Add a native D3D12 FSR Ray Regeneration route for gfx1200/gfx1201, using the game's device and queue with embedded HLSL guide preparation and output composition. Preserve the vkd3d compatibility route for RDNA3 and expose explicit route overrides and diagnostics. Improve Windows runtime compatibility: - Retry rejected application-ID initialization with a project identity. - Defer expensive RR backend creation until feature creation. - Recover from transient submission failures without disabling later frames. - Add shared-buffer export fallbacks through explicit heaps and a private device on the same adapter. - Safely evict old RR descriptor sets when input views change. - Treat requested RR presets as hints when a configured model is available. - Expand SR/RR input formats and handle diffuse/specular hit distances, packed normals, depth-stencil resources, and unused FSR guides. Reduce SR overhead by batching output downloads and format conversion under one completion wait. Optimize strict M enc2/tube attention with bounded FP32 accumulation and the original FP64 fallback for ambiguous rounding cases. Extend RDNA4 tooling with gfx1200/gfx1201 native FP8 RR model builds, capability checks, architecture-aware deployment, and experimental K/M kernel paths. Require matching model and runtime arithmetic families. Add optional embedded FSR shader optimizations keyed by shader hash and exact original SPIR-V contents. Preserve reflected metadata and record optimizer provenance, artifact hashes, and cache identity. Simplify Windows installation with automatic backend dependency downloads, download progress and ETA, backed-up absolute OptiScaler library paths, and recovery from interrupted RR model builds. Add build-kernels.bat and a kernels-only build matrix for all eleven registered RDNA targets, with code-object target validation and SHA256 manifests. Expand conversion probes, numerical stress tests, temporal benchmark tooling, installer/deployment tests, and setup documentation. Document synthetic regression and timing results, experimental build options, and remaining RDNA4 hardware validation requirements.
151 lines
7.4 KiB
Python
151 lines
7.4 KiB
Python
"""Verify and deploy a locally built RR model; optionally select it in d4r.ini.
|
|
|
|
Existing models are preserved. Configuration changes receive an external backup.
|
|
No NVIDIA DLL, compiler, driver, package archive or system setting is changed.
|
|
"""
|
|
import argparse
|
|
import hashlib
|
|
import json
|
|
import os
|
|
from pathlib import Path
|
|
import re
|
|
import shutil
|
|
import uuid
|
|
|
|
|
|
def sha(data):
|
|
return hashlib.sha256(data).hexdigest()
|
|
|
|
|
|
def select_native(data, model):
|
|
bom = data.startswith(b'\xef\xbb\xbf')
|
|
text = data.decode('utf-8-sig')
|
|
newline = '\r\n' if '\r\n' in text else '\n'
|
|
result, in_rr, found, pending = [], False, False, set()
|
|
values = {'backend': 'native', 'modeldir': model}
|
|
def flush():
|
|
if pending and result and not result[-1].endswith(('\n', '\r')):
|
|
result.append(newline)
|
|
for key in ('backend', 'modeldir'):
|
|
if key in pending:
|
|
result.append(('Backend' if key == 'backend' else 'ModelDir') + ' = ' + values[key] + newline)
|
|
pending.clear()
|
|
for line in text.splitlines(keepends=True):
|
|
stripped = line.strip()
|
|
if stripped.startswith('['):
|
|
if in_rr:
|
|
flush()
|
|
in_rr = stripped.split(']', 1)[0][1:].strip().casefold() == 'rayreconstruction'
|
|
if in_rr:
|
|
found = True
|
|
pending = set(values)
|
|
if in_rr and '=' in line and not stripped.startswith((';', '#')):
|
|
key = line.split('=', 1)[0].strip().casefold()
|
|
if key in values:
|
|
prefix, body = line.split('=', 1)
|
|
ending = '\r\n' if body.endswith('\r\n') else '\n' if body.endswith('\n') else ''
|
|
body = body.removesuffix(ending) if ending else body
|
|
comment = re.search(r'\s[;#]', body)
|
|
suffix = body[comment.start():] if comment else ''
|
|
result.append(prefix + '= ' + values[key] + suffix + ending)
|
|
pending.discard(key)
|
|
continue
|
|
result.append(line)
|
|
if in_rr:
|
|
flush()
|
|
if not found:
|
|
if result and not result[-1].endswith('\n'):
|
|
result.append(newline)
|
|
result.extend((newline, '[RayReconstruction]' + newline, 'Backend = native' + newline,
|
|
'ModelDir = ' + model + newline))
|
|
return (b'\xef\xbb\xbf' if bom else b'') + ''.join(result).encode('utf-8')
|
|
|
|
|
|
def model_arithmetic(identity, runtime):
|
|
"""Check the model's arithmetic and target before copying any products."""
|
|
architecture = identity.get('architecture')
|
|
arithmetic = {'gfx1101': 'gfx11', 'gfx1200': 'gfx12', 'gfx1201': 'gfx12'}.get(architecture)
|
|
if identity.get('preset') not in ('D', 'E') or not arithmetic or identity.get('arithmetic') != 'exact-' + arithmetic:
|
|
raise RuntimeError('Unsupported local RR cache identity')
|
|
target_file = runtime / 'gpu-target.json'
|
|
if target_file.is_file():
|
|
if json.loads(target_file.read_text(encoding='utf-8-sig')).get('architecture') != architecture:
|
|
raise RuntimeError('RR model architecture does not match runtime target')
|
|
elif arithmetic == 'gfx12':
|
|
raise RuntimeError('RDNA4 RR deployment requires runtime gpu-target.json')
|
|
return arithmetic
|
|
|
|
|
|
def main():
|
|
parser = argparse.ArgumentParser(description=__doc__)
|
|
parser.add_argument('--model', required=True, type=lambda p: Path(p).resolve())
|
|
parser.add_argument('--runtime', required=True, type=lambda p: Path(p).resolve())
|
|
parser.add_argument('--enable', action='store_true')
|
|
parser.add_argument('--backup-root', type=lambda p: Path(p).resolve(),
|
|
default=Path(os.environ.get('LOCALAPPDATA', str(Path.home()))) / 'd4r/backups')
|
|
args = parser.parse_args()
|
|
config = args.runtime / 'd4r.ini'
|
|
if not config.is_file():
|
|
parser.error('--runtime must contain d4r.ini')
|
|
if args.enable and args.backup_root.resolve().is_relative_to(args.runtime.parent):
|
|
parser.error('configuration backups must stay outside the game/runtime parent directory')
|
|
cache_data = (args.model / 'windows-native-python-cache.json').read_bytes()
|
|
cache = json.loads(cache_data)
|
|
identity = cache['identity']
|
|
arithmetic = model_arithmetic(identity, args.runtime)
|
|
products = cache['products']
|
|
if not {'rr.txt', 'enc0.args', 'hkpn.args'}.issubset(products):
|
|
raise RuntimeError('RR cache omits required manifest/templates')
|
|
for name, expected in products.items():
|
|
if Path(name).name != name or name in ('.', '..') or '\\' in name or ':' in name:
|
|
raise RuntimeError('Invalid model product path: ' + name)
|
|
if sha((args.model / name).read_bytes()) != expected:
|
|
raise RuntimeError('Model product changed: ' + name)
|
|
manifest = (args.model / 'rr.txt').read_text()
|
|
stages = [line.split() for line in manifest.splitlines() if line.startswith('stage ')]
|
|
expected_stages = {'enc0', 'enc1', 'enc2', 'enc3', 'enc4', 'enc5', 'dec4', 'dec3', 'dec2', 'dec1', 'dec0', 'hkpn'}
|
|
if (not manifest.startswith('D4RR0001\n') or
|
|
f'preset {4 if identity["preset"] == "D" else 5}' not in manifest.splitlines() or
|
|
len(stages) != 12 or {fields[1] for fields in stages} != expected_stages or
|
|
f'arithmetic {arithmetic}' not in manifest.splitlines()):
|
|
raise RuntimeError('Incomplete RR model manifest')
|
|
for fields in stages:
|
|
if fields[2] not in products or fields[4] not in products:
|
|
raise RuntimeError('Stage product not covered by cache')
|
|
name = 'rr-' + identity['preset'].lower() + '-' + sha(cache_data)[:12]
|
|
destination = args.runtime / 'engine' / name
|
|
if not destination.resolve().is_relative_to(args.runtime):
|
|
raise RuntimeError('RR destination must stay within the selected runtime')
|
|
reused = destination.exists()
|
|
if not reused:
|
|
destination.mkdir(parents=True)
|
|
for product in products:
|
|
shutil.copy2(args.model / product, destination / product)
|
|
(destination / 'd4r-private-rr.json').write_text(json.dumps({'privateLocalModel': True, 'cacheSha256': sha(cache_data), 'products': products}, indent=2) + '\n')
|
|
for product, expected in products.items():
|
|
if sha((destination / product).read_bytes()) != expected:
|
|
raise RuntimeError('Existing/deployed RR model differs; preserving it: ' + str(destination))
|
|
backup = None
|
|
if args.enable:
|
|
original = config.read_bytes()
|
|
updated = select_native(original, destination.relative_to(args.runtime).as_posix())
|
|
if updated != original:
|
|
args.backup_root.mkdir(parents=True, exist_ok=True)
|
|
backup = args.backup_root / ('rr-config-' + uuid.uuid4().hex + '.ini')
|
|
with backup.open('xb') as stream:
|
|
stream.write(original)
|
|
temporary = config.with_name('d4r.ini.' + uuid.uuid4().hex + '.tmp')
|
|
with temporary.open('xb') as stream:
|
|
stream.write(updated)
|
|
if config.read_bytes() != original:
|
|
temporary.unlink()
|
|
raise RuntimeError('d4r.ini changed during deployment; preserved current configuration')
|
|
os.replace(temporary, config)
|
|
print(json.dumps({'deployed': str(destination), 'preset': identity['preset'], 'productsVerified': len(products),
|
|
'reused': reused, 'enabled': args.enable, 'configBackup': str(backup) if backup else None,
|
|
'privateLocalModel': True, 'published': False}, indent=2))
|
|
|
|
|
|
if __name__ == '__main__':
|
|
main()
|