feat: support prim::If in automatic fallback #447
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Bo Wang <[email protected]>
Signed-off-by: Bo Wang <[email protected]>
Signed-off-by: Bo Wang <[email protected]>
Signed-off-by: Bo Wang <[email protected]>
Signed-off-by: Bo Wang <[email protected]>
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
There are some changes that do not conform to C++ style guidelines:
diff --git a/workspace/core/compiler.cpp b/tmp/changes.txt
index dfcd69d..2f2f202 100644
--- a/workspace/core/compiler.cpp
+++ b/tmp/changes.txt
@@ -203,8 +203,8 @@ void AddIfBlockToGraph(
auto cur_block_mapping = graph_and_mapping.second;
std::unordered_map<torch::jit::Value*, torch::jit::Value*> block_graph_to_new_g;
for (auto& i : cur_block_mapping) {
- // for every pair in then_mapping, old_value => mini graph value, if old_value also appears in old_to_new_g, then it's
- // mini graph's input
+ // for every pair in then_mapping, old_value => mini graph value, if old_value also appears in old_to_new_g, then
+ // it's mini graph's input
if (old_to_new_g.count(i.first)) {
block_graph_to_new_g[i.second] = old_to_new_g[i.first];
}
@@ -317,8 +317,7 @@ torch::jit::script::Module CompileGraphWithFallback(const torch::jit::script::Mo
input_ranges.insert({g->inputs()[i], cfg.convert_info.input_ranges[i]});
}
auto input_ivalues_map = partitioning::generateRandomInputs(input_ranges);
- auto graph_and_mapping =
- ConstructFallbackGraph(new_mod, g->block(), input_ivalues_map, cfg, named_params);
+ auto graph_and_mapping = ConstructFallbackGraph(new_mod, g->block(), input_ivalues_map, cfg, named_params);
new_g = graph_and_mapping.first;
LOG_INFO(*new_g << "(FallbackGraph)\n");
diff --git a/workspace/core/partitioning/partitioning.cpp b/tmp/changes.txt
index 692878a..6590cd5 100644
--- a/workspace/core/partitioning/partitioning.cpp
+++ b/tmp/changes.txt
@@ -2,8 +2,8 @@
#include <queue>
#include "core/conversion/conversion.h"
-#include "torch/csrc/jit/passes/constant_pooling.h"
#include "core/partitioning/shape_analysis.h"
+#include "torch/csrc/jit/passes/constant_pooling.h"
#include "torch/csrc/jit/passes/dead_code_elimination.h"
namespace trtorch {
@@ -150,7 +150,8 @@ void resolveNonTensorInputs(PartitionedGraph& segmented_blocks) {
if (!updated_segments.count(first_torch_id)) {
auto to_inject_blocks = injectNodesForNonTensorInputs(segmented_blocks[first_torch_id]);
segmented_blocks.erase(segmented_blocks.begin() + first_torch_id);
- segmented_blocks.insert(segmented_blocks.begin() + first_torch_id, to_inject_blocks.begin(), to_inject_blocks.end());
+ segmented_blocks.insert(
+ segmented_blocks.begin() + first_torch_id, to_inject_blocks.begin(), to_inject_blocks.end());
updated_segments.insert(first_torch_id);
}
} else {
ERROR: Some files do not conform to style guidelines
There was a problem hiding this comment.
