Skip to content

Commit a90e6db

Browse files
narendasannarendasan
committed
fix(pooling): fix the tests and the 1D pooling cases
Signed-off-by: Naren Dasan <naren@narendasan.com> Signed-off-by: Naren Dasan <narens@nvidia.com>
1 parent 7dc4af4 commit a90e6db

File tree

2 files changed

+271
-30
lines changed

2 files changed

+271
-30
lines changed

‎core/conversion/converters/impl/pooling.cpp‎

Lines changed: 77 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -23,23 +23,23 @@ bool MaxPoolingConverter(ConversionCtx* ctx, const torch::jit::Node* n, args& ar
2323
}
2424

2525

26-
auto kernel_size = util::toDimsHW(args[1].unwrapToIntList());
26+
auto kernel_size = util::toDims(args[1].unwrapToIntList());
2727
LOG_DEBUG("kernel_size: " << kernel_size);
28-
auto padding = util::toDimsHW(args[3].unwrapToIntList());
28+
auto padding = util::toDims(args[3].unwrapToIntList());
2929
LOG_DEBUG("padding: " << padding);
3030
auto stride = util::toDims(args[2].unwrapToIntList());
3131
LOG_DEBUG("stride: " << stride);
3232

3333
auto dilation = util::toDims(args[4].unwrapToIntList());
3434

35-
TRTORCH_ASSERT(dilation == util::toDims(std::vector<int64_t>({1,1})), "Pooling dilation is not supported in TensorRT");
35+
TRTORCH_ASSERT(dilation == util::toDims(std::vector<int64_t>(dilation.nbDims, 1)), "Pooling dilation is not supported in TensorRT");
3636

3737
LOG_DEBUG("dilation: " << dilation);
3838
LOG_WARNING("Dilation not used in max pooling converter");
3939
bool ceil_mode = args[5].unwrapToBool();
4040

4141
auto new_layer = ctx->net->addPoolingNd(*in, nvinfer1::PoolingType::kMAX, kernel_size);
42-
TRTORCH_CHECK(new_layer, "Unable to create Max Pool 2D layer from node: " << *n);
42+
TRTORCH_CHECK(new_layer, "Unable to create Max Pooling layer from node: " << *n);
4343

4444
new_layer->setName(util::node_info(n).c_str());
4545
new_layer->setPaddingNd(padding);
@@ -77,9 +77,9 @@ bool AvgPoolingConverter(ConversionCtx* ctx, const torch::jit::Node* n, args& ar
7777
}
7878

7979

80-
auto kernel_size = util::toDimsHW(args[1].unwrapToIntList());
80+
auto kernel_size = util::toDims(args[1].unwrapToIntList());
8181
LOG_DEBUG("kernel_size: " << kernel_size);
82-
auto padding = util::toDimsHW(args[3].unwrapToIntList());
82+
auto padding = util::toDims(args[3].unwrapToIntList());
8383
LOG_DEBUG("padding: " << padding);
8484
auto stride = util::toDims(args[2].unwrapToIntList());
8585
LOG_DEBUG("stride: " << stride);
@@ -88,7 +88,7 @@ bool AvgPoolingConverter(ConversionCtx* ctx, const torch::jit::Node* n, args& ar
8888
bool count_inlcude_pad = args[5].unwrapToBool();
8989

9090
auto new_layer = ctx->net->addPoolingNd(*in, nvinfer1::PoolingType::kAVERAGE, kernel_size);
91-
TRTORCH_CHECK(new_layer, "Unable to create Avg Pool 2D layer from node: " << *n);
91+
TRTORCH_CHECK(new_layer, "Unable to create Avg Pooling layer from node: " << *n);
9292

9393
new_layer->setName(util::node_info(n).c_str());
9494
new_layer->setPaddingNd(padding);
@@ -118,12 +118,67 @@ bool AvgPoolingConverter(ConversionCtx* ctx, const torch::jit::Node* n, args& ar
118118

