core/d3d11+d3d12: consume the pre-baked xBR batch FS blobs

Wires GPU_HW_D3D12::GetBatchPipeline and GPU_HW_D3D11::GetBatch
PixelShader to a new PickBatchTexturedXBRFS picker that serves
the 144 xBR-family DXBC blobs landed at the e07ce04 foundation
commit. Same shape as the Bilinear / JINC2 activation pairs
(269a2a0 + f2ae92a, 17a0c66 + 6afe8c2) - identical picker
signature with the `binalpha` parameter selecting between xBR
and xBRBinAlpha via the BINALPHA -D arm of the shared HLSL
template.

xBR is the last filter template in the batch FS pre-bake arc.
After this commit, all 5 batch FS slices are pre-baked-active
on both D3D11 and D3D12:

  - Untextured (texture_mode == Disabled): 12 variants
  - Textured + Nearest:                    72 variants
  - Textured + Bilinear / BilinearBinAlpha: 144 variants
  - Textured + JINC2 / JINC2BinAlpha:       144 variants
  - Textured + xBR / xBRBinAlpha:           144 variants
  Total:                                    516 variants live

The shadergen-fallback else arm in both backends' batch FS
selection becomes unreachable as a result and is removed in this
commit. Follow-up cleanup (separate commit) will delete the
now-dead GetBatchFragmentShader method on D3D12 and the matching
member-field state (m_batch_fragment_shader_blobs,
m_batch_fragment_shader_blobs_fastpath, m_batch_shader_mutex),
plus the m_shader_cache.GetPixelShader path that's no longer
reached on D3D11.

What changes:

  src/common/d3d_common/embedded_shaders.h
    288 new extern declarations for the 144 xBR blobs (uint8_t
    array + size_t pair per blob). Section header documents that
    xBR is the last filter family + describes per-variant size
    rationale (5x5 sample fan-out + 4-quadrant blend tree).

    1 new picker function declaration:
      Bytecode PickBatchTexturedXBRFS(
        uint8_t lookup_mode, bool binalpha,
        bool use_dual_source, uint32_t multisamples,
        bool per_sample_shading, bool disable_color_perspective);
    Signature identical to PickBatchTexturedBilinearFS /
    PickBatchTexturedJINC2FS.

  src/common/d3d_common/embedded_shaders.cpp
    144 new #include lines pulling in the .inc files inside the
    D3DCommon::EmbeddedShaders namespace.

    1 new picker implementation - same 5-dim 6x2x2x3x2 static
    table layout as Bilinear / JINC2:
      static const Bytecode k_blobs[6][2][2][3][2] =
        // [tm][binalpha][use_dual_source][interp][persp]
    tm_idx / interp_idx / persp_idx / dual_idx / b_idx derivation
    identical to the other filter pickers.

  src/core/gpu_hw_d3d12.cpp
    GetBatchPipeline fragment-shader fork: the existing
    shadergen-fallback else arm is replaced with the xBR picker
    call. The arm covers exactly {xBR, xBRBinAlpha} - the only
    GPUTextureFilter values that aren't matched by the four
    preceding pre-baked arms. binalpha drives the BINALPHA -D
    macro arm of the xBR template; otherwise the picker takes
    the same input set as the other batch FS pickers.

    The dead `ComPtr<ID3DBlob> fs_blob` local (formerly held the
    shadergen path's blob lifetime) is removed.

    precompile_sync's pre_baked predicate is simplified to
    unconditional skip - every reachable batch FS cell is now
    pre-baked. The progress.Increment() call still ticks per
    reachable cell so the user-facing progress bar count is
    unchanged from previous releases; cells complete near-
    instantly because no D3DCompile runs at warmup time. The
    GetBatchFragmentShader call inside the if (!pre_baked) block
    is removed; the function itself remains in place for the
    cleanup follow-up.

  src/core/gpu_hw_d3d11.cpp
    GetBatchPixelShader: the existing else arm (constructing
    tmp_shadergen + calling m_shader_cache.GetPixelShader) is
    replaced with the xBR picker call. Same input set + same
    binalpha derivation as the D3D12 arm; same picker output.
    The tmp_shadergen construction is gone from this commit; the
    GPU_HW_ShaderGen include is no longer needed by this
    function but remains in place for the m_shader_cache uses
    elsewhere in the file (non-batch shaders). The cleanup
    follow-up will revisit whether m_shader_cache + the include
    are still needed in this TU.

