Its per-frame display + full-VRAM readback stalled the render thread on the
GPU (about a third of its time in an R4 race). PSX_DISPLAY_RING=1 forces it on,
=0 off; unset it follows the render thread (unchanged when it is off).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[video] frame_generation_method = "redraw" (default) | "reprojection",
PSX_FRAME_GEN_METHOD overrides; replaces frame_generation_redraw.
- Redraw runs and allocates nothing of reprojection: no snapshots, depth
image, programs, HUD matching; the cost-spike clamp and the dynamic
resolution load formula apply to reprojection only. Switching to redraw
or turning Smooth motion off frees what reprojection held.
- Uniform locations and PSX_RP_DEBUG/PSX_RP_NOPREV are read once with the
programs.
- Each method gets its own breaker settings, set again on a switch;
frame_gen_hold takes a kind (step down / over budget), not a string.
- The warp's depth is cleared before the cars are drawn (PGXP depth tested
them LEQUAL against it).
- The warp restores the renderer's texture bindings on units 1 and 2 and
never leaves a sampler on texture 0.
- gl_frame_gen_test: redraw asserts nothing of reprojection ran or was
allocated (its phase 0.5 digest matches master's); reprojection runs
check the warp's pixels on a parallax pan (vs the redraw of the pair, no
holes, HUD exact), real frames unchanged, and no warp depth with PGXP.
- FRAME_GENERATION.md: the key, the default, reprojection and its
R4-tuned object rules.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A rival car filling half of a 2P view left fewer than half of the pairs
on the camera, rejecting ~7% of 2P game frames; reprojection redraws the
cars itself, so 30% inliers suffice. Overload holds are 0.1-1 s while
in-between frames are reprojected.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fixes the 2P band near the minimap: a 2D sky gradient that shifts with the
camera was treated as a HUD needle and drawn on top. Object-mode road
corners the camera alone explains are warped as world; only textured
triangles that really moved are redrawn as cars, after their view's first
world triangle.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Dynamic resolution: Smooth motion's share inflates the measured cost
instead of shrinking the period, so guest-bound detection works again
(it had pinned the level at the ceiling).
- Cost estimate: a sample more than 4x the estimate counts as 4x (one
stall no longer stops generation for many seconds); compiling the
reprojection programs is outside the timed frame.
- Reprojection source: the surface snapshotted at the real frame's present
(two slots: newer and older); the HUD and mirror of both lists are
'keep' in the depth image, drawn last, so a HUD drawn after its list
closed is kept too. Tunnel A/G/B worst ratio 2.9 -> 1.3.
- HUD needles: 2D triangles that move between the frames are drawn at
their in-between position; under them the older frame's dial shows.
- Split screen: a view whose camera cannot be fitted keeps the real image
instead of rejecting the frame (FgCamParams.keep_partial).
- Car redraw skips object triangles before the first world triangle (the
sky dome). Redundant base copies skipped when reprojecting.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each in-between frame warps the newer real frame's finished image by the
in-between camera instead of drawing the scene again:
- per game frame: a snapshot of the list's finished image (taken when the
list closes; R4 flips to a buffer before drawing into it, so the live
surface is not a stable source) and a 2x native image of camera 1/z,
pixel kind and view rasterized from the recorded triangles in painter
order (world / car-or-dome / 2D-HUD-small-view);
- per in-between frame: one grid draw (a vertex per native pixel) that
unprojects world pixels with their depth, moves them by the view's
in-between camera (fg_view_affine) and samples the snapshot at full
resolution, nearest wins; cells over depth edges or non-world pixels are
left to the snapshot underneath; 2D/HUD pixels are put back; cars are
drawn again at their own in-between position.
Measured: ~0.3-0.8 ms CPU and ~0.3-2 ms GPU per in-between frame (was 2-4
ms each), A/G/B max out-of-range 0.3-3 %, worst ratio 1.1-1.5.
[video] frame_generation_redraw / PSX_FRAME_GEN_REDRAW=1 keeps the old path.
