tensor_size(sample_id));\r\n",
139-
" }\r\n",
140-
" tp.RunAll();\r\n",
141-
"}\r\n",
142-
"\r\n",
143-
"} // namespace other_ns\r\n",
144-
"\r\n",
145-
"DALI_REGISTER_OPERATOR(CustomDummy, ::other_ns::Dummy<::dali::CPUBackend>,\r\n",
146-
" ::dali::CPU);\r\n",
147-
"\r\n",
148-
"DALI_SCHEMA(CustomDummy)\r\n",
149-
" .DocStr(\"Make a copy of the input tensor\")\r\n",
150-
" .NumInput(1)\r\n",
151-
" .NumOutput(1);\r\n"
118+
"#include \"dummy.h\"\n",
119+
"\n",
120+
"namespace other_ns {\n",
121+
"\n",
122+
"template <>\n",
123+
"void Dummy<::dali::CPUBackend>::RunImpl(::dali::Workspace &ws) {\n",
124+
" const auto &input = ws.Input<::dali::CPUBackend>(0);\n",
125+
" auto &output = ws.Output<::dali::CPUBackend>(0);\n",
126+
"\n",
127+
" ::dali::TypeInfo type = input.type_info();\n",
128+
" auto &tp = ws.GetThreadPool();\n",
129+
" const auto &in_shape = input.shape();\n",
130+
" for (int sample_id = 0; sample_id < in_shape.num_samples(); sample_id++) {\n",
131+
" tp.AddWork(\n",
132+
" [&, sample_id](int thread_id) {\n",
133+
" type.Copy<::dali::CPUBackend, ::dali::CPUBackend>(\n",
134+
" output.raw_mutable_tensor(sample_id),\n",
135+
" input.raw_tensor(sample_id),\n",
136+
" in_shape.tensor_size(sample_id), 0);\n",
137+
" },\n",
138+
" in_shape.tensor_size(sample_id));\n",
139+
" }\n",
140+
" tp.RunAll();\n",
141+
"}\n",
142+
"\n",
143+
"} // namespace other_ns\n",
144+
"\n",
145+
"DALI_REGISTER_OPERATOR(CustomDummy, ::other_ns::Dummy<::dali::CPUBackend>,\n",
146+
" ::dali::CPU);\n",
147+
"\n",
148+
"DALI_SCHEMA(CustomDummy)\n",
149+
" .DocStr(\"Make a copy of the input tensor\")\n",
150+
" .NumInput(1)\n",
151+
" .NumOutput(1);\n"
152152
]
153153
}
154154
],
@@ -180,30 +180,30 @@
180180
"name": "stdout",
181181
"output_type": "stream",
182182
"text": [
183-
"#include <cuda_runtime_api.h>\r\n",
184-
"#include \"dummy.h\"\r\n",
185-
"\r\n",
186-
"namespace other_ns {\r\n",
187-
"\r\n",
188-
"template<>\r\n",
189-
"void Dummy<::dali::GPUBackend>::RunImpl(::dali::Workspace &ws) {\r\n",
190-
" const auto &input = ws.Input<::dali::GPUBackend>(0);\r\n",
191-
" const auto &shape = input.shape();\r\n",
192-
" auto &output = ws.Output<::dali::GPUBackend>(0);\r\n",
193-
" for (int sample_idx = 0; sample_idx < shape.num_samples(); sample_idx++) {\r\n",
194-
" CUDA_CALL(cudaMemcpyAsync(\r\n",
195-
" output.raw_mutable_tensor(sample_idx),\r\n",
196-
" input.raw_tensor(sample_idx),\r\n",
197-
" shape[sample_idx].num_elements() * input.type_info().size(),\r\n",
198-
" cudaMemcpyDeviceToDevice,\r\n",
199-
" ws.stream()));\r\n",
200-
" }\r\n",
201-
"}\r\n",
202-
"\r\n",
203-
"} // namespace other_ns\r\n",
204-
"\r\n",
205-
"DALI_REGISTER_OPERATOR(CustomDummy, ::other_ns::Dummy<::dali::GPUBackend>,\r\n",
206-
" ::dali::GPU);\r\n"
183+
"#include <cuda_runtime_api.h>\n",
184+
"#include \"dummy.h\"\n",
185+
"\n",
186+
"namespace other_ns {\n",
187+
"\n",
188+
"template<>\n",
189+
"void Dummy<::dali::GPUBackend>::RunImpl(::dali::Workspace &ws) {\n",
190+
" const auto &input = ws.Input<::dali::GPUBackend>(0);\n",
191+
" const auto &shape = input.shape();\n",
192+
" auto &output = ws.Output<::dali::GPUBackend>(0);\n",
193+
" for (int sample_idx = 0; sample_idx < shape.num_samples(); sample_idx++) {\n",
194+
" CUDA_CALL(cudaMemcpyAsync(\n",
195+
" output.raw_mutable_tensor(sample_idx),\n",
196+
" input.raw_tensor(sample_idx),\n",
197+
" shape[sample_idx].num_elements() * input.type_info().size(),\n",