What doesn't change in this commit:

  - Visual output: zero diff. fxc emits identical DXBC from the
    same HLSL whether invoked at edit-time via the regen tool or
    at runtime via D3DCompile.

  - The D3D12 GetBatchFragmentShader method (gpu_hw_d3d12.cpp
    lines 1151-1255) and its supporting state
    (m_batch_fragment_shader_blobs matrix,
    m_batch_fragment_shader_blobs_fastpath atomic array,
    m_batch_shader_mutex) stay in place. Now dead code, but
    removing them is a separate cleanup commit to keep this
    activation focused on the picker wiring.

  - The D3D11 m_shader_cache field + the GPU_HW_ShaderGen
    include in gpu_hw_d3d11.cpp also stay - both are still used
    by the non-batch shader paths (display, copy, vram_*).
    Whether m_shader_cache survives at all depends on whether
    those non-batch paths get pre-baked-consumer wiring too;
    that's a separate audit.

  - m_batch_pixel_shaders matrix dimensions on D3D11. The
    [dithering] / [interlacing] dims hold redundant slots post-
    cbuffer-routing; collapsing them is a separate follow-up
    that touches the matrix declaration + the indexing in every
    use site.

  - m_supports_adaptive_downsampling on D3D12 stays false until
    the Phase 2 downsample port lands.

Behavioural impact for users (D3D11 + D3D12 with xBR or
xBRBinAlpha):

  - CompilePipelines (Enabled mode) and the background
    ShaderCompileThreadEntryPoint (Lazy mode) skip D3DCompile on
    6 (texture_mode) x 2 (binalpha) x 5 (transparency_mode) x 4
    (render_mode) x 2 (depth_test) = 480 PSO build cells per xBR
    filter session. Cold-start D3DCompile cost per xBR cell is
    the highest of any filter (50-80 ms; xBR's 5x5 fan-out + 4-
    quadrant decision tree produces ~5x more SSA values than
    JINC2 to schedule and register-allocate). Cumulative savings:
    20-40 seconds of init time depending on m_multisamples and
    m_per_sample_shading.

  - On-disk shader cache shrinks for xBR sessions. Existing xBR-
    cell entries from older versions remain on disk but go
    unconsumed (the picker bypasses m_shader_cache).

  - DrawBatchVertices is unchanged - same PSO lookup, same fast-
    path atomic load, same draw call. Only the source of the
    bound ID3D11PixelShader / ID3D12PipelineState's FS changes
    from m_shader_cache-managed to picker-driven.

  - precompile_sync (D3D12 Enabled mode) completes faster
    overall - the entire batch FS warmup no longer touches
    D3DCompile, just the picker tables. Cold-start GPU_HW init
    on a fresh shader cache drops by minutes for Enabled mode at
    high m_resolution_scale + m_multisamples combinations.

Test:
  - g++ -std=c++17 -c on src/common/d3d_common/embedded_shaders.cpp
    standalone compiles cleanly. nm shows all 9 pickers in
    D3DCommon::EmbeddedShaders (5 batch FS - Untextured, Nearest,
    Bilinear, JINC2, xBR - plus 4 downsample). .o size grows
    from 5.6 MiB (post-JINC2-activation) to 14 MiB; the delta is
    the embedded xBR DXBC payload (~7 MiB) plus per-blob symbol
    overhead.