119119
auto pooling_registrations TRTORCH_UNUSED = RegisterNodeConversionPatterns()
120120
.pattern({
121-
"aten::max_pool1d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=[0, 0], int[2] dilation=[1, 1], bool ceil_mode=False) -> (Tensor)",
121+
"aten::max_pool1d(Tensor self, int[1] kernel_size, int[1] stride=[], int[1] padding=[], int[1] dilation=[], bool ceil_mode=False) -> (Tensor)",
122122
[](ConversionCtx* ctx, const torch::jit::Node* n, args& args) -> bool {
123-
return MaxPoolingConverter(ctx, n, args);
123+
auto in = args[0].ITensor();
124+
auto shape = util::toVec(in->getDimensions());
125+
126+
// Max Pool needs at least 4D input
127+
if (shape.size() < 4) {
128+
auto new_shape = util::toDimsPad(shape, 4);
129+
LOG_DEBUG("Input shape is less than 4D got: " << util::toDims(shape) << ", inserting shuffle layer to reshape to 4D tensor shape: " << new_shape);
130+
auto shuffle = ctx->net->addShuffle(*in);
131+
shuffle->setReshapeDimensions(new_shape);
132+
shuffle->setName((util::node_info(n) + " [Reshape to " + util::toStr(new_shape) + ']').c_str());
133+
in = shuffle->getOutput(0);
134+
}
135+
136+
auto kernel_vec = args[1].unwrapToIntList().vec();
137+
kernel_vec.insert(kernel_vec.begin(), 1);
138+
auto kernel_size = util::toDims(kernel_vec);
139+
LOG_DEBUG("kernel_size: " << kernel_size);
140+
auto stride_vec = args[2].unwrapToIntList().vec();
141+
stride_vec.insert(stride_vec.begin(), 1);
142+
auto stride = util::toDims(stride_vec);
143+
LOG_DEBUG("stride: " << stride);
144+
auto padding_vec = args[3].unwrapToIntList().vec();
145+
padding_vec.insert(padding_vec.begin(), 0);
146+
auto padding = util::toDims(padding_vec);
147+
LOG_DEBUG("padding: " << padding);
148+
149+
auto dilation = util::toDims(args[4].unwrapToIntList());
150+
151+
TRTORCH_ASSERT(dilation == util::toDims(std::vector<int64_t>(dilation.nbDims, 1)), "Pooling dilation is not supported in TensorRT");
152+
153+
LOG_DEBUG("dilation: " << dilation);
154+
LOG_WARNING("Dilation not used in max pooling converter");
155+
bool ceil_mode = args[5].unwrapToBool();
156+
157+
auto new_layer = ctx->net->addPoolingNd(*in, nvinfer1::PoolingType::kMAX, kernel_size);
158+
TRTORCH_CHECK(new_layer, "Unable to create Max Pooling layer from node: " << *n);
159+
160+
new_layer->setName(util::node_info(n).c_str());
161+
new_layer->setPaddingNd(padding);
162+
if (stride.nbDims != 2 && ctx->settings.device == nvinfer1::DeviceType::kDLA) {
163+
if (!ctx->settings.allow_gpu_fallback) {
164+
TRTORCH_THROW_ERROR("DLA Pooling stride is limited to 2D, allow GPU fallback");
165+
} else {
166+
LOG_WARNING("DLA Pooling stride is limited to 2D, will run on GPU");
167+
}
168+
}
169+
new_layer->setStrideNd(stride);
170+
171+
auto padding_mode = ceil_mode ? nvinfer1::PaddingMode::kEXPLICIT_ROUND_UP : nvinfer1::PaddingMode::kEXPLICIT_ROUND_DOWN;
172+
new_layer->setPaddingMode(padding_mode);
173+
174+
new_layer->setName(util::node_info(n).c_str());
175+
auto out_tensor = ctx->AssociateValueAndTensor(n->outputs()[0], new_layer->getOutput(0));
176+
177+
LOG_DEBUG("Output tensor shape: " << out_tensor->getDimensions());
178+
return true;
124179
}
125180
}).pattern({
126-
"aten::avg_pool1d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=[0, 0], bool ceil_mode=False, bool count_include_pad=True) -> (Tensor)",
181+
"aten::avg_pool1d(Tensor self, int[1] kernel_size, int[1] stride=[], int[1] padding=0, bool ceil_mode=False, bool count_include_pad=True) -> Tensor",
127182
[](ConversionCtx* ctx, const torch::jit::Node* n, args& args) -> bool {
128183
auto in = args[0].ITensor();
129184
auto shape = util::toVec(in->getDimensions());
@@ -139,12 +194,18 @@ auto pooling_registrations TRTORCH_UNUSED = RegisterNodeConversionPatterns()
139194
}
140195

141196

142-
auto kernel_size = util::toDimsHW(args[1].unwrapToIntList());
197+
auto kernel_vec = args[1].unwrapToIntList().vec();
198+
kernel_vec.insert(kernel_vec.begin(), 1);
199+
auto kernel_size = util::toDims(kernel_vec);
143200
LOG_DEBUG("kernel_size: " << kernel_size);
144-
auto padding = util::toDimsHW(args[3].unwrapToIntList());
145-
LOG_DEBUG("padding: " << padding);
146-
auto stride = util::toDims(args[2].unwrapToIntList());
201+
auto stride_vec = args[2].unwrapToIntList().vec();
202+
stride_vec.insert(stride_vec.begin(), 1);
203+
auto stride = util::toDims(stride_vec);
147204
LOG_DEBUG("stride: " << stride);
205+
auto padding_vec = args[3].unwrapToIntList().vec();
206+
padding_vec.insert(padding_vec.begin(), 0);
207+
auto padding = util::toDims(padding_vec);
208+
LOG_DEBUG("padding: " << padding);
148209

149210
bool ceil_mode = args[4].unwrapToBool();
150211
bool count_inlcude_pad = args[5].unwrapToBool();
@@ -187,12 +248,12 @@ auto pooling_registrations TRTORCH_UNUSED = RegisterNodeConversionPatterns()
187248
return AvgPoolingConverter(ctx, n, args);
188249
}
189250
}).pattern({
190-
"aten::max_pool3d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=[0, 0], int[2] dilation=[1, 1], bool ceil_mode=False) -> (Tensor)",
251+
"aten::max_pool3d(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=[], int[3] dilation=[], bool ceil_mode=False) -> (Tensor)",
191252
[](ConversionCtx* ctx, const torch::jit::Node* n, args& args) -> bool {
192253
return MaxPoolingConverter(ctx, n, args);
193254
}
194255
}).pattern({
195-
"aten::avg_pool3d(Tensor self, int[2] kernel_size, int[2] stride=[], int[2] padding=[0, 0], bool ceil_mode=False, bool count_include_pad=True, int? divisor_override=None) -> (Tensor)",
256+
"aten::avg_pool3d(Tensor self, int[3] kernel_size, int[3] stride=[], int[3] padding=[], bool ceil_mode=False, bool count_include_pad=True, int? divisor_override=None) -> (Tensor)",
196257
[](ConversionCtx* ctx, const torch::jit::Node* n, args& args) -> bool {
197258
return AvgPoolingConverter(ctx, n, args);
198259
}

0 commit comments

Comments
 (0)