NR Auto Scale v0.1.0-pre.2 — native-resolution progress

v0.1.0-pre.2 — Native-resolution performance checkpoint

This preview publishes the code behind the latest manual result: approximately
19 base FPS at 2560x1440 on an RX 9070 XT. The complete cadence recording
contains 1,112 changed RGB submissions in 60.238 seconds (18.46/s).
With 2x frame generation, 19 x 2 is nominally 38 output FPS; LSFG/display FPS
was not independently measured. The 60 FPS target is not yet met.

Changes since pre.1

• Native-resolution capture/forwarding, including preserving native geometry
  and WGC routing when the Lossless Scaling profile changes.
• Shared GPU transport, exact RGB duplicate suppression, direct full-strength
  display and reduced redundant copies/comparisons, with existing fences kept.
• Bounded HIP host timing and worker-wait hints that pace the neural feed.
• Version/artifact manifests, research notes, a manual-run ledger and an
  offline analyzer that exports sanitized, source-hashed measurements.

Selected observational intervals show 17.53 -> 18.69 changed RGB/s and
105.49 -> 26.08 GPU feed/check iterations/s. The large drop in redundant work
is not a proportional neural speedup: mean HIP device waits remain approximately
52 ms. These are separate gameplay trials, not controlled A/B benchmarks.
See progress and measurement limits.

What this package is

The package contains the same executable used in the latest manual run:

DlssNrBridge.exe SHA-256:
d902dc89c6977a9012028844da6acd788a2386e2c1bb43da71fa805cd230a85f

The project-built Lossless.dll is also included. AMD/NVIDIA runtimes, model
weights, original paid Lossless Scaling files, private configuration, profiles
and raw logs are excluded. CHECKSUMS.json verifies package entries;
RELEASE.json identifies the project artifacts and source fingerprints.
Supply the existing runtime separately; the exercised upstream version is
v0.2.15, build f47996ae.

For an existing installation, stop scaling before updating project binaries
and retain the previous binaries for rollback. Do not replace model/runtime
files or restore old profiles. Fresh setup remains fixed 1280x720 by default;
use Setup.cmd -NativeResolution 1 for the native path measured here. Existing
NativeResolution=1 installations already follow the captured application size.

Verification and remaining work

Production compilation and artifact/hash verification passed. The user manually
exercised the native shared-GPU and completion-aware feed path. Full hardware
pixel equivalence, general game compatibility, end-to-end latency, and a 60 FPS
result are not established. No tests, synthetic scenes, GPU benchmarks or
playback were run by the agent for this release. Native dimensions, model,
precision, effect strength, Inline=0 and the installed profile stay unchanged.

Next work is attribution of the remaining neural GPU cost and version-tagged
manual comparison, not further claims based on repeated presentations.

Source commit: 2762c22

NR Auto Scale v0.1.0-pre.1 — experimental preview

Experimental preview of NR Auto Scale for Windows x64.

Download the ZIP, extract it outside the Lossless Scaling installation, close Lossless Scaling, and run Setup.cmd. Supply your own AMD compatibility proxy and NVIDIA NR runtime files; install the required HIP/runtime prerequisites described in the documentation.

• Native-resolution bridge mode is the default; no geometry upscaling is requested.
• Ctrl+Alt+F6 toggles NR, F7 decreases blend, and F8 increases blend.
• The README includes matching crops from a frozen 2560x1440 CS2 frame. Crops contain no account details and have no resizing or color edits.
• Native bridge pixels are verified. Automatic Scale/Unscale was tested with the earlier fixed-size path; the final native-resolution Lossless Scaling end-to-end test is incomplete. This is a preview for testing, not a stable release.
• Native neural jobs took about 63 ms each on the tested GPU. This is not recommended for competitive CS2.

The ZIP includes only this project's binaries, scripts, documentation, notices, and approved comparison crops. No paid Lossless Scaling application files, AMD/NVIDIA runtime DLLs, models, logs, settings, or personal files are included. Runtime licenses still apply; the project does not grant NVIDIA/AMD runtime permissions.