Planner/holds: dynres 'over budget' no longer stops cheap reprojected
frames; guest-late trips ignored while they are cheap; allocation and
shader compilation are kept out of the cost estimate. frame_gen JSON:
rp_ms, rp_zn, reprojected, np (why a flip planned none); debug buffer 4 KiB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Dynamic resolution budgets the real frames with the share of the game
frame Smooth motion leaves them (gl_renderer_frame_gen_real_share), so the
scale drops until the in-between frames fit the display refresh.
- In-between frames are timed from the game frame's arrival, not from the end
of its composition; the last one per game frame is no longer dropped by a
quarter-interval margin.
- Views under 15 % of the largest (a rear-view mirror) stay as the real frame
drew them: a wrong mirror camera fit threw its lane marks across the
screen as a dashed line.
- HUD: 2D triangles both frames draw with the same key that moved a little
(tachometer needle) are interpolated; changed digits stay as drawn.
- frame_gen JSON: end_ahead / end_bp / end_phase counters.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
In-between frames go on one global grid at the display refresh (or the
VRR panel maximum), never faster, across game frames; the interpolation
phase comes from the host clock, so 90/100/144/165/240 Hz and other rates
that are no multiple of 30 work. Up to 15 generated frames per game frame.
Fix: a display buffer that changed size at the same origin (e.g. the first
frame after a save-state load) left a stale entry that every list closed
with, so no two frames ever matched and Smooth motion stayed off for the
rest of the session. The entry is replaced now. frame_gen JSON reports
why the last flip planned nothing (noplan_why).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Variable refresh (G-Sync/FreeSync, ProMotion): present each frame when it is
ready (swap interval 0). Smooth motion targets the panel's maximum refresh at
the current size (psx_sdl_display_max_refresh) and never exceeds it.
PSX_HOST_REFRESH_HZ / PSX_HOST_REFRESH_MAX_HZ simulate a display for headless
tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review of #561. Guest time now has two entry points (psx_cycles.h): the
CPU charge psx_cpu_charge() (instruction base, load fudge/region/completion,
i-cache refills, BIOS HLE costs), which applies [timing] guest_cycle_scale,
and the unscaled device advance psx_advance_cycles() (DMA, idle skips,
mul/div and GTE deadline waits). The emitters' block-mode charges emit
psx_cpu_charge; overlay DLLs accumulate it raw and publish through the
cpu_charge callback (was advance_cycles, same slot) into the same host
function, so nothing is scaled at the DLL boundary. The batched
psx_cyc_charge path, the dirty-RAM interpreter, psx_cyc_readmem, i-cache
refills and bios_hle.c all charge through it. At scale 1 it is exactly
psx_advance_cycles.
Gate in psxrecomp: [timing] guest_cycle_scale_gate, RAM predicates judged
at every VBlank edge (deterministic, works online, no mod needed), plus
psx_mod_set_guest_cycle_scale_gate() in mod_plugins.h for custom logic or
to tie the scale to a mod (guest_cycle_scale_gated = true; every mod
session reset shuts it). The carried fraction and both gate states are a
boot_state section (BS_SEC_GCS, written only when the scale is not 1), so
savestates and the netplay rollback snapshot re-simulate the same cycles.
psx_guest_cycle_scale_gate_test pins tier identity, the gates and rollback
re-simulation against the real psx_cycles.c.
