Files
d4r/tools/windows/player_config_probe.cpp
T
ayo a617c818de 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.
2026-10-17 01:43:06 -04:00

165 lines
12 KiB
C++

// Checks the drag-in d4r.ini parser and NVIDIA DLL pin decisions with
// fixtures. No GPU work, no NVIDIA binaries.
#include "player_config.h"
#include <functional>
namespace player = d4r::win::player;
namespace {
int failures = 0;
void expect(bool condition, const std::string& what) {
std::printf("%s %s\n", condition ? "ok " : "FAIL", what.c_str());
if (!condition) ++failures;
}
bool has_note(const std::vector<std::string>& notes, const std::string& text) {
for (const auto& note : notes) if (note.find(text) != std::string::npos) return true;
return false;
}
player::Settings settings(const std::string& ini, std::vector<std::string>& notes) {
notes.clear(); return player::read_settings(player::parse_ini(ini), notes);
}
std::string error_of(const std::function<void()>& fn) {
try { fn(); } catch (const std::exception& error) { return error.what(); }
return {};
}
}
int wmain(int argc, wchar_t** argv) {
if (argc != 2) { std::printf("usage: player_config_probe FIXTURE_DIR\n"); return 2; }
const std::filesystem::path dir(argv[1]);
std::vector<std::string> notes;
auto s = settings("", notes);
expect(s.preset == 11 && !s.asyncInterop && !s.validateOutput && !s.verbose && notes.empty(), "empty d4r.ini uses K, sync, no validation");
s = settings("\xEF\xBB\xBF[DLSS]\nModel = M ; DLSS 4.5\n", notes);
expect(s.preset == 13 && notes.empty(), "BOM and inline comment; Model = M");
s = settings("[dlss]\nmodel = dlss4.5\n", notes);
expect(s.preset == 13, "case-insensitive section/key and alias DLSS4.5");
for (const char* model : {"game", "7"}) {
s = settings(std::string("[DLSS]\nModel = ") + model + "\n", notes);
expect(s.preset == 11 && has_note(notes, "not available on Windows"), std::string("Model = ") + model + " is rejected with a note and K");
}
s = settings("[DLSS]\nModel = E\n", notes);
expect(s.preset == 5 && has_note(notes, "no standalone Windows engine"), "Model = E selects NGX preset 5 with limitation note");
s = settings("[DLSS]\nModel = L\n", notes);
expect(s.preset == 12 && has_note(notes, "not validated"), "Model = L selects NGX preset 12 with limitation note");
s = settings("[DLSS]\nModel = 12\n", notes);
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");
s = settings("[Kernels]\nNativeFp8 = on\n[Latency]\nFrameAge = 1\n", notes);
expect(s.nativeFp8 && !has_note(notes, "[Kernels] NativeFp8") && has_note(notes, "[Latency] FrameAge"),
"native FP8 is parsed while Linux-only settings are reported as ignored");
s = settings("", notes);
expect(!s.showWatermark, "watermark is hidden by default");
s = settings("[DLSS]\nShowWatermark = true\n", notes);
expect(s.showWatermark && notes.empty(), "[DLSS] ShowWatermark = true keeps NVIDIA's stamp");
s = settings("[DLSS]\nShowWatermark = sometimes\n", notes);
expect(!s.showWatermark && has_note(notes, "ShowWatermark must be true or false"), "invalid ShowWatermark falls back to hidden with a note");
s = settings("[Paths]\nCacheDir = D:\\d4r-cache\n", notes);
expect(s.cacheDir == L"D:\\d4r-cache" && notes.empty(), "CacheDir path");
s = settings("[Paths]\nNativeDir = private-native\n", notes);
expect(s.nativeDir == L"private-native" && notes.empty(), "NativeDir path");
expect(player::native_code_object_dir(dir, &s) == dir / L"private-native", "relative NativeDir resolves beside d4r.ini");
expect(!error_of([&] { player::verify_code_objects(dir, &s); }).empty(), "incomplete private native directory rejected");
s.cacheDir = (dir / L"cache").wstring();
player::apply(dir, s);
expect(std::wstring(_wgetenv(L"D4R_ZLUDA_NATIVE_DIR")) == (dir / L"private-native").wstring(),
"player environment selects configured private native directory");
s = settings("[Paths]\nNativeDir = auto\n", notes);
expect(s.nativeDir.empty() && notes.empty(), "NativeDir auto preserves packaged kernels");
s = settings("[RayReconstruction]\nBackend = NATIVE\nModelDir = local-rr\n", notes);
expect(s.rrBackend == "native" && s.rrModelDir == L"local-rr" && notes.empty(), "native RR settings parse independently of SR");