198+
" cudaMemcpyDeviceToDevice,\n",
199+
" ws.stream()));\n",
200+
" }\n",
201+
"}\n",
202+
"\n",
203+
"} // namespace other_ns\n",
204+
"\n",
205+
"DALI_REGISTER_OPERATOR(CustomDummy, ::other_ns::Dummy<::dali::GPUBackend>,\n",
206+
" ::dali::GPU);\n"
207207
]
208208
}
209209
],
@@ -239,7 +239,7 @@
239239
"name": "stdout",
240240
"output_type": "stream",
241241
"text": [
242-
"/home/jlisiecki/Dali/dali/compile/dali/python/nvidia/dali/include\n"
242+
"/usr/local/lib/python3.8/dist-packages/nvidia/dali/include\n"
243243
]
244244
}
245245
],
@@ -256,7 +256,7 @@
256256
"name": "stdout",
257257
"output_type": "stream",
258258
"text": [
259-
"/home/jlisiecki/Dali/dali/compile/dali/python/nvidia/dali\n"
259+
"/usr/local/lib/python3.8/dist-packages/nvidia/dali\n"
260260
]
261261
}
262262
],
@@ -273,7 +273,7 @@
273273
"name": "stdout",
274274
"output_type": "stream",
275275
"text": [
276-
"['-I/home/jlisiecki/Dali/dali/compile/dali/python/nvidia/dali/include', '-D_GLIBCXX_USE_CXX11_ABI=1']\n"
276+
"['-I/usr/local/lib/python3.8/dist-packages/nvidia/dali/include', '-D_GLIBCXX_USE_CXX11_ABI=1']\n"
277277
]
278278
}
279279
],
@@ -290,7 +290,7 @@
290290
"name": "stdout",
291291
"output_type": "stream",
292292
"text": [
293-
"['-L/home/jlisiecki/Dali/dali/compile/dali/python/nvidia/dali', '-ldali']\n"
293+
"['-L/usr/local/lib/python3.8/dist-packages/nvidia/dali', '-ldali']\n"
294294
]
295295
}
296296
],
@@ -317,42 +317,42 @@
317317
"name": "stdout",
318318
"output_type": "stream",
319319
"text": [
320-
"cmake_minimum_required(VERSION 3.10)\r\n",
321-
"set(CMAKE_CUDA_ARCHITECTURES \"50;60;70;80;90\")\r\n",
322-
"\r\n",
323-
"project(custom_dummy_plugin LANGUAGES CUDA CXX C)\r\n",
324-
"\r\n",
325-
"set(CMAKE_CXX_STANDARD 17)\r\n",
326-
"set(CMAKE_CXX_STANDARD_REQUIRED ON)\r\n",
327-
"set(CMAKE_CXX_EXTENSIONS OFF)\r\n",
328-
"set(CMAKE_C_STANDARD 11)\r\n",
329-
"\r\n",
330-
"# TODO(klecki): When the test container gets a CMake that supports C++17 as a proper option,\r\n",
331-
"# swap those lines\r\n",
332-
"# set(CMAKE_CUDA_STANDARD 17)\r\n",
333-
"# set(CMAKE_CUDA_STANDARD_REQUIRED ON)\r\n",
334-
"set(CMAKE_CUDA_FLAGS \"${CMAKE_CUDA_FLAGS} -std=c++17\")\r\n",
335-
"\r\n",
336-
"include_directories(SYSTEM \"${CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES}\")\r\n",
337-
"\r\n",
338-
"execute_process(\r\n",
339-
" COMMAND python -c \"import nvidia.dali as dali; \\\r\n",
340-
" print(dali.sysconfig.get_lib_dir())\"\r\n",
341-
" OUTPUT_VARIABLE DALI_LIB_DIR)\r\n",
342-
"string(STRIP ${DALI_LIB_DIR} DALI_LIB_DIR)\r\n",
343-
"\r\n",
344-
"execute_process(\r\n",
345-
" COMMAND python -c \"import nvidia.dali as dali; print(\\\" \\\r\n",
346-
" \\\".join(dali.sysconfig.get_compile_flags()))\"\r\n",
347-
" OUTPUT_VARIABLE DALI_COMPILE_FLAGS)\r\n",
348-
"string(STRIP ${DALI_COMPILE_FLAGS} DALI_COMPILE_FLAGS)\r\n",
349-
"\r\n",
350-
"set(CMAKE_CXX_FLAGS \"${CMAKE_CXX_FLAGS} ${DALI_COMPILE_FLAGS} \")\r\n",
351-
"set(CMAKE_CUDA_FLAGS \"${CMAKE_CUDA_FLAGS} ${DALI_COMPILE_FLAGS} \")\r\n",
352-
"link_directories(\"${DALI_LIB_DIR}\")\r\n",
353-
"\r\n",
354-
"add_library(dali_customdummy SHARED dummy.cc dummy.cu)\r\n",
355-
"target_link_libraries(dali_customdummy dali)\r\n"
320+
"cmake_minimum_required(VERSION 3.10)\n",
321+
"set(CMAKE_CUDA_ARCHITECTURES \"50;60;70;80;90\")\n",
322+
"\n",
323+
"project(custom_dummy_plugin LANGUAGES CUDA CXX C)\n",
324+
"\n",
325+