There are some changes that do not conform to C++ style guidelines:
diff --git a/workspace/core/compiler.cpp b/tmp/changes.txt
index dfcd69d..2f2f202 100644
--- a/workspace/core/compiler.cpp
+++ b/tmp/changes.txt
@@ -203,8 +203,8 @@ void AddIfBlockToGraph(
auto cur_block_mapping = graph_and_mapping.second;
std::unordered_map<torch::jit::Value*, torch::jit::Value*> block_graph_to_new_g;
for (auto& i : cur_block_mapping) {
- // for every pair in then_mapping, old_value => mini graph value, if old_value also appears in old_to_new_g, then it's
- // mini graph's input
+ // for every pair in then_mapping, old_value => mini graph value, if old_value also appears in old_to_new_g, then
+ // it's mini graph's input
if (old_to_new_g.count(i.first)) {
block_graph_to_new_g[i.second] = old_to_new_g[i.first];
}
@@ -317,8 +317,7 @@ torch::jit::script::Module CompileGraphWithFallback(const torch::jit::script::Mo
input_ranges.insert({g->inputs()[i], cfg.convert_info.input_ranges[i]});
}
auto input_ivalues_map = partitioning::generateRandomInputs(input_ranges);
- auto graph_and_mapping =
- ConstructFallbackGraph(new_mod, g->block(), input_ivalues_map, cfg, named_params);
+ auto graph_and_mapping = ConstructFallbackGraph(new_mod, g->block(), input_ivalues_map, cfg, named_params);
new_g = graph_and_mapping.first;
LOG_INFO(*new_g << "(FallbackGraph)\n");
diff --git a/workspace/core/partitioning/partitioning.cpp b/tmp/changes.txt
index 692878a..6590cd5 100644
--- a/workspace/core/partitioning/partitioning.cpp
+++ b/tmp/changes.txt
@@ -2,8 +2,8 @@
#include <queue>
#include "core/conversion/conversion.h"
-#include "torch/csrc/jit/passes/constant_pooling.h"
#include "core/partitioning/shape_analysis.h"
+#include "torch/csrc/jit/passes/constant_pooling.h"
#include "torch/csrc/jit/passes/dead_code_elimination.h"
namespace trtorch {
@@ -150,7 +150,8 @@ void resolveNonTensorInputs(PartitionedGraph& segmented_blocks) {
if (!updated_segments.count(first_torch_id)) {
auto to_inject_blocks = injectNodesForNonTensorInputs(segmented_blocks[first_torch_id]);
segmented_blocks.erase(segmented_blocks.begin() + first_torch_id);
- segmented_blocks.insert(segmented_blocks.begin() + first_torch_id, to_inject_blocks.begin(), to_inject_blocks.end());
+ segmented_blocks.insert(
+ segmented_blocks.begin() + first_torch_id, to_inject_blocks.begin(), to_inject_blocks.end());
updated_segments.insert(first_torch_id);
}
} else {
ERROR: Some files do not conform to style guidelines
There was a problem hiding this comment.
There are some changes that do not conform to C++ style guidelines:
diff --git a/workspace/core/compiler.cpp b/tmp/changes.txt
index dfcd69d..2f2f202 100644
--- a/workspace/core/compiler.cpp
+++ b/tmp/changes.txt
@@ -203,8 +203,8 @@ void AddIfBlockToGraph(
auto cur_block_mapping = graph_and_mapping.second;
std::unordered_map<torch::jit::Value*, torch::jit::Value*> block_graph_to_new_g;
for (auto& i : cur_block_mapping) {
- // for every pair in then_mapping, old_value => mini graph value, if old_value also appears in old_to_new_g, then it's
- // mini graph's input
+ // for every pair in then_mapping, old_value => mini graph value, if old_value also appears in old_to_new_g, then
+ // it's mini graph's input
if (old_to_new_g.count(i.first)) {
block_graph_to_new_g[i.second] = old_to_new_g[i.first];
}
@@ -317,8 +317,7 @@ torch::jit::script::Module CompileGraphWithFallback(const torch::jit::script::Mo
input_ranges.insert({g->inputs()[i], cfg.convert_info.input_ranges[i]});
}
auto input_ivalues_map = partitioning::generateRandomInputs(input_ranges);
- auto graph_and_mapping =
- ConstructFallbackGraph(new_mod, g->block(), input_ivalues_map, cfg, named_params);
+ auto graph_and_mapping = ConstructFallbackGraph(new_mod, g->block(), input_ivalues_map, cfg, named_params);
new_g = graph_and_mapping.first;
LOG_INFO(*new_g << "(FallbackGraph)\n");
diff --git a/workspace/core/partitioning/partitioning.cpp b/tmp/changes.txt
index 692878a..6590cd5 100644
--- a/workspace/core/partitioning/partitioning.cpp
+++ b/tmp/changes.txt
@@ -2,8 +2,8 @@
#include <queue>
#include "core/conversion/conversion.h"
-#include "torch/csrc/jit/passes/constant_pooling.h"
#include "core/partitioning/shape_analysis.h"
+#include "torch/csrc/jit/passes/constant_pooling.h"
#include "torch/csrc/jit/passes/dead_code_elimination.h"
namespace trtorch {
@@ -150,7 +150,8 @@ void resolveNonTensorInputs(PartitionedGraph& segmented_blocks) {
if (!updated_segments.count(first_torch_id)) {
auto to_inject_blocks = injectNodesForNonTensorInputs(segmented_blocks[first_torch_id]);
segmented_blocks.erase(segmented_blocks.begin() + first_torch_id);
- segmented_blocks.insert(segmented_blocks.begin() + first_torch_id, to_inject_blocks.begin(), to_inject_blocks.end());
+ segmented_blocks.insert(
+ segmented_blocks.begin() + first_torch_id, to_inject_blocks.begin(), to_inject_blocks.end());
updated_segments.insert(first_torch_id);
}
} else {
ERROR: Some files do not conform to style guidelines
Signed-off-by: Bo Wang <[email protected]>
Signed-off-by: Dheeraj Peri <[email protected]>
There was a problem hiding this comment.
