Repository navigation
Release 0.6.5: version 0.6.5 and its release notes - #48
Merged
Merged
Conversation
- Version defaults 0.6.0 -> 0.6.1: the front end and WrongGPU.asi (src/version.h), the launcher (WRONGGPU_VERSION), WrongGPU-RR.asi (release_version of the 2.7, 2.8 and 2.14 binding tables) and the RR runtime's reported version. - One version source for a release build: CMake -DWG_VERSION=x.y.z for the front end and WrongGPU.asi (writes wg_version_override.h), build.py --release-version x.y.z for the RR plugin. Without them the defaults are used. - clean/installer/RELEASE-NOTES.md for 0.6.1: the fixes since 0.6.0 (#7, #8, #9, #14 and the launcher issues), thanks to j0nsh1n for the reports, and the 0.6.0 notes carried over without performance figures (no 0.6.1 measurement backs them). Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
- #21, #22 and #23 move to the next release: their lines are out of the notes. - No claim of a 0.6.1 game test until it has a record; the tested-games list points to COMPATIBILITY.md for the version each result comes from. - Requirements name the RX 9060 as well, as Known issues do. Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
phantoslayer21
force-pushed
the
release-0.6.1
branch
from
November 13, 2026 19:14
0cf0b5d to
566db78
Compare
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com> # Conflicts: # clean/fsr/CMakeLists.txt
- Version defaults 0.6.1 -> 0.6.5 in the places the 0.6.1 commit set: the front end and WrongGPU.asi (src/version.h), the launcher (WRONGGPU_VERSION), WrongGPU-RR.asi (release_version of the 2.7, 2.8 and 2.14 binding tables, and its test) and the RR runtime's reported version. Comments that named 0.6.1 as the first release with the one version source say 0.6.5 (0.6.1 is not released). - clean/installer/RELEASE-NOTES.md for 0.6.5: Model L (off by default; converted on the PC from a DLSS file 310.5 or newer; model convert --preset=L, set <game> model=L, OptiScaler's WrongGPU Model L; Direct3D 12; 2.25 ms against Model M's 1.17 ms in Cyberpunk 2077 at 1080p Quality, from clean/l/SPEED.md and its record), its known limit (the SDR path untested in a game), the tested games, and the fixes planned for 0.6.1. Refs #89 Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Every game tried (Cyberpunk 2077, RESIDENT EVIL requiem with HDR off, LEGO Batman, Ratchet & Clank: Rift Apart) hands the upscaler HDR colour, so the SDR path (#176) has offline records only; the Known issues line says so plainly. Refs #89 Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
… player features (pending R1) Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…e as debugging options, off by default Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…st; the old Preset K note goes Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…ecord Model K's network takes 1.01 ms per frame at 1080p Quality on a Radeon RX 9070 XT (walking, the runtime's D3D12 timer, median of 9 windows of 300 frames). The record is clean/k/records/cp2077-model-k-timing-2026-11-15.json. It was a different session and scene from the Model L / Model M figures, and the notes say so. The README's model section now describes Model L and Model K as 0.6.5's per-game choices, off by default. Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…'s builds Model L ships in a new form and Model K's figure was not measured the same way as the others, so the Model L, Model M and Model K times are marked pending. They come back from in-game measurements of the release builds, with their records, before the release. Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
… with the record In Cyberpunk 2077 at 1080p Quality on a Radeon RX 9070 XT, standing still at the same spot and measured one right after the other, the model's GPU time per frame is 1.03 ms with Model K and 1.17 ms with Model M. The record is clean/k/records/cp2077-model-k-timing-2026-11-15.json. Model L's time stays pending until it is measured in this release's form. Model K is added to the Cyberpunk 2077 line of the tested games. Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…0.6.1 Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
k_approx=a1,a2 (runtime_options) selects close approximations of Model K's input stage (logarithm, exponent, divisions) and output stage (divisions). Model K, 1440p Quality, test harness (direct route): 1.37 ms per frame with k_approx=a1,a2 and 1.58 ms on the exact path of the same build; rated very, very close on both test sets with no drift. Off by default (exact Model K), native Direct3D 12 route only; frames on the legacy input path keep the exact input stage. Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…g check's terms Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Model L (the hybrid form, the default on Direct3D 12) in Cyberpunk 2077 at 1080p Quality on a Radeon RX 9070 XT: 2.67 ms per frame, with Model M at 1.17 ms in the same sitting (record clean/l/records/cp2077-model-l-timing-2026-11-16.json). README, release notes and clean/l/SPEED.md carry it; SPEED.md's harness section now shows Model L's 1440p Quality figure from the hybrid record and keeps Model M's earlier harness rows in their own table. Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
… Model M's earlier harness record dated Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…release-0.6.1 Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…s) into release-0.6.1 Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
…measured), known issue: rare crash at game start Signed-off-by: phantoslayer21 <156617979+phantoslayer21@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Version 0.6.5 and its release notes. 0.6.1 was skipped; 0.6.5 ships Models M, L and K and Ray Reconstruction (milestone 0.6.5, tracked in the release blocker list). Draft: merges last, after the in-game test of the release payload.