"set(CMAKE_CXX_STANDARD 17)\n",
326+
"set(CMAKE_CXX_STANDARD_REQUIRED ON)\n",
327+
"set(CMAKE_CXX_EXTENSIONS OFF)\n",
328+
"set(CMAKE_C_STANDARD 11)\n",
329+
"\n",
330+
"# TODO(klecki): When the test container gets a CMake that supports C++17 as a proper option,\n",
331+
"# swap those lines\n",
332+
"# set(CMAKE_CUDA_STANDARD 17)\n",
333+
"# set(CMAKE_CUDA_STANDARD_REQUIRED ON)\n",
334+
"set(CMAKE_CUDA_FLAGS \"${CMAKE_CUDA_FLAGS} -std=c++17\")\n",
335+
"\n",
336+
"include_directories(SYSTEM \"${CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES}\")\n",
337+
"\n",
338+
"execute_process(\n",
339+
" COMMAND python -c \"import nvidia.dali as dali; \\\n",
340+
" print(dali.sysconfig.get_lib_dir())\"\n",
341+
" OUTPUT_VARIABLE DALI_LIB_DIR)\n",
342+
"string(STRIP ${DALI_LIB_DIR} DALI_LIB_DIR)\n",
343+
"\n",
344+
"execute_process(\n",
345+
" COMMAND python -c \"import nvidia.dali as dali; print(\\\" \\\n",
346+
" \\\".join(dali.sysconfig.get_compile_flags()))\"\n",
347+
" OUTPUT_VARIABLE DALI_COMPILE_FLAGS)\n",
348+
"string(STRIP ${DALI_COMPILE_FLAGS} DALI_COMPILE_FLAGS)\n",
349+
"\n",
350+
"set(CMAKE_CXX_FLAGS \"${CMAKE_CXX_FLAGS} ${DALI_COMPILE_FLAGS} \")\n",
351+
"set(CMAKE_CUDA_FLAGS \"${CMAKE_CUDA_FLAGS} ${DALI_COMPILE_FLAGS} \")\n",
352+
"link_directories(\"${DALI_LIB_DIR}\")\n",
353+
"\n",
354+
"add_library(dali_customdummy SHARED dummy.cc dummy.cu)\n",
355+
"target_link_libraries(dali_customdummy dali)\n"
356356
]
357357
}
358358
],
@@ -376,9 +376,9 @@
376376
"name": "stdout",
377377
"output_type": "stream",
378378
"text": [
379-
"-- The CUDA compiler identification is NVIDIA 12.5.40\n",
380-
"-- The CXX compiler identification is GNU 12.3.0\n",
381-
"-- The C compiler identification is GNU 12.3.0\n",
379+
"-- The CUDA compiler identification is NVIDIA 12.4.131\n",
380+
"-- The CXX compiler identification is GNU 9.4.0\n",
381+
"-- The C compiler identification is GNU 9.4.0\n",
382382
"-- Detecting CUDA compiler ABI info\n",
383383
"-- Detecting CUDA compiler ABI info - done\n",
384384
"-- Check for working CUDA compiler: /usr/local/cuda/bin/nvcc - skipped\n",
@@ -394,9 +394,9 @@
394394
"-- Check for working C compiler: /usr/bin/cc - skipped\n",
395395
"-- Detecting C compile features\n",
396396
"-- Detecting C compile features - done\n",
397-
"-- Configuring done (6.0s)\n",
397+
"-- Configuring done (5.2s)\n",
398398
"-- Generating done (0.0s)\n",
399-
"-- Build files have been written to: /home/jlisiecki/Dali/dali/docs/examples/custom_operations/custom_operator/customdummy/build\n",
399+
"-- Build files have been written to: /dali/docs/examples/custom_operations/custom_operator/customdummy/build\n",
400400
"[ 33%] \u001b[32mBuilding CXX object CMakeFiles/dali_customdummy.dir/dummy.cc.o\u001b[0m\n",
401401
"[ 66%] \u001b[32mBuilding CUDA object CMakeFiles/dali_customdummy.dir/dummy.cu.o\u001b[0m\n",
402402
"[100%] \u001b[32m\u001b[1mLinking CXX shared library libdali_customdummy.so\u001b[0m\n",
@@ -428,7 +428,7 @@
428428
"name": "stdout",
429429
"output_type": "stream",
430430
"text": [
431-
"customdummy/build/libdali_customdummy.so\r\n"
431+
"customdummy/build/libdali_customdummy.so\n"
432432
]
433433
}
434434
],
@@ -574,7 +574,7 @@
574574
"name": "python",
575575
"nbconvert_exporter": "python",
576576
"pygments_lexer": "ipython3",
577-
"version": "3.10.12"
577+
"version": "3.8.10"
578578
}
579579
},
580580
"nbformat": 4,

0 commit comments

Comments
 (0)