There are some changes that do not conform to Python style guidelines:
Reformatting /workspace/cpp/ptq/training/vgg16/vgg16.py
Reformatting /workspace/cpp/ptq/training/vgg16/main.py
Reformatting /workspace/cpp/ptq/training/vgg16/export_ckpt.py
Reformatting /workspace/py/trtorch/ptq.py
Reformatting /workspace/py/trtorch/_compiler.py
Reformatting /workspace/py/trtorch/logging.py
Reformatting /workspace/py/trtorch/_compile_spec.py
Reformatting /workspace/py/trtorch/__init__.py
Reformatting /workspace/py/trtorch/_types.py
Reformatting /workspace/py/setup.py
--- /workspace/tests/modules/hub.py (original)
+++ /workspace/tests/modules/hub.py (reformatted)
@@ -99,6 +99,7 @@
# Sample Conditional Model (for testing partitioning and fallback in conditionals)
class FallbackIf(torch.nn.Module):
+
def __init__(self):
super(FallbackIf, self).__init__()
self.relu1 = torch.nn.ReLU()
@@ -120,6 +121,7 @@
x = self.conv1(x)
return x
+
conditional_model = FallbackIf().eval().cuda()
conditional_script_model = torch.jit.script(conditional_model)
torch.jit.save(conditional_script_model, "conditional_scripted.jit.pt")
Reformatting /workspace/tests/modules/hub.py
Reformatting /workspace/tests/py/test_to_backend_api.py
Reformatting /workspace/tests/py/test_ptq_to_backend.py
Reformatting /workspace/tests/py/model_test_case.py
Reformatting /workspace/tests/py/test_trt_intercompatability.py
Reformatting /workspace/tests/py/test_ptq_dataloader_calibrator.py
Reformatting /workspace/tests/py/test_api.py
Reformatting /workspace/tests/py/test_multi_gpu.py
Reformatting /workspace/tests/py/test_api_dla.py
Reformatting /workspace/tests/py/test_ptq_trt_calibrator.py
ERROR: Some files do not conform to style guidelines
|
@narendasan this is ready for your review |
Signed-off-by: Dheeraj Peri <[email protected]>
There was a problem hiding this comment.
There are some changes that do not conform to Python style guidelines:
Reformatting /workspace/cpp/ptq/training/vgg16/export_ckpt.py
Reformatting /workspace/cpp/ptq/training/vgg16/vgg16.py
Reformatting /workspace/cpp/ptq/training/vgg16/main.py
--- /workspace/tests/modules/hub.py (original)
+++ /workspace/tests/modules/hub.py (reformatted)
@@ -108,6 +108,7 @@
# Sample Conditional Model (for testing partitioning and fallback in conditionals)
class FallbackIf(torch.nn.Module):
+
def __init__(self):
super(FallbackIf, self).__init__()
self.relu1 = torch.nn.ReLU()
@@ -129,6 +130,7 @@
x = self.conv1(x)
return x
+
conditional_model = FallbackIf().eval().cuda()
conditional_script_model = torch.jit.script(conditional_model)
torch.jit.save(conditional_script_model, "conditional_scripted.jit.pt")
Reformatting /workspace/tests/modules/hub.py
Reformatting /workspace/tests/py/test_to_backend_api.py
Reformatting /workspace/tests/py/test_ptq_trt_calibrator.py
Reformatting /workspace/tests/py/test_api_dla.py
Reformatting /workspace/tests/py/test_trt_intercompatability.py
Reformatting /workspace/tests/py/model_test_case.py
Reformatting /workspace/tests/py/test_multi_gpu.py
Reformatting /workspace/tests/py/test_ptq_to_backend.py
Reformatting /workspace/tests/py/test_api.py
Reformatting /workspace/tests/py/test_ptq_dataloader_calibrator.py
ERROR: Some files do not conform to style guidelinesSigned-off-by: Dheeraj Peri <[email protected]>
There was a problem hiding this comment.