OpenBIOS regenerated (emitter stamp); retail SCPH1001 is not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
No CPU is emulated: recompiled code charges the guest clock a fixed cost
per instruction. guest_cycle_scale = N charges 1/N of it (instruction,
load, mul/div, GTE latencies, overlay DLL batches) while VBlank, timers,
CD, SPU and DMA keep hardware time, so the game never runs out of frame
time. A title constant in game.toml (1 = faithful, 1..64), never a player
setting; guest_cycle_scale_gated applies it only while the title's mod
opens the gate (psx_guest_cycle_scale_gate_open). PSX_GUEST_CYCLE_SCALE
overrides for testing. A scale other than 1 is folded into the netplay
content fingerprint as a safety net.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[video] pgxp_seam = "off" | "fine" | "wide" (default off; PSX_PGXP_SEAM,
PSX_PGXP_SEAM_PX). Above 1x, a vertex that sits on a neighbour's edge only to
the PS1's precision leaves a hairline crack. Opaque 3D triangles move each
edge outward (mitred, limited at sharp corners): fine = 1 output px, wide =
half a native px; UVs, colour, q and SZ extrapolate with the barycentric
coordinates of the new corners so textures do not slide.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[video] pgxp_color_correction (default off; PSX_PGXP_COLOR overrides). Gouraud
colour on PGXP 3D triangles interpolates with 1/SZ; textured triangles with
perspective UVs share their w. The flag rides in the per-vertex PGXP code the
depth buffer already carries (sz + 65536).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Triangles with a vertex nearer than SZ 1024 (PSX_PGXP_DEPTH_NEAR) stay out of
the depth buffer, the tolerance gains an absolute 48 SZ, and every displayed
frame starts with a cleared depth buffer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A 3D batch is two passes: colour tests against depth pulled toward the
camera by 2% of the distance (PSX_PGXP_DEPTH_TOL) and writes none; a
colourless pass then writes the true depth. Lane markings drawn after the
road a hair off its plane stay whole.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[video] pgxp_depth_buffer (default off; PSX_PGXP_DEPTH overrides) and
pgxp_depth_threshold. gpu.c hands the GL backend each triangle's GTE SZ when
all three vertices are dataflow-precise (gr_set_depth_triangle); the backend
draws those opaque triangles with a LEQUAL depth test and write. 2D,
unproven, semi-transparent polygons, lines and rectangles never test or
write. Depth mode is a batch key and rides in every high-resolution-window,
native-wide and render-thread replay command. Clears are colourless
depth-only draws through the flat batch: after a drawing-area change or a
fill, and when the average SZ rises by the threshold (4096, as DuckStation).
Generated (Smooth motion) frames draw without depth.
Docs: ENHANCEMENTS.md G1.14.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Under HD texture-pack authority (and dual-raster netplay) every draw was
also rasterized into CPU VRAM unless a render pass was active. Replayed
draws of a generated frame were not exempt, so each in-between frame wrote
interpolated geometry into the guest's native VRAM. They are presentation
only, like a pass. Fixture: compose under HD authority, VRAM unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Log lines and doc headings say "Smooth motion (frame generation)", the
player-facing name. Config key [video] frame_generation and PSX_FRAME_GEN
are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
render_thread, present_thread and frame_generation are read from the
player's settings.toml [video] (over the game.toml default; PSX_* env
still overrides one run and is never saved), written back by
save_user_settings, and round-tripped through the recomp-ui Display rows
when the launcher defines RECOMP_LAUNCHER_HAS_RENDER_PIPELINE. The
pipeline starts once at boot: a first-boot launcher choice applies
immediately, an in-game change is saved and applies at next launch.
Regenerated the OpenBIOS emitter stamp for the config_loader change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gpu.c's precise-triangle path references Smooth motion's vertex sources;
the software-only fixture never reaches them (abort if it does).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gpu.c asks the GL backend and gte.cpp for Smooth motion's vertex sources;
master's native_wide_projection_test (cd9981cd) links gpu.c with these stubs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The remaining tunnel cracks came from vertex sources, not from missing
depth: 46 % of R4 lookups were marked ambiguous because the same 3D point
is projected several times per frame (corners shared by separate meshes,
other functions), so those vertices fell back to a triangle-mean guess.
- The same point projected again is one vertex (the smaller identity, so
the choice does not follow draw order).
- Up to two different points on one pixel are kept; gpu.c picks the one
projected by the same function as the triangle's other vertices, else
the one nearest their depth.