  - g++ -std=c++17 syntax check on the new D3D12 xBR arm using a
    shim TU (ComPtr / ID3DBlob / D3D12_SHADER_BYTECODE + the
    real embedded_shaders.h) compiles cleanly. All 5 picker arms
    + the untextured branch type-check.

Cumulative pre-bake state across both backends:

  Pre-arc:                                    75 264 (every axis a -D)
  Post-cbuffer-routing arc:                   1 176
  Post-c01f8ae / a7f5717 (Untextured + Nearest, both backends):
                                              168 cell-equivalents
  Post-2fcf4cc (D3D11 downsample live):       22 additional cells
  Post-269a2a0 + f2ae92a (Bilinear-family live):
                                              +288 cell-equivalents
  Post-17a0c66 + 6afe8c2 (JINC2-family live):
                                              +288 cell-equivalents
  Post-this-commit (xBR-family live):         +288 cell-equivalents
                                              (144 cells x 2 backends)
  Total batch FS variants pre-baked + live:   1 032 cell-equivalents
                                              (516 cells x 2 backends)

The batch FS pre-bake arc is now complete on both D3D11 and
D3D12. No remaining textured filter or untextured slice goes
through runtime shadergen + D3DCompile at GPU_HW init or at
draw time.

Next:
  - Cleanup commit: delete GetBatchFragmentShader (D3D12) and
    its supporting member state (m_batch_fragment_shader_blobs,
    m_batch_fragment_shader_blobs_fastpath, m_batch_shader_mutex).
    Audit D3D11 m_shader_cache usage and remove if no
    consumers remain. ~150 lines net deletion.
  - m_batch_pixel_shaders matrix dim collapse on D3D11. The
    [dithering] / [interlacing] dims (post-cbuffer-routing
    redundant) can drop, reducing the matrix from 7 x 4 x 9 x 2
    x 2 to 7 x 4 x 9 = 252 unique cells. The pre-baked picker
    doesn't go through this matrix at all anyway, so the
    collapse is a code-cleanliness win without a runtime impact.
  - Shader cache unification: move D3D11::ShaderCache and
    D3D12::ShaderCache to a shared D3DCommon::ShaderCache,
    drop the d3d12_ prefix on the bytecode cache filename so
    D3D11 and D3D12 share a single d3d_shaders_*.bin file
    (pipeline_library_*.bin stays D3D12-only). Bump FILE_VERSION
    in the merged class to invalidate existing on-disk caches
    one final time.
  - Independent: Phase 2 D3D12 downsample port - PSOs, explicit
    resource barriers, texture / mip-view creation. Uses the
    pre-baked blobs from 251e457 + the d3d_common rename at
    0ed7261.
This commit is contained in:
LibretroAdmin 2026-05-20 03:27:16 +00:00 • committed by U-DESKTOP-SPFP6AQ\twistedtechre
commit f2620c1cc1
4 changed files with 1059 additions and 52 deletions