There are some changes that do not conform to Python style guidelines:
Reformatting /workspace/cpp/ptq/training/vgg16/export_ckpt.py
Reformatting /workspace/cpp/ptq/training/vgg16/vgg16.py
Reformatting /workspace/cpp/ptq/training/vgg16/main.py
--- /workspace/tests/modules/hub.py (original)
+++ /workspace/tests/modules/hub.py (reformatted)
@@ -108,6 +108,7 @@
# Sample Conditional Model (for testing partitioning and fallback in conditionals)
class FallbackIf(torch.nn.Module):
+
def __init__(self):
super(FallbackIf, self).__init__()
self.relu1 = torch.nn.ReLU()
@@ -129,6 +130,7 @@
x = self.conv1(x)
return x
+
conditional_model = FallbackIf().eval().cuda()
conditional_script_model = torch.jit.script(conditional_model)
torch.jit.save(conditional_script_model, "conditional_scripted.jit.pt")
Reformatting /workspace/tests/modules/hub.py
Reformatting /workspace/tests/py/test_to_backend_api.py
Reformatting /workspace/tests/py/test_ptq_trt_calibrator.py
Reformatting /workspace/tests/py/test_ptq_dataloader_calibrator.py
Reformatting /workspace/tests/py/test_api_dla.py
Reformatting /workspace/tests/py/test_trt_intercompatability.py
Reformatting /workspace/tests/py/model_test_case.py
Reformatting /workspace/tests/py/test_multi_gpu.py
Reformatting /workspace/tests/py/test_ptq_to_backend.py
Reformatting /workspace/tests/py/test_api.py
ERROR: Some files do not conform to style guidelines
There was a problem hiding this comment.
There are some changes that do not conform to C++ style guidelines:
diff --git a/workspace/core/compiler.cpp b/tmp/changes.txt
index 4dea4a7..9a63c9c 100644
--- a/workspace/core/compiler.cpp
+++ b/tmp/changes.txt
@@ -332,46 +332,46 @@ torch::jit::script::Module CompileGraphWithFallback(const torch::jit::script::Mo
return mod;
}
-// <<<<<<< HEAD
-// =======
-// std::unordered_map<torch::jit::Value*, torch::jit::Value*> old_to_new_g;
-// // add global graph's input to old_to_new_g mapping
-// for (auto input : g->inputs()) {
-// util::getOrAddInputForValue(input, new_g, old_to_new_g);
-// }
-// for (auto& seg_block : segmented_blocks) {
-// std::string cur_block_target =
-// seg_block.target() == partitioning::SegmentedBlock::kTensorRT ? "TensorRT" : "Torch";
-// LOG_INFO(*seg_block.g() << "(Sub Graph" << cur_block_target << "Block)\n");
-// std::ostringstream trt_engine_id;
-// trt_engine_id << reinterpret_cast<const int*>(&seg_block);
-// if (seg_block.target() == partitioning::SegmentedBlock::kTensorRT) {
-// std::vector<ir::Input> inputs;
-// for (auto& shape : seg_block.in_shape()) {
-// inputs.push_back(ir::Input(shape));
-// }
-// // update the input ranges for each segments
-// convert_cfg.inputs = inputs;
-// auto engine = conversion::ConvertBlockToEngine(seg_block.block(), convert_cfg, named_params);
-// auto temp_g = std::make_shared<torch::jit::Graph>();
-// auto device_spec = convert_cfg.engine_settings.device;
-// auto cuda_device = runtime::CudaDevice(device_spec.gpu_id, device_spec.device_type);
-// AddEngineToGraph(new_mod, temp_g, engine, cuda_device, trt_engine_id.str(), true);
-//
-// seg_block.update_graph(temp_g);
-// AddSegmentedBlockToGraph(new_g, seg_block, old_to_new_g);
-// } else {
-// AddSegmentedBlockToGraph(new_g, seg_block, old_to_new_g);
-// }
-// }
-//
-// for (auto& output : g->outputs()) {
-// new_g->registerOutput(old_to_new_g[output]);
-// }
-//
-// LOG_INFO(*new_g << "(FallbackGraph)\n");
-//
-// >>>>>>> master
+ // <<<<<<< HEAD
+ // =======
+ // std::unordered_map<torch::jit::Value*, torch::jit::Value*> old_to_new_g;
+ // // add global graph's input to old_to_new_g mapping
+ // for (auto input : g->inputs()) {
