TL;DR: Add an experimental FSR Ray Regeneration backend on Windows, run the K

network natively on HIP with Vulkan->HIP shared memory, extend K kernels to
gfx11 (RDNA3), and add dependency/packaging/test tooling around them.

Windows: FSR RR backend, native HIP K runner, gfx11 K kernels, tooling

FSR Ray Regeneration (experimental)
- Add fsr_rr_backend.{h,cpp}, fsr_rr_passes.h and the fsr_rr/ prepare and
  compose compute shaders. They run AMD's FidelityFX denoiser
  (amd_fidelityfx_denoiser_dx12.dll, SDK 2.3, not bundled) on a private
  vkd3d-proton device.
- Add the amdxc64 shim (tools/windows/amdxc64_shim) and two vkd3d-proton
  patches for libcxx includes. AMD's amdxc64.dll is never loaded.
- d4r_nvngx_windows: RrFeature now selects between the native DLSS RR
  backend and the FSR backend (D4R_RR_BACKEND=fsr). After the denoiser it
  runs an internal Super Sampling stage (feature 1) to upscale. If SR is
  unavailable and output != render size, the frame fails; it is never
  returned noisy.
- d4r.ini / player_config: new [RayReconstruction] FsrDenoiser,
  FsrStabilityBias (auto or >= 0) and FsrDetail (0-1), with validation, and
  matching D4R_FSR_* environment variables.
- Add the FSR RR fixture, output-check and ngx test scripts, and a shader
  compile script.

Native HIP K runner (Windows)
- CMake: add d4r_engine_k_hip (d4r-k-hip) plus separate volk/engine copies,
  because VK_USE_PLATFORM_WIN32_KHR changes volk's entry points.
- Share Vulkan memory with HIP via Win32 opaque handles on Windows
  (VK_KHR_external_memory_win32) and via fds on Linux. Handle types are now
  intptr_t, and the platform choice is in D4R_EXT_MEMORY_* macros.
- HipNetwork: add debugRepeatHashes and debugNodeMicros, exposed as
  D4R_HIP_REPEAT_CHECK, D4R_HIP_REPEAT_ALL and D4R_HIP_NODE_PROFILE in
  replay.cpp.
- replay.cpp: display-resolution-motion frames (FrameFlagDisplayMotion) no
  longer require near2rnd == out2rnd. Both values must still be in (0, 1].

K kernels on gfx11
- Allow position-token and enc1 tiling on __GFX11__ as well as gfx12.
- Add an opt-in ds_bpermute cross-half exchange
  (D4R_WINDOWS_GFX11_DS_EXCHANGE, enabled for K on Windows gfx11). M keeps
  its LDS path.
- Fix a block-barrier race in the position layer. Without POS_VSHARE a fast
  wave could overwrite its hb rows while a slower wave still read them for a
  later head, giving rare wrong results on gfx1101 (enc0 outputs differing
  between identical launches).

Packaging and install
- install.bat, d4r.ini, WINDOWS_README: FSR option, and
  install_dependencies.py to install pinned NVIDIA dependencies with staged,
  hash-verified replacement and rollback.
- Update the release package, local package refresh and test scripts.
- Add the InteropBatchInputCopies option (default false). Batching is faster
  on some synthetic runs, but player temporal exactness is unresolved.

Tooling and tests
- K model: capture_k_weights, build_k_model, compare_k_outputs,
  benchmark_k_hip, build_capture_runtime, test_k_model and the
  zluda-k-output-capture patch. Captured weights and models are private
  local assets and must not enter release archives.
- Synthetic probes: test_m_ngx, test_sr_dynamic,
  prepare_rr_dynamic_fixture and the d3d12/rr_ngx probe updates.
- Add tests/windows/test_install_dependencies.py.

Docs
- Update the windows-fsr-rr, k-performance, local-package, rr-interop and
  rr-model-cache docs. Add windows-parity-handoff.md and
  windows-parity-continuation-prompt.md.