View file

@ -471,6 +471,151 @@ namespace D3DCommon::EmbeddedShaders {
#include "embedded_dxbc/batch_textured_jinc2_ps_p8r1_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_jinc2_ps_p8r1_d1_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d0_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r0_d1_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d0_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p0r1_d1_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d0_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r0_d1_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d0_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p4r1_d1_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d0_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r0_d1_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d0_sample_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_centroid_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_centroid_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_centroid_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_centroid_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_none_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_none_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_none_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_none_n1_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_sample_n0_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_sample_n0_b1.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_sample_n1_b0.inc"
#include "embedded_dxbc/batch_textured_xbr_ps_p8r1_d1_sample_n1_b1.inc"
#include "embedded_dxbc/adaptive_downsample_blur_ps.inc"
#include "embedded_dxbc/adaptive_downsample_composite_ps_s16.inc"
#include "embedded_dxbc/adaptive_downsample_composite_ps_s2.inc"
@ -1914,6 +2059,525 @@ Bytecode PickBatchTexturedJINC2FS(
return k_blobs[tm_idx][b_idx][dual_idx][interp_idx][persp_idx];
}
Bytecode PickBatchTexturedXBRFS(
uint8_t lookup_mode, bool binalpha, bool use_dual_source,
uint32_t multisamples, bool per_sample_shading,
bool disable_color_perspective)
{
// Identical encoding to PickBatchTexturedBilinearFS /
// PickBatchTexturedJINC2FS - the three filter families only
// differ in their HLSL bodies, not in the axis layout that the
// picker indexes. See the Bilinear picker for the tm_idx /
// interp_idx / persp_idx / dual_idx / b_idx derivation; the same
// logic applies verbatim here.
const uint8_t actual_mode = lookup_mode & 0x3u;
const bool raw = (lookup_mode & 0x4u) != 0;
unsigned tm_idx;
if (actual_mode == 2u) // Direct16Bit family
tm_idx = raw ? 1u : 0u;
else if (actual_mode == 0u) // Palette4Bit family
tm_idx = raw ? 3u : 2u;
else // actual_mode == 1u (Palette8Bit family)
tm_idx = raw ? 5u : 4u;
const unsigned interp_idx =
per_sample_shading ? 2u : ((multisamples > 1u) ? 1u : 0u);
const unsigned persp_idx = disable_color_perspective ? 1u : 0u;
const unsigned dual_idx = use_dual_source ? 1u : 0u;
const unsigned b_idx = binalpha ? 1u : 0u;
static const Bytecode k_blobs[6][2][2][3][2] = {
// tm = p0r0 (Direct16Bit)
{
// binalpha = xBR
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r0_d0_none_n0_b0,
k_batch_textured_xbr_ps_p0r0_d0_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d0_none_n1_b0,
k_batch_textured_xbr_ps_p0r0_d0_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b0,
k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b0,
k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b0,
k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b0,
k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b0_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r0_d1_none_n0_b0,
k_batch_textured_xbr_ps_p0r0_d1_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d1_none_n1_b0,
k_batch_textured_xbr_ps_p0r0_d1_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b0,
k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b0,
k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b0,
k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b0,
k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b0_size_bytes}},
},
},
// binalpha = xBRBinAlpha
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r0_d0_none_n0_b1,
k_batch_textured_xbr_ps_p0r0_d0_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d0_none_n1_b1,
k_batch_textured_xbr_ps_p0r0_d0_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b1,
k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b1,
k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b1,
k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b1,
k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b1_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r0_d1_none_n0_b1,
k_batch_textured_xbr_ps_p0r0_d1_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d1_none_n1_b1,
k_batch_textured_xbr_ps_p0r0_d1_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b1,
k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b1,
k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b1,
k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b1,
k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b1_size_bytes}},
},
},
},
// tm = p0r1 (RawDirect16Bit)
{
// binalpha = xBR
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r1_d0_none_n0_b0,
k_batch_textured_xbr_ps_p0r1_d0_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d0_none_n1_b0,
k_batch_textured_xbr_ps_p0r1_d0_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b0,
k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b0,
k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b0,
k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b0,
k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b0_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r1_d1_none_n0_b0,
k_batch_textured_xbr_ps_p0r1_d1_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d1_none_n1_b0,
k_batch_textured_xbr_ps_p0r1_d1_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b0,
k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b0,
k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b0,
k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b0,
k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b0_size_bytes}},
},
},
// binalpha = xBRBinAlpha
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r1_d0_none_n0_b1,
k_batch_textured_xbr_ps_p0r1_d0_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d0_none_n1_b1,
k_batch_textured_xbr_ps_p0r1_d0_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b1,
k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b1,
k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b1,
k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b1,
k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b1_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p0r1_d1_none_n0_b1,
k_batch_textured_xbr_ps_p0r1_d1_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d1_none_n1_b1,
k_batch_textured_xbr_ps_p0r1_d1_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b1,
k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b1,
k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b1,
k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b1,
k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b1_size_bytes}},
},
},
},
// tm = p4r0 (Palette4Bit)
{
// binalpha = xBR
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r0_d0_none_n0_b0,
k_batch_textured_xbr_ps_p4r0_d0_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d0_none_n1_b0,
k_batch_textured_xbr_ps_p4r0_d0_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b0,
k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b0,
k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b0,
k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b0,
k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b0_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r0_d1_none_n0_b0,
k_batch_textured_xbr_ps_p4r0_d1_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d1_none_n1_b0,
k_batch_textured_xbr_ps_p4r0_d1_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b0,
k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b0,
