Disable cudnn option (#123)

* add disable cudnn option

* correct comment

* clang format

* add to readme

---------

Co-authored-by: jamied <[email protected]>

README.md

@@ -144,6 +144,26 @@ key: "INFERENCE_MODE"
 }
 ```
 
+* `DISABLE_CUDNN`: Boolean flag to disable the cuDNN library. By default, cuDNN is enabled.
+
+[cuDNN](https://developer.nvidia.com/cudnn) is a GPU-accelerated library of primitives for 
+deep neural networks. cuDNN provides highly tuned implementations for standard routines.
+
+Typically, models run with cuDNN enabled are faster. However there are some exceptions
+where using cuDNN can be slower, cause higher memory usage or result in errors. 
+
+
+The section of model config file specifying this parameter will look like:
+
+```
+parameters: {
+key: "DISABLE_CUDNN"
+    value: {
+    string_value: "true"
+    }
+}
+```
+
 * `ENABLE_WEIGHT_SHARING`: Boolean flag to enable model instances on the same device to
 share weights. This optimization should not be used with stateful models. If not specified,
 weight sharing is disabled.

src/libtorch.cc

@@ -104,6 +104,7 @@ class ModelState : public BackendModel {
     return enable_jit_executor_pair_;
   }
   bool EnabledInferenceMode() { return enable_inference_mode_; }
+  bool EnabledCudnn() { return enable_cudnn_; }
   bool EnabledCacheCleaning() { return enable_cache_cleaning_; }
 
   bool EnabledWeightSharing() { return enable_weight_sharing_; }
@@ -125,6 +126,9 @@ class ModelState : public BackendModel {
   // Flag to indicate whether inference mode is enabled. Defaults to false.
   bool enable_inference_mode_;
 
+  // Flag to indicate whether cudnn is enabled. Defaults to true.
+  bool enable_cudnn_;
+
   // Flag to indicate whether cache cleaning after each run is enabled.
   // Defaults to false.
   bool enable_cache_cleaning_;
@@ -227,8 +231,9 @@ ModelState::Create(TRITONBACKEND_Model* triton_model, ModelState** state)
 
 ModelState::ModelState(TRITONBACKEND_Model* triton_model)
     : BackendModel(triton_model), enable_optimized_execution_(true),
-      enable_inference_mode_(true), enable_cache_cleaning_(false),
-      enable_weight_sharing_(false), enable_tensor_fuser_pair_({false, true}),
+      enable_inference_mode_(true), enable_cudnn_(true),
+      enable_cache_cleaning_(false), enable_weight_sharing_(false),
+      enable_tensor_fuser_pair_({false, true}),
       enable_jit_profiling_pair_({false, true}),
       enable_jit_executor_pair_({false, true})
 {
@@ -393,6 +398,24 @@ ModelState::ParseParameters()
          " for model instance '" + Name() + "'")
             .c_str());
 
+    // If 'DISABLE_CUDNN' is not present in 'parameters' then no update is made
+    // to 'enable_cudnn_'.
+    bool disable_cudnn = false;
+    err = ParseParameter(params, "DISABLE_CUDNN", &disable_cudnn);
+    if (err != nullptr) {
+      if (TRITONSERVER_ErrorCode(err) != TRITONSERVER_ERROR_NOT_FOUND) {
+        return err;
+      } else {
+        TRITONSERVER_ErrorDelete(err);
+      }
+    }
+    enable_cudnn_ = !disable_cudnn;
+    LOG_MESSAGE(
+        TRITONSERVER_LOG_INFO,
+        (std::string("cuDNN is ") + (enable_cudnn_ ? "enabled" : "disabled") +
+         " for model instance '" + Name() + "'")
+            .c_str());
+
     // If 'ENABLE_TENSOR_FUSER' is not present in 'parameters' then no
     // update is made to 'enable_tensor_fuser'.
     bool enable_tensor_fuser = false;
@@ -1562,6 +1585,9 @@ ModelInstanceState::Execute(
     // enable/disable inference mode - supersedes NoGradGuard
     torch::InferenceMode infer_guard(model_state_->EnabledInferenceMode());
 
+    // enable/disable cudnn
+    at::globalContext().setUserEnabledCuDNN(model_state_->EnabledCudnn());
+
     // JIT. No change is made unless parameter is explicitly set.
     if (std::get<0>(model_state_->EnabledJitProfiling())) {
       torch::jit::getProfilingMode() =