Repository navigation
feat: smoother play, HDR10, a frame rate cap, Discord status, the Debug tab, sharper shadows - #1485
Merged
Merged
Conversation
…ug tab, sharper shadows Tested together in the v2.0.12 candidate builds (verify215-219: flash checks, smoothness walks, an 8 GB card simulated). - Stutter: a frame's uploads have a budget (40 ms): new ASTC textures a draw samples past it come a frame or more later (RX_VK_UPLOAD_BUDGET_MS); descriptors are rebuilt when a texture is destroyed (RX_VK_DESC_REUSE_DGEN=0: off). - VRAM: on a card of 8 GB or less the textures' budget is 45% of what the driver allows, not 70% (the 3060 Ti/3070 slowdowns); RX_VK_TEST_ALLOWED_MB=<n> simulates a smaller card. - Shadows: depth-compare samplers filter linearly (percentage-closer): shadow edges no longer in stair steps (RX_VK_SHADOW_PCF=0: off). Nearest samplers named in the log; RX_VK_FORCE_LINEAR=1 (issue 1117's A/B). - Translator: Ampere LDSM between subgroup barriers (MEMBAR's subgroup scope); EDEN_NX2_GM_SPLIT/GM_PORT_PROGS. - HDR10 out (Insert menu: HDR Off / On / Native, paper white, peak brightness), applied at once. - A Frame rate cap row (60, 90, 120, 144, 165, 240; HOS_FPS_MAX), read at the start. - Discord status: "Playing Hookshot" with the time played through a local Discord client (Gameplay > Discord status). - The Debug tab: the OoT debug tools (#785) turned on from the menu (HOS_OOT_CHEATS=1 as before). - Diagnostics: the frame trace (RX_VK_FRAME_TRACE=1), slow uploads, self-sampling draws, slow program preparation and staging arena creation in the log; RX_VK_ASYNC_IMPORT=1 (opt-in) and RX_VK_PASS_BARRIER=game (off) to try. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
KrysGG
added a commit
to KrysGG/Hookshot
that referenced
this pull request
Nov 10, 2026
…st, version 12 prompt_face (main) and wide_menus (this PR) both came last in MkwVkHost: main's first, then ours, and the API version goes to 12. The present's path runs for the HDR picture or the menu sides; the RX_VK_SHOT_FX shot keeps main's headless menu on top. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
KrysGG
added a commit
to KrysGG/Hookshot
that referenced
this pull request
Nov 10, 2026
NotRexed#1485 gave the OoT debug tools (NotRexed#785) a tab of their own, DEBUG, with the free camera turned on from the menu. So no Cheats/Mods tab: the Screen filter and Filter strength rows join the Debug tab, the free camera row goes (main's Debug tools has it), and the HUD hiding stays, following the free camera. The menu's width and tab code are main's. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
msxdan
added a commit
to msxdan/Hookshot
that referenced
this pull request
Nov 10, 2026
…ispatch A compute program's pipeline was always built on a worker, and the dispatch that started it waited up to 50 ms for it. On a first visit the workers are busy with graphics pipelines, so each compute program new to the run held its frame for 50 ms and was left out after it: frames of 105-130 ms with two of them, though the driver had the pipeline in its cache from the precompile (since NotRexed#1485 the precompile makes only the compute programs dispatched in earlier runs, so every area new to the player has them). The driver is asked first with VK_PIPELINE_CREATE_FAIL_ON_PIPELINE_COMPILE_REQUIRED_BIT (pipelineCreationCacheControl): it answers at once either way; one it has is made there and the dispatch runs in its frame, one it would have to compile goes to a worker as before. RX_VK_COMPUTE_ASK=0: always to a worker. Measured on v2.0.11-408 with the two commits before this one and RX_VK_GPL=1 (RTX 5070, Linux, shaders prepared, no pipelines, loading into Kokiri Forest and 32 s of walking, one run a side, RX_VK_FRAME_TRACE): slow submits with 10 ms or more of "dispatch" 32 -> 7; in the walk, time lost in frames over 25 ms 1387 -> 504 ms and frames over 50 ms 13 -> 2; in the load 4936 -> 2895 ms, the longest frame 1546 -> 595 ms.
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.
Tested together in the v2.0.12 candidate builds (verify215-219: flash checks, smoothness walks, an 8 GB card simulated).
(RX_VK_UPLOAD_BUDGET_MS); descriptors are rebuilt when a texture is destroyed (RX_VK_DESC_REUSE_DGEN=0: off).
slowdowns); RX_VK_TEST_ALLOWED_MB= simulates a smaller card.
(RX_VK_SHADOW_PCF=0: off). Nearest samplers named in the log; RX_VK_FORCE_LINEAR=1 (issue 1117's A/B).
staging arena creation in the log; RX_VK_ASYNC_IMPORT=1 (opt-in) and RX_VK_PASS_BARRIER=game (off) to try.
🤖 Generated with Claude Code