k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b0,
k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b0,
k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b0_size_bytes}},
},
},
// binalpha = xBRBinAlpha
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r0_d0_none_n0_b1,
k_batch_textured_xbr_ps_p4r0_d0_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d0_none_n1_b1,
k_batch_textured_xbr_ps_p4r0_d0_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b1,
k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b1,
k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b1,
k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b1,
k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b1_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r0_d1_none_n0_b1,
k_batch_textured_xbr_ps_p4r0_d1_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d1_none_n1_b1,
k_batch_textured_xbr_ps_p4r0_d1_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b1,
k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b1,
k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b1,
k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b1,
k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b1_size_bytes}},
},
},
},
// tm = p4r1 (RawPalette4Bit)
{
// binalpha = xBR
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r1_d0_none_n0_b0,
k_batch_textured_xbr_ps_p4r1_d0_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d0_none_n1_b0,
k_batch_textured_xbr_ps_p4r1_d0_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b0,
k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b0,
k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b0,
k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b0,
k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b0_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r1_d1_none_n0_b0,
k_batch_textured_xbr_ps_p4r1_d1_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d1_none_n1_b0,
k_batch_textured_xbr_ps_p4r1_d1_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b0,
k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b0,
k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b0,
k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b0,
k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b0_size_bytes}},
},
},
// binalpha = xBRBinAlpha
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r1_d0_none_n0_b1,
k_batch_textured_xbr_ps_p4r1_d0_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d0_none_n1_b1,
k_batch_textured_xbr_ps_p4r1_d0_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b1,
k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b1,
k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b1,
k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b1,
k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b1_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p4r1_d1_none_n0_b1,
k_batch_textured_xbr_ps_p4r1_d1_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d1_none_n1_b1,
k_batch_textured_xbr_ps_p4r1_d1_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b1,
k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b1,
k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b1,
k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b1,
k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b1_size_bytes}},
},
},
},
// tm = p8r0 (Palette8Bit)
{
// binalpha = xBR
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r0_d0_none_n0_b0,
k_batch_textured_xbr_ps_p8r0_d0_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d0_none_n1_b0,
k_batch_textured_xbr_ps_p8r0_d0_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b0,
k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b0,
k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b0,
k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b0,
k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b0_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r0_d1_none_n0_b0,
k_batch_textured_xbr_ps_p8r0_d1_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d1_none_n1_b0,
k_batch_textured_xbr_ps_p8r0_d1_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b0,
k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b0,
k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b0,
k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b0,
k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b0_size_bytes}},
},
},
// binalpha = xBRBinAlpha
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r0_d0_none_n0_b1,
k_batch_textured_xbr_ps_p8r0_d0_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d0_none_n1_b1,
k_batch_textured_xbr_ps_p8r0_d0_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b1,
k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b1,
k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b1,
k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b1,
k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b1_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r0_d1_none_n0_b1,
k_batch_textured_xbr_ps_p8r0_d1_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d1_none_n1_b1,
k_batch_textured_xbr_ps_p8r0_d1_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b1,
k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b1,
k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b1,
k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b1,
k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b1_size_bytes}},
},
},
},
// tm = p8r1 (RawPalette8Bit)
{
// binalpha = xBR
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r1_d0_none_n0_b0,
k_batch_textured_xbr_ps_p8r1_d0_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d0_none_n1_b0,
k_batch_textured_xbr_ps_p8r1_d0_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b0,
k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b0,
k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b0,
k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b0,
k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b0_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r1_d1_none_n0_b0,
k_batch_textured_xbr_ps_p8r1_d1_none_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d1_none_n1_b0,
k_batch_textured_xbr_ps_p8r1_d1_none_n1_b0_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b0,
k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b0,
k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b0_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b0,
k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b0_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b0,
k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b0_size_bytes}},
},
},
// binalpha = xBRBinAlpha
{
// no dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r1_d0_none_n0_b1,
k_batch_textured_xbr_ps_p8r1_d0_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d0_none_n1_b1,
k_batch_textured_xbr_ps_p8r1_d0_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b1,
k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b1,
k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b1,
k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b1,
k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b1_size_bytes}},
},
// dual
{
// interp = none
{{k_batch_textured_xbr_ps_p8r1_d1_none_n0_b1,
k_batch_textured_xbr_ps_p8r1_d1_none_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d1_none_n1_b1,
k_batch_textured_xbr_ps_p8r1_d1_none_n1_b1_size_bytes}},
// interp = centroid
{{k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b1,
k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b1,
k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b1_size_bytes}},
// interp = sample
{{k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b1,
k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b1_size_bytes},
{k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b1,
k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b1_size_bytes}},
},
},
},
};
return k_blobs[tm_idx][b_idx][dual_idx][interp_idx][persp_idx];
}
// ---- Downsample pre-bake pickers -----------------------------------
//
// Tiny tables - 1, 2, 4, and 15 slots respectively. The

