settings: drop the netplay claim from the s16 fastpath sublabel

The 'Resample to Fixed Integer' sublabel introduced alongside the
hint stated that bit-identical resampler output 'helps netplay
determinism'. It does not: the resampler sits downstream of the core,
its output never feeds back into core state, and netplay
synchronises inputs and savestate checksums, not audio. The claim
could mislead users into enabling the hint (or, worse, treating it
as a netplay troubleshooting step) for a benefit it cannot provide.

Replace it with the accurate property: bit-identical output makes
audio reproducible across runs and systems, which is what the
validation/regression workflows rely on. Verified that
intl/h2json.py parses the edited row and emits the corrected string
under MENU_ENUM_SUBLABEL_AUDIO_FASTPATH_S16 for the Crowdin source;
the generated msg_hash_us.json is gitignored, so the def row remains
the only tracked carrier. Source-string change will invalidate
existing translations of this sublabel, as intended.
This commit is contained in:
libretroadmin authored and libretroadmin committed 2026-07-19 22:02:01 +02:00
1 parent c4b38a2811
commit b0aceb81e0
1 file changed
+1 -1
@@ -13,4 +13,4 @@ S_BOOL(audio_fastpath_s16, AUDIO_FASTPATH_S16,
"audio_fastpath_s16",
DEFAULT_AUDIO_FASTPATH_S16, SD_FLAG_ADVANCED, 0, CMD_EVENT_NONE,
"Resample to Fixed Integer (Hint)",
"Use the fixed-point (integer) resampler instead of the floating-point one when a core outputs 16-bit audio. Avoids the integer-to-float round-trip and produces bit-identical output on every platform, which helps netplay determinism. Has no effect on cores that output floating-point audio, and falls back to the floating-point path while an incompatible DSP filter is active.")
"Use the fixed-point (integer) resampler instead of the floating-point one when a core outputs 16-bit audio. Avoids the integer-to-float round-trip and produces bit-identical output on every platform, making audio output reproducible across runs and systems. Has no effect on cores that output floating-point audio, and falls back to the floating-point path while an incompatible DSP filter is active.")