WrongGPU.asi(clean/fsr/src/version.h), the launcher(
WRONGGPU_VERSION),WrongGPU-RR.asi(release_versionof the 2.7, 2.8 and 2.14 binding tables;test_gen_forwardingexpects it) and the RR runtime's reported version.-DWG_VERSION=x.y.zfor the front end andWrongGPU.asi,build.py --release-version x.y.zfor the RR plugin (as before; comments now say "since 0.6.5").clean/fsr/CMakeLists.txt(the release'sWG_VERSIONblock and main'sWG_MODEL_Loption at the same place):both kept.
clean/installer/RELEASE-NOTES.mdfor 0.6.5:(
model convert --preset=L); chosen withset <game> model=Lor OptiScaler's WrongGPU Model L; Direct3D 12.the player's games (
model convert --preset=K); chosen withset <game> model=Kor OptiScaler's WrongGPU ModelK; Direct3D 12, tested in Cyberpunk 2077 at 1080p Quality. An older install on Preset K is no longer asked to
switch to Model M (the prompt is compiled out in a build with Model K).
discarded by uninstall / update / repair / relocation (Uninstall, Update, Repair and relocation can discard a backup that was meant to be kept #70, A hand-edited wronggpu.ini is reported "Left in place" and then deleted #73), coexistence with another OptiScaler setup (Launcher: coexist with other OptiScaler setups #23),
RR memory released after the GPU's last frame (RR releases its descriptor heap and query heap at destroy without waiting for the GPU #63), RR off when another RR plugin is identified (Ray Reconstruction plugin: stay off when another Ray Reconstruction plugin is already active #22), RR under
Proton (RR runtime: compile thread dies of a stack overflow under Proton with current vkd3d-proton (fix included) #149, thanks to j0nsh1n for RR runtime: give the pipeline compile thread a 64 MB stack; log failed pipelines #196).
executor=hipand frame capture as debugging options that use HIP, both off by default. The launcher settings that still use HIP (sparkle guard, Quality 5 / 9 output filters, all off by default) are not advertised.tested offline only, because every game we tried sends HDR colour. Tested games list Model L for both.
Number gate (every figure in the new text):
clean/l/SPEED.mdin-game table →clean/l/records/cp2077-model-l-timing-2026-11-14.json(network total, gameplay median 2.250 / 1.174).1440p Quality 4.00 / 2.17) →
clean/l/records/l-vs-m-shapes-main.json.clean/installer/src/wg_core/model.cpp).(9c0fcc8); the only new numbers are issue and PR numbers, all in the notes.
replaced if it differs.
Not in the notes yet: these items come only once they are merged:
--model-lwithWRONGGPU_MODEL_L/WG_MODEL_Lon (blocker A4).Tested: public gates on the branch head. CI runs on push.