- Per-vertex motion is no longer rounded to 1/8 px: at high internal
scales the rounding opened hairlines along long T-junction seams.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Corner and tunnel runs showed light cracks along seams and near the view
edge in generated frames:
- weld: every placed vertex at one screen position of a view moves alike
(the same corner arrives from several projections);
- a vertex without a projection on an edge between placed vertices moves
as that point of the edge (T-junctions from the game's polygon splits);
- a vertex the in-between camera passes is projected at a clamped depth,
and such a game frame is not generated (no near-plane clipping: tunnel
ceilings, overhead bridges);
- debug: "clamped", "guessed" (CPU-built vertices placed by a triangle
mean over a >1 px motion spread).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Pairing triangles of consecutive draw lists by key and draw order paired
the wrong ones whenever the game reordered, culled or switched LODs (R4:
a grandstand of one repeated texture turned into a staircase; on a
synthetic version 225 of 226 pairs were wrong). No primitive is paired by
draw order any more.
gte.cpp notes each RTPS/RTPT output (only while generation is on) with an
identity (issuing ra + model-space vertex), its camera-space position and
H, indexed by screen word; gpu.c records each polygon vertex's source
ahead of the triangle (RTH_FG_SRC, never drawn). Per view the camera
motion between the two frames is fitted from the paired camera-space
vertices (RANSAC + least squares); one real frame is redrawn from the
in-between camera: world vertices re-projected, vertices moving against
the world (cars) lerped in camera space, CPU-built vertices moved with
their triangle, HUD/2D unchanged. Moving forward the older frame is
redrawn; uncovered areas show the newer real frame. A verdict rejects
the generated frame (the real one shows) without projections, with few
pairs/inliers, or with a motion no camera makes. Real frames are
unchanged (gl_frame_gen_test).
PSX_PRESENT_SHOT_BURST=N captures the next N composed frames, real and
generated, for checking in-between frames against their neighbours.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A render thread a frame behind lowers the plan's ceiling by one (recovering
one per 2 s) instead of tripping the breaker; the guest waiting on the queue
trips it only when the wait was a stall (a quarter of the game frame). The
real-frame cost plans with CPU time; the GPU span is reported only. The
generated-frame blit copies colour only and marks the stencil stale.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Both swap paths go through gl_window_swap. The temporal-blending ownership
guard allows the present thread's context (pt_begin only). The frame
generation fixture used setenv, which MinGW lacks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Diagnostic: a staged present_shot waits for the next generated frame, so the
in-between images can be inspected in game.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Armed at init (session start, render-thread start, a new resolution) and on
a savestate load, game entry or window resize. While armed, an overrun jumps
straight to the highest level whose pure-area prediction is at or below
0.78, then is judged like any step; a fit after a step, an up step, a failed
judgement or 6 s disarm it and single steps fine-tune. A fixed cost only
makes the target dearer than predicted, so a jump never lands below the
level that fits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Opt-in under the render thread. Composed frames go to a ring of offscreen
slots at the window's size (binds of framebuffer 0 on the composing context
land on the current slot); a present thread with a second, shared context on
the same window waits on the slot's fence, copies it 1:1 and swaps. The
compositor's wait moves off rendering; the composing side waits only while
every slot is queued or on screen. Falls back to direct swaps when the shared
context or sync objects are unavailable.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
master's HD texture packs (#557 base) added gpu_hd_textures_* calls to the GL
renderer and the gpu_render.c facade; the fixture runs without a pack. The
command index is regenerated for frame_gen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[video] frame_generation (opt-in, with render_thread): the render thread keeps
each game frame's replayed draw list, matches the triangles of the last two
(frame_gen.c), and presents in-between frames at the display refresh by
redrawing the newer list into separate surfaces with matched vertices
interpolated. It never runs guest code. Real frames are composed at their
flip and presented after the generated ones, with the same pixels and
order. Generation uses only measured surplus: a plan from render-thread CPU,
generated-frame cost and swap time, a breaker after backpressure or late
guest frames, and a pause while dynamic resolution is over budget. The
render-thread core gains a timed tick hook. Frame-cost queries pause around
generation, so dynamic resolution measures real frames only.