View file

@ -147,6 +147,27 @@ Bytecode PickBatchTexturedJINC2FS(uint8_t lookup_mode, bool binalpha,
bool per_sample_shading,
bool disable_color_perspective);
// Textured + xBR-family batch FS variant picker. Selects from the
// 144 blobs at embedded_dxbc/batch_textured_xbr_ps_*.inc. Identical
// signature to the Bilinear / JINC2 pickers (`binalpha` selects
// between xBR and xBRBinAlpha enum values via the BINALPHA -D
// macro arm). The runtime cost difference vs Bilinear / JINC2 is
// in the FS body, not in the picker: xBR samples a 5x5 neighbourhood,
// runs a 4-quadrant blend decision tree on YCbCr distances, and
// applies per-quadrant line-blend special cases.
//
// 6 (tex_mode) x 2 (binalpha) x 2 (dual) x 3 (interp) x 2 (persp)
// = 144 entries.
//
// xBR is the last filter template to be pre-baked. After this
// activation lands, the shadergen-fallback else arm in
// GetBatchPipeline / GetBatchPixelShader becomes unreachable and
// is deleted (along with the tmp_shadergen helper on D3D11).
Bytecode PickBatchTexturedXBRFS(uint8_t lookup_mode, bool binalpha,
bool use_dual_source, uint32_t multisamples,
bool per_sample_shading,
bool disable_color_perspective);
// --------------------------------------------------------------------
// ---- Downsample pre-bake pickers -----------------------------------
@ -1335,6 +1356,322 @@ extern const size_t k_batch_textured_jinc2_ps_p8r1_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_jinc2_ps_p8r1_d1_sample_n1_b1[];
extern const size_t k_batch_textured_jinc2_ps_p8r1_d1_sample_n1_b1_size_bytes;
// Textured + xBR-family batch FS pre-baked DXBC blobs.
// 144 .inc files at src/common/d3d_common/embedded_dxbc/
// batch_textured_xbr_ps_*.inc.
//
// Variant axes (5, all -D macros to fxc) - identical shape to
// the Bilinear / JINC2 families:
// * 6 texture mode combos (3 -D: PALETTE_4_BIT / PALETTE_8_BIT /
// RAW_TEXTURE; PALETTE_4_BIT and PALETTE_8_BIT mutually exclusive)
// * USE_DUAL_SOURCE (0/1)
// * INTERP_CENTROID / INTERP_SAMPLE (none/centroid/sample tri-state)
// * NOPERSP (0/1)
// * BINALPHA (0/1) - gates the ialpha >= 0.5 quantisation in
// FilteredSampleFromVRAM. BINALPHA=0 => xBR; BINALPHA=1 =>
// xBRBinAlpha.
//
// 6 x 2 x 3 x 2 x 2 = 144 blobs. Variant suffix:
// pXrY_d{0,1}_{none,centroid,sample}_n{0,1}_b{0,1}
//
// Per-variant DXBC size is the largest of any pre-baked batch FS
// template (48-60 KiB), because xBR samples a 5x5 neighbourhood,
// runs a 4-quadrant blend decision tree on YCbCr distances, and
// applies per-quadrant line-blend special cases. ~7 MiB embedded
// total. fxc /O3 cannot fold the data-dependent control flow even
// with constant BINALPHA.
//
// Foundation commit: e07ce04. Consumed by PickBatchTexturedXBRFS.
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d0_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r0_d1_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d0_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p0r1_d1_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d0_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r0_d1_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d0_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p4r1_d1_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d0_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r0_d1_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d0_sample_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_centroid_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_none_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_none_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_none_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_none_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_none_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_none_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_none_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_none_n1_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_sample_n0_b1_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b0[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b0_size_bytes;
extern const uint8_t k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b1[];
extern const size_t k_batch_textured_xbr_ps_p8r1_d1_sample_n1_b1_size_bytes;
// Downsample fragment shaders. Backport of the 4 Vulkan downsample