expect(player::rr_model_directory(dir, s) == std::filesystem::absolute(dir / L"local-rr").lexically_normal(), "relative RR model resolves beside d4r.ini");
s.cacheDir = (dir / L"cache").wstring(); player::apply(dir, s);
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 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") && !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"; }
expect(player::compat_nvapi_path(dir) == dir / L"nvapi" / L"nvapi64.dll", "player mode accepts quick-test NVAPI layout");
const auto archNvapi = dir / L"nvapi" / d4r::diag::wide(d4r::diag::HipApi::target_arch()) / L"nvapi64.dll";
std::filesystem::create_directories(archNvapi.parent_path());
{ std::ofstream(archNvapi, std::ios::binary) << "architecture NVAPI fixture"; }
expect(player::compat_nvapi_path(dir) == archNvapi, "player mode prefers architecture-specific release NVAPI");
std::filesystem::create_directories(dir / L"zluda");
const auto colocatedNvapi = dir / L"zluda" / L"nvapi64.dll";
{ std::ofstream(colocatedNvapi, std::ios::binary) << "colocated compatibility NVAPI fixture"; }
expect(player::compat_nvapi_path(dir) == colocatedNvapi, "player mode reuses NVAPI colocated with nvcuda");
std::filesystem::remove(colocatedNvapi);
std::filesystem::remove(archNvapi);
expect(player::raw_nvapi_backend_path(dir) == dir / L"nvapi" / L"backend" / L"nvapi64.dll",
"raw ZLUDA NVAPI backend has a separate explicit path");
const auto archNative = dir / L"native" / d4r::diag::wide(d4r::diag::HipApi::target_arch());
std::filesystem::create_directories(archNative);
expect(player::native_code_object_dir(dir) == archNative, "player mode prefers architecture-specific release kernels");
std::filesystem::remove_all(archNative);
std::filesystem::create_directories(dir / L"native");
expect(player::native_code_object_dir(dir) == dir / L"native", "player mode accepts target-specific flat kernel package");
// Pins: fixture DLL contents are ordinary text; their hashes are computed here.
std::filesystem::create_directories(dir / L"ngx");
std::filesystem::create_directories(dir / L"dlss");
{ std::ofstream(dir / L"dlss" / L"nvngx_dlss.dll", std::ios::binary) << "validated dlss fixture"; }
{ std::ofstream(dir / L"ngx" / L"_nvngx.dll", std::ios::binary) << "unvalidated core fixture"; }
const auto dlss = player::sha256(dir / L"dlss" / L"nvngx_dlss.dll"), core = player::sha256(dir / L"ngx" / L"_nvngx.dll");
const auto pins = player::parse_pins("# comment\ndlss " + dlss + " 310.9.1 validated\ndlssd " + std::string(64, 'B') +
" 310.9.1 unvalidated\nngx " + core + " 32.0.15.9636 unvalidated\n" +
"dlss " + std::string(64, 'A') + " 310.7.0 unvalidated\n");
expect(pins.size() == 4, "installer-only dlssd pin is accepted");
notes.clear();
expect(error_of([&] { player::verify_dll(dir, pins, "dlss", L"dlss\\nvngx_dlss.dll", "nvngx_dlss.dll", notes); }).empty() && notes.empty(),
"validated DLSS pin accepted silently");
expect(error_of([&] { player::verify_dll(dir, pins, "ngx", L"ngx\\_nvngx.dll", "_nvngx.dll", notes); }).empty() &&
has_note(notes, "not yet validated on Windows"), "unvalidated pin accepted with a note");
{ std::ofstream(dir / L"dlss" / L"nvngx_dlss.dll", std::ios::binary) << "changed"; }
auto error = error_of([&] { player::verify_dll(dir, pins, "dlss", L"dlss\\nvngx_dlss.dll", "nvngx_dlss.dll", notes); });
expect(error.find("Wrong nvngx_dlss.dll: you have version") != std::string::npos && error.find("310.9.1 or 310.7.0") != std::string::npos,
"changed DLL rejected with you-have/you-need: " + error);
std::filesystem::remove(dir / L"dlss" / L"nvngx_dlss.dll");
error = error_of([&] { player::verify_dll(dir, pins, "dlss", L"dlss\\nvngx_dlss.dll", "nvngx_dlss.dll", notes); });
expect(error.find("is missing. Copy NVIDIA's nvngx_dlss.dll") != std::string::npos, "missing DLL explains where to copy it");
expect(!error_of([] { player::parse_pins("dlss short 1 validated\n"); }).empty(), "damaged pins file rejected");
expect(!error_of([] { player::parse_pins("other " + std::string(64, 'A') + " 1 validated\n"); }).empty(), "unknown pin kind rejected");
std::printf("%s PLAYER_CONFIG failures=%d\n", failures ? "FAIL" : "PASS", failures);
return failures ? 1 : 0;
}