Tests: frame_gen_test, render_thread_test (tick), gl_frame_gen_test (real
frames identical off/on; phase 0/1 equal the endpoints). Docs:
docs/FRAME_GENERATION.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A step grows the scratch to hold the old level's displayed rect and draw
area, and it stayed that size: every tiled stencil rebuild renders into it,
so on GL-on-Metal its size cost bandwidth on each one. Return it to the size
a surface allocated at the new level starts with.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The hr (VRAM) surface stayed allocated at the ceiling. On GL-on-Metal every
render pass loads and stores the whole attachment, so its cost followed the
ceiling rather than the level: R4 2P VS at 5x ran 40-45 Hz for its first
seconds with a 10x ceiling and 60 Hz with surfaces allocated at 5x. Allocate
a new hr surface at each step and reseed it from the 1x image and the saved
rects; a failed allocation falls back to the in-place step down, or refuses
a step up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A native-wide surface was allocated at the dynamic-resolution ceiling and
rendered into its lower-left corner. On GL-on-Metal every render pass loads
and stores the whole attachment, so the wide surfaces cost the ceiling's
bandwidth at every level. Allocate them at the level they render at and
reallocate at each step, rescaling the presented rows and margins from the
old surface. The limit check stays at the ceiling.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The step record's replay sat inside the debug-tools block, so release
builds dropped every recorded step. gl_render_thread_test gains a dynres
mode: the same level steps between frames with the thread off (applied at
once) and on (recorded), identical guest readbacks, VRAM, frame and wide
surface, and the render thread's frame cost measured. Also drop the
duplicate gpu_ws_background_requires_full_composite stub the #508 merge
left in the scale-invariance fixture.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
master's HD texture packs (#557 base) added gpu_hd_textures_* calls to the GL
renderer and the gpu_render.c facade; the fixture runs without a pack.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
render_thread_test is registered on MinGW (winpthreads provides the
fixture's pthreads and nanosleep; the core takes its Win32 path).
gl_render_thread_test gains the MinGW link path run_gl_scale_invariance.py
uses on test/gl-scale-invariance-windows (static SDL3, opengl32 and its
Win32 libraries, LTO, probe.exe), and a host with no window or GL 3.3 core
context (headless, Windows over SSH) exits 77 from a preflight run, reported
by CTest as a skip.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An empty CPU->VRAM rect fell through to glb_vram_transfer_in on the
emulation thread while the render thread owned the context; it is now
recorded like any other upload. A projected triangle whose record cannot
be reserved takes the context first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
--headless-opengl captured the presented image into the ring every frame
through gl_renderer_capture_display_hires, a sync point, so a headless
run (fp_identity) held the context for most of every frame. The capture
is now recorded at the same point of the frame (gl_renderer_ring_capture)
and done by the render thread; ring readers already sync first. A rect the
GL surface cannot serve is skipped there rather than resolved through a
CPU readback, and ensure_cpu refuses to run on the render thread at all:
its readback would land in the private upload source and mark the CPU
side current without gpu.c's array seeing the pixels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The OSD query at the two present-skip sites now goes through
ov_needs_present(), which falls through to host_osd_needs_present()
unless the render thread is replaying a present with its recorded
overlay state.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A default-QoS pthread is scheduled onto efficiency cores when the host is
busy, which put the render thread, now on every frame's critical path,
behind the emulation thread it is supposed to relieve.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
R4 races present new content every other vblank, so at a saturated 4K
widescreen frame two vblanks behind was common and about half the
presents were dropped. Skipping is a catch-up step, not a frame-rate
policy: one skipped present is always followed by a shown one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
One frame behind is the normal pipelined state when the render thread is
the slower side; skipping on it dropped nearly every present at 4K
widescreen. rt_frames_ahead() counts the complete frames queued after
the one being replayed; the backend skips only at two, where the
emulation thread is at the in-flight bound waiting on the render thread.
The unit test checks the count never exceeds the bound and that both
cases occur under a slow consumer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Shokunin