+ // util::getOrAddInputForValue(input, new_g, old_to_new_g);
+ // }
+ // for (auto& seg_block : segmented_blocks) {
+ // std::string cur_block_target =
+ // seg_block.target() == partitioning::SegmentedBlock::kTensorRT ? "TensorRT" : "Torch";
+ // LOG_INFO(*seg_block.g() << "(Sub Graph" << cur_block_target << "Block)\n");
+ // std::ostringstream trt_engine_id;
+ // trt_engine_id << reinterpret_cast<const int*>(&seg_block);
+ // if (seg_block.target() == partitioning::SegmentedBlock::kTensorRT) {
+ // std::vector<ir::Input> inputs;
+ // for (auto& shape : seg_block.in_shape()) {
+ // inputs.push_back(ir::Input(shape));
+ // }
+ // // update the input ranges for each segments
+ // convert_cfg.inputs = inputs;
+ // auto engine = conversion::ConvertBlockToEngine(seg_block.block(), convert_cfg, named_params);
+ // auto temp_g = std::make_shared<torch::jit::Graph>();
+ // auto device_spec = convert_cfg.engine_settings.device;
+ // auto cuda_device = runtime::CudaDevice(device_spec.gpu_id, device_spec.device_type);
+ // AddEngineToGraph(new_mod, temp_g, engine, cuda_device, trt_engine_id.str(), true);
+ //
+ // seg_block.update_graph(temp_g);
+ // AddSegmentedBlockToGraph(new_g, seg_block, old_to_new_g);
+ // } else {
+ // AddSegmentedBlockToGraph(new_g, seg_block, old_to_new_g);
+ // }
+ // }
+ //
+ // for (auto& output : g->outputs()) {
+ // new_g->registerOutput(old_to_new_g[output]);
+ // }
+ //
+ // LOG_INFO(*new_g << "(FallbackGraph)\n");
+ //
+ // >>>>>>> master
auto new_method = new_mod._ivalue()->compilation_unit()->create_function(method.name(), new_g);
auto schema = util::GenerateGraphSchema(new_method->name(), new_g);
new_mod.type()->addMethod(new_method);
ERROR: Some files do not conform to style guidelinesSigned-off-by: Dheeraj Peri <[email protected]>
There was a problem hiding this comment.
There are some changes that do not conform to Python style guidelines:
Reformatting /workspace/cpp/ptq/training/vgg16/export_ckpt.py
Reformatting /workspace/cpp/ptq/training/vgg16/vgg16.py
Reformatting /workspace/cpp/ptq/training/vgg16/main.py
--- /workspace/tests/modules/hub.py (original)
+++ /workspace/tests/modules/hub.py (reformatted)
@@ -100,6 +100,7 @@
# Sample Conditional Model (for testing partitioning and fallback in conditionals)
class FallbackIf(torch.nn.Module):
+
def __init__(self):
super(FallbackIf, self).__init__()
self.relu1 = torch.nn.ReLU()
@@ -121,6 +122,7 @@
x = self.conv1(x)
return x
+
conditional_model = FallbackIf().eval().cuda()
conditional_script_model = torch.jit.script(conditional_model)
torch.jit.save(conditional_script_model, "conditional_scripted.jit.pt")
Reformatting /workspace/tests/modules/hub.py
Reformatting /workspace/tests/py/test_to_backend_api.py
Reformatting /workspace/tests/py/test_api.py
Reformatting /workspace/tests/py/test_api_dla.py
Reformatting /workspace/tests/py/test_trt_intercompatability.py
Reformatting /workspace/tests/py/model_test_case.py
Reformatting /workspace/tests/py/test_multi_gpu.py
Reformatting /workspace/tests/py/test_ptq_to_backend.py
Reformatting /workspace/tests/py/test_ptq_trt_calibrator.py
Reformatting /workspace/tests/py/test_ptq_dataloader_calibrator.py
ERROR: Some files do not conform to style guidelines
Signed-off-by: Naren Dasan <[email protected]> Signed-off-by: Naren Dasan <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Support prim::If in automatic fallback.
Previously, all the prim::If operators will be fallback to PyTorch nodes when automatic fallback feature is enabled. This could be optimized since some operators in prim::If block could also be converted to TensorRT engines.
Type of change
Checklist: