ayo
34e6b0fe71
Add an experimental FSR Ray Regeneration backend for D3D12 RR requests: - Convert depth, motion, normals and albedos into AMD denoiser inputs. - Split and recompose diffuse/specular signals, then run Super Sampling. - Add stability/detail controls, demodulation guards and halo limiting. - Handle AMD provider discovery under NVIDIA GPU spoofing. - Add runtime probes, synthetic reference scenes and per-dispatch profiling. Add exact gfx11 DLSS-RR inference using f16 cooperative matrices: - Reproduce FP8 conversions, accumulation and camera reprojection arithmetic. - Widen model weights and avoid requiring VK_EXT_shader_float8. - Extend model building, replay tools and descriptor metadata. Expand motion preprocessing and experimental AW2 native-motion support: - Add tiled LDS dilation, padded allocation handling and thin-only sky fill. - Order dilation with input waits and NGX on the selected stream. - Add jitter-aligned native composition, hybrid routing and wire history repair. - Add prepared masks, sparse repair and independent motion/repair controls. Add optional fused R11G11B10 input/output conversion, sampled stage timings and profile summarization. Extend configuration, packaging, documentation, FidelityFX headers and CPU/GPU regression coverage.
49 lines
2.7 KiB
Python
49 lines
2.7 KiB
Python
"""Keep startup, feature instances and resolution changes out of shared medians."""
|
|
import importlib.util
|
|
from pathlib import Path
|
|
import unittest
|
|
|
|
ROOT = Path(__file__).resolve().parents[1]
|
|
spec = importlib.util.spec_from_file_location('profile_summary', ROOT/'scripts/summarize_d4r_profile.py')
|
|
summary = importlib.util.module_from_spec(spec)
|
|
spec.loader.exec_module(summary)
|
|
|
|
|
|
class ProfileSummaryTests(unittest.TestCase):
|
|
def test_warm_groups_and_unavailable_timings(self):
|
|
def sample(frame, render='853x480', reset=0, gpu='2.2'):
|
|
return (f'D4R_PROFILE frame={frame} render={render} output=2560x1440 '
|
|
f'preset=11 quality=3 flags=0x43 input_wait=cpu reset={reset} '
|
|
f'vk_total=2.8 dlss_gpu={gpu} native_motion_gpu=-1 wire_repair_gpu=-1')
|
|
create='NVSDK_NGX_D3D12_CreateFeature(feature=1)'
|
|
log='\n'.join((create, sample(1), sample(300, reset=1), sample(600),
|
|
sample(900, gpu='nan'), sample(1200, render='1706x960'),
|
|
create, sample(300)))
|
|
rows=summary.summarize(log)
|
|
self.assertEqual(len(rows), 3)
|
|
self.assertEqual([r['samples'] for r in rows], [2, 1, 1])
|
|
self.assertEqual([r['feature_instance'] for r in rows], [1, 1, 2])
|
|
self.assertEqual(rows[0]['timings_ms']['dlss_gpu']['n'], 1)
|
|
self.assertIsNone(rows[0]['timings_ms']['wire_repair_gpu'])
|
|
self.assertEqual(rows[0]['timings_ms']['vk_total']['median'], 2.8)
|
|
|
|
def test_routing_changes_get_separate_groups(self):
|
|
base = ('D4R_PROFILE frame=300 render=853x480 output=2560x1440 '
|
|
'preset=11 quality=3 flags=0x43 input_wait=cpu reset=0 '
|
|
'feature=29184 native_field=1 wire_requested=1 gpu_completion=0 ')
|
|
rows = summary.summarize(base+'native_substitute=0\n'+base+'native_substitute=1')
|
|
self.assertEqual(len(rows), 2)
|
|
self.assertEqual([r['native_substitute'] for r in rows], ['0', '1'])
|
|
|
|
def test_scene_window_and_actual_output_mode(self):
|
|
base = ('D4R_PROFILE frame=600 render=853x480 output=2560x1440 '
|
|
'preset=11 quality=3 flags=0x43 input_wait=cpu reset=0 ')
|
|
log = ('[ 10.000] '+base+'vk_total=8 vk_output_mode=transfer\n'
|
|
'[ 20.000] '+base+'vk_total=3 vk_output_mode=compute\n'
|
|
'[ 21.000] '+base+'vk_total=4 vk_output_mode=transfer\n'
|
|
'[ 30.000] '+base+'vk_total=9 vk_output_mode=compute')
|
|
rows = summary.summarize(log, since_clock=19, until_clock=22)
|
|
self.assertEqual(len(rows), 2)
|
|
self.assertEqual([r['vk_output_mode'] for r in rows], ['compute', 'transfer'])
|
|
self.assertEqual([r['timings_ms']['vk_total']['median'] for r in rows], [3, 4])
|