[NvTensorRTRTX EP] Skip GPU JIT during compile-only sessions

[umangb-09]

Description

Add an internal session config entry, "session.compile_only", set by CompileModel() before
session initialization. The NvTensorRTRTX EP reads it in
NvExecutionProviderInfo::FromProviderOptions() and, when set, skips deserializeCudaEngine() /
createExecutionContext() in CreateNodeComputeInfoFromGraph().

The EP context node is still saved — that path uses the serialized engine buffer directly and does
not depend on the deserialized engine. A stub compute function is registered to satisfy the
framework; it returns NOT_IMPLEMENTED if called, which cannot happen in practice because
compile-only sessions are destroyed without inference.

Motivation and Context

OrtCompileAPI::CompileModel() creates an InferenceSession solely to drive EP::Compile() and
write out the EPContext model, then destroys it without running inference. During that session, the
NvTensorRTRTX EP was performing a full deserializeCudaEngine() and createExecutionContext() —
uploading engine weights to the GPU and JIT-ing the engine, only to free everything when the session
was destroyed.

When the user then loads the EPContext model in a real session, the same JIT and upload happen again.
Net effect on the typical "compile, then load and run" flow:

ONNX model
    → CompileModel()         [JIT + GPU upload #1 — discarded]
    → EP context model saved to disk
    → Session from EP context model
                             [JIT + GPU upload #2 — necessary]
    → Inference

JIT and GPU upload run twice.

[Copilot]

Pull request overview

This PR introduces an internal compile-only session flag (session.compile_only) that is set by OrtCompileAPI::CompileModel() to let EPs (specifically NvTensorRTRTX) skip expensive GPU engine deserialization and execution context creation during compile-only sessions that are never used for inference.

Changes:

• Set internal session config session.compile_only=1 when CompileModel() creates the temporary InferenceSession.
• Plumb the flag into NvTensorRTRTX EP option parsing and provider state (compile_only_mode_).
• In NvTensorRTRTX EP, skip deserializeCudaEngine() / createExecutionContext() during graph compilation and register a stub compute function for compile-only sessions.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
onnxruntime/core/session/utils.cc	Clones session options in CompileModel() and sets session.compile_only for the compilation-only session.
include/onnxruntime/core/session/onnxruntime_session_options_config_keys.h	Adds the internal kOrtSessionOptionCompileOnly config key string.
onnxruntime/core/providers/nv_tensorrt_rtx/nv_execution_provider_info.h	Adds compile_only_mode to EP info struct.
onnxruntime/core/providers/nv_tensorrt_rtx/nv_execution_provider_info.cc	Reads session.compile_only from session config into compile_only_mode.
onnxruntime/core/providers/nv_tensorrt_rtx/nv_execution_provider.h	Adds compile_only_mode_ member to the EP instance.
onnxruntime/core/providers/nv_tensorrt_rtx/nv_execution_provider.cc	Skips GPU deserialization/context creation in compile-only mode and returns a stub NodeComputeInfo.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.