View file

@ -1447,27 +1447,28 @@ ID3D11PixelShader* GPU_HW_D3D11::GetBatchPixelShader(GPUTextureFilter filter, ui
}
else
{
// xBR / xBRBinAlpha still go through shadergen + D3DCompile.
// The else arm shrinks to just the xBR family after this
// commit; the final foundation + activation pair removes it
// entirely.
// Fifth and final pre-baked batch FS slice (e07ce04 foundation
// + this commit's activation). The remaining filter values are
// xBR / xBRBinAlpha by elimination. Picker structure identical
// to Bilinear / JINC2; the body of the picked DXBC is xBR's
// 5x5 neighbourhood + 4-quadrant blend decision tree + per-
// quadrant line-blend special cases. binalpha drives the
// BINALPHA -D arm: xBRBinAlpha quantises the blend-weighted
// alpha to {0, 1} before the `ialpha < 0.5 ? discard : ...`
// test.
//
// Construct a per-call shadergen bound to the requested filter
// rather than reusing m_shadergen (which is pinned to the
// runtime-current m_texture_filtering for the non-batch helpers).
// Filter is the only setting that differs between this shadergen
// and m_shadergen - all other inputs come from member state that
// is single-valued per session. Construction is a handful of POD
// copies; the expensive work is GenerateBatchFragmentShader +
// D3DCompile + CreatePixelShader that follow, all unchanged from
// the pre-dim-cache path.
GPU_HW_ShaderGen tmp_shadergen(
m_host_display->GetRenderAPI(), m_resolution_scale, m_multisamples, m_per_sample_shading, m_true_color,
m_scaled_dithering, filter, m_using_uv_limits, m_pgxp_depth_buffer, m_disable_color_perspective,
m_supports_dual_source_blend);
const std::string ps = tmp_shadergen.GenerateBatchFragmentShader(
static_cast<BatchRenderMode>(render_mode), static_cast<GPUTextureMode>(lookup_mode), dithering, interlacing);
fresh = m_shader_cache.GetPixelShader(m_device.Get(), ps);
// GPUTextureFilter::Count is unreachable here (the filter enum
// is set from settings and validated at parse time), so this
// arm covers exactly {xBR, xBRBinAlpha}. The shadergen +
// tmp_shadergen + m_shader_cache.GetPixelShader fallback path
// that used to live here is deleted - all batch FS variants
// now consume pre-baked DXBC via D3DCommon::EmbeddedShaders
// pickers.
const bool binalpha = (filter == GPUTextureFilter::xBRBinAlpha);
const auto bc = D3DCommon::EmbeddedShaders::PickBatchTexturedXBRFS(
lookup_mode, binalpha, use_dual_source, m_multisamples,
m_per_sample_shading, m_disable_color_perspective);
fresh = D3D11::ShaderCompiler::CreatePixelShader(m_device.Get(), bc.data, bc.size);
}
if (!fresh)

