Preview branch that addresses a number of reported bugs

[narendasan]

Description

Previews complex tensor support, rms norm support, and latest main + porting the profiler from the legacy profiler.

Fixes #4137 #4135

Type of change

Please delete options that are not relevant and/or add your own.

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

Checklist:

• My code follows the style guidelines of this project (You can use the linters)
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas and hacks
• I have made corresponding changes to the documentation
• I have added tests to verify my fix or my feature
• New and existing unit tests pass locally with my changes
• I have added the relevant labels to my PR in so that relevant reviewers are notified

[github-actions]

There are some changes that do not conform to Python style guidelines:

--- /home/runner/work/TensorRT/TensorRT/tests/py/dynamo/conversion/test_index_bool_split_aten.py	2026-03-19 00:28:21.633730+00:00
+++ /home/runner/work/TensorRT/TensorRT/tests/py/dynamo/conversion/test_index_bool_split_aten.py	2026-03-19 00:28:42.141701+00:00
@@ -4,10 +4,11 @@
1. `index_has_bool_indices` validator correctly distinguishes bool vs int indices.
2. Integer-indexed `aten.index.Tensor` routes to the converter WITHOUT output allocator.
3. Boolean-indexed `aten.index.Tensor` routes to the converter WITH output allocator.
4. Both paths produce correct results.
"""
+
import unittest
from unittest.mock import MagicMock

import torch
import torch.nn as nn
@@ -58,13 +59,11 @@
        node = _make_index_node([None, torch.tensor([True, False])])
        self.assertTrue(index_has_bool_indices(node))

    def test_mixed_int_and_bool_returns_true(self):
        """If any index is bool, the function should return True."""
-        node = _make_index_node(
-            [torch.tensor([0, 1]), torch.tensor([True, False])]
-        )
+        node = _make_index_node([torch.tensor([0, 1]), torch.tensor([True, False])])
        self.assertTrue(index_has_bool_indices(node))

    def test_all_none_returns_false(self):
        node = _make_index_node([None, None])
        self.assertFalse(index_has_bool_indices(node))

[github-actions]

There are some changes that do not conform to C++ style guidelines:

diff --git a/home/runner/work/TensorRT/TensorRT/core/runtime/execute_engine.cpp b/tmp/changes.txt
index 94264f0..fdc06a1 100644
--- a/home/runner/work/TensorRT/TensorRT/core/runtime/execute_engine.cpp
+++ b/tmp/changes.txt
@@ -412,24 +412,24 @@ std::vector<at::Tensor> execute_engine(std::vector<at::Tensor> inputs, c10::intr
    { // Output Collection
      RECORD_USER_SCOPE("torch_tensorrt_execute_engine::OutputCollection");
      for (size_t i = 0; i < compiled_engine->out_binding_names.size(); i++) {
-      auto name = compiled_engine->out_binding_names[i];
-      auto dims = compiled_engine->output_allocator->getShapes().at(name);
-      auto dtype =
-          util::TRTDataTypeToScalarType(compiled_engine->exec_ctx->getEngine().getTensorDataType(name.c_str()));
-      at::Tensor output = compiled_engine->output_allocator->getBuffers().at(name).clone().detach();
-      int64_t prod = 1;
-      for (int i = 0; i < dims.nbDims; ++i) {
-        prod *= dims.d[i];
-      }
-      std::vector<int64_t> shape(dims.nbDims);
-      for (int i = 0; i < dims.nbDims; ++i) {
-        shape[i] = dims.d[i];
+        auto name = compiled_engine->out_binding_names[i];
+        auto dims = compiled_engine->output_allocator->getShapes().at(name);
+        auto dtype =
+            util::TRTDataTypeToScalarType(compiled_engine->exec_ctx->getEngine().getTensorDataType(name.c_str()));
+        at::Tensor output = compiled_engine->output_allocator->getBuffers().at(name).clone().detach();
+        int64_t prod = 1;
+        for (int i = 0; i < dims.nbDims; ++i) {
+          prod *= dims.d[i];
+        }
+        std::vector<int64_t> shape(dims.nbDims);
+        for (int i = 0; i < dims.nbDims; ++i) {
+          shape[i] = dims.d[i];
+        }
+        // When using the OutputAllocator, the allocated buffer might be larger than the size of the output,
+        // so we need to reshape the buffer to the output shape
+        output = output.reshape(-1).view(dtype).slice(0, 0, prod).reshape(shape);
+        outputs.push_back(output);
      }
-      // When using the OutputAllocator, the allocated buffer might be larger than the size of the output,
-      // so we need to reshape the buffer to the output shape
-      output = output.reshape(-1).view(dtype).slice(0, 0, prod).reshape(shape);
-      outputs.push_back(output);
-    }
    } // End Output Collection

    if (compiled_engine->profile_execution) {
ERROR: Some files do not conform to style guidelines