tools/windows/player_config_probe.cpp

@@ -46,6 +46,16 @@ int wmain(int argc, wchar_t** argv) {
     expect(s.preset == 12 && has_note(notes, "not validated"), "Model = 12 selects NGX preset 12 with limitation note");
     s = settings("[Interop]\nAsyncInterop = yes\n[Debug]\nValidateOutput = on\nLog = verbose\n", notes);
     expect(s.asyncInterop && s.validateOutput && s.verbose && notes.empty(), "boolean spellings and Log = verbose");
+    s = settings("[Interop]\nBatchInputCopies = yes\n", notes);
+    expect(s.batchInputCopies && notes.empty(), "input batching selected through player INI");
+    s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
+    expect(std::getenv("D4R_BATCH_INPUT_COPIES") != nullptr, "player environment enables input batching");
+    s = settings("[Interop]\nBatchInputCopies = invalid\n", notes);
+    expect(!s.batchInputCopies && has_note(notes, "BatchInputCopies must be true or false"), "invalid batching falls back to unbatched with a note");
+    s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
+    expect(std::getenv("D4R_BATCH_INPUT_COPIES") == nullptr, "player environment clears earlier input batching");
+    s = settings("", notes);
+    expect(!s.batchInputCopies && notes.empty(), "older INIs retain unbatched behavior");
     s = settings("[Debug]\nValidateOutput = maybe\nLog = loud\n", notes);
     expect(!s.validateOutput && !s.verbose && has_note(notes, "must be true or false, not 'maybe'") &&
         has_note(notes, "must be normal or verbose"), "invalid values fall back with plain notes");
@@ -77,11 +87,30 @@ int wmain(int argc, wchar_t** argv) {
     expect(std::string(std::getenv("D4R_RR_BACKEND")) == "native" &&
         std::filesystem::path(_wgetenv(L"D4R_RR_MODEL_DIR")) == player::rr_model_directory(dir, s), "native RR environment selected without a launcher");
     s = settings("[RayReconstruction]\nBackend = fsr\n", notes);
-    expect(s.rrBackend == "fsr" && has_note(notes, "FSR RR is unavailable"), "unavailable FSR selection remains explicit");
+    expect(s.rrBackend == "fsr" && has_note(notes, "FSR RR is experimental"), "FSR selection is reported as experimental");
+    s = settings("[RayReconstruction]\nBackend = fsr\nFsrDenoiser = amd\\denoiser.dll\n", notes);
+    s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
+    expect(std::string(std::getenv("D4R_RR_BACKEND")) == "fsr" &&
+        std::filesystem::path(_wgetenv(L"D4R_FSR_RR_DLL")) == std::filesystem::absolute(dir / L"amd" / L"denoiser.dll").lexically_normal() &&
+        std::filesystem::path(_wgetenv(L"D4R_FSR_VKD3D_CORE")) == dir / L"fsr" / L"d4r_vkd3d_core.dll" &&
+        std::filesystem::path(_wgetenv(L"D4R_FSR_AMDXC_SHIM")) == dir / L"fsr" / L"amdxc64.dll" &&
+        std::filesystem::path(_wgetenv(L"D4R_FSR_CACHE_DIR")) == dir / L"cache" / L"vkd3d" && !std::getenv("D4R_RR_MODEL_DIR"),
+        "FSR environment selected from d4r.ini without a launcher");
+    s = settings("[RayReconstruction]\nBackend = fsr\nFsrStabilityBias = 0\nFsrDetail = 0.25\n", notes);
+    s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
+    expect(notes.empty() || has_note(notes, "experimental"), "FSR tuning keys are recognized");
+    expect(std::string(std::getenv("D4R_FSR_RR_STABILITY_BIAS")) == "0" && std::string(std::getenv("D4R_FSR_RR_DETAIL")) == "0.25",
+        "FSR tuning reaches the environment");
+    s = settings("[RayReconstruction]\nBackend = fsr\nFsrStabilityBias = auto\nFsrDetail = 2\n", notes);
+    s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
+    expect(has_note(notes, "FsrDetail must be a number from 0 to 1") && !std::getenv("D4R_FSR_RR_STABILITY_BIAS") && !std::getenv("D4R_FSR_RR_DETAIL"),
+        "auto bias leaves AMD's default and an out-of-range detail is rejected");
     s = settings("[RayReconstruction]\nBackend = typo\n", notes);
     expect(s.rrBackend == "off" && has_note(notes, "Backend must be"), "invalid RR backend stays off");
     s = settings("", notes); s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
-    expect(!std::getenv("D4R_RR_BACKEND") && !std::getenv("D4R_RR_MODEL_DIR"), "default RR off clears stale selection");
+    expect(!std::getenv("D4R_RR_BACKEND") && !std::getenv("D4R_RR_MODEL_DIR") && !std::getenv("D4R_FSR_RR_DLL") &&
+        !std::getenv("D4R_FSR_VKD3D_CORE") && !std::getenv("D4R_FSR_AMDXC_SHIM") && !std::getenv("D4R_FSR_CACHE_DIR") &&
+        !std::getenv("D4R_FSR_RR_STABILITY_BIAS") && !std::getenv("D4R_FSR_RR_DETAIL"), "default RR off clears stale selection");
 
     std::filesystem::create_directories(dir / L"nvapi");
     { std::ofstream(dir / L"nvapi" / L"nvapi64.dll", std::ios::binary) << "generic nvapi fixture"; }