View file

@ -892,34 +892,24 @@ bool GPU_HW_D3D12::CompilePipelines()
if (!IsBatchShaderReachable(static_cast<BatchRenderMode>(render_mode), texture_mode, dual_source))
continue;
// Pre-baked variants skip the GetBatchFragmentShader cache-
// warm step - the bytecode is statically linked, no
// D3DCompile to run. Pre-baked templates so far:
// All batch FS variants are pre-baked as of the e07ce04 +
// <this-commit> arc completion. No shadergen + D3DCompile
// call needed during precompile_sync's warmup; the
// GetBatchPipeline path consumes pre-baked DXBC via
// D3DCommon::EmbeddedShaders pickers in all cases:
// - Untextured (texture_mode == Disabled): commits
// 9e6f933 / b6d1903 / c01f8ae (re-bake).
// - Textured + Nearest: commits 9e4c33d + a7f5717.
// - Textured + Bilinear / BilinearBinAlpha: commits
// 269a2a0 + f2ae92a.
// - Textured + JINC2 / JINC2BinAlpha: commits
// 17a0c66 + this one.
// The one remaining textured filter template (xBR /
// xBRBinAlpha) is still runtime-compiled and lands in the
// final foundation + activation pair. progress.Increment
// ticks regardless so the user-facing progress bar reads the
// same count.
const bool pre_baked =
(static_cast<GPUTextureMode>(texture_mode) == GPUTextureMode::Disabled) ||
(cur_filter == GPUTextureFilter::Nearest) ||
(cur_filter == GPUTextureFilter::Bilinear) ||
(cur_filter == GPUTextureFilter::BilinearBinAlpha) ||
(cur_filter == GPUTextureFilter::JINC2) ||
(cur_filter == GPUTextureFilter::JINC2BinAlpha);
if (!pre_baked)
{
ComPtr<ID3DBlob> blob = GetBatchFragmentShader(cur_filter, render_mode, texture_mode);
if (!blob)
return false;
}
// 17a0c66 + 6afe8c2.
// - Textured + xBR / xBRBinAlpha: commits e07ce04 +
// this one.
// progress.Increment ticks for each reachable cell so the
// user-facing progress bar reads the same count it did pre-
// pre-bake; the cells just complete quickly because no
// D3DCompile runs.
progress.Increment();
}
}
@ -1329,7 +1319,6 @@ GPU_HW_D3D12::ComPtr<ID3D12PipelineState> GPU_HW_D3D12::GetBatchPipeline(GPUText
filter != GPUTextureFilter::Nearest);
D3D12_SHADER_BYTECODE fs_bc;
ComPtr<ID3DBlob> fs_blob; // null on the pre-baked path; held for ref-life otherwise
if (textured)
{
if (filter == GPUTextureFilter::Nearest)
@ -1386,15 +1375,31 @@ GPU_HW_D3D12::ComPtr<ID3D12PipelineState> GPU_HW_D3D12::GetBatchPipeline(GPUText
}
else
{
// xBR / xBRBinAlpha still go through shadergen + D3DCompile.
// The else arm shrinks to just the xBR family until its
// foundation + activation pair lands; after that the else
// becomes unreachable and is deleted.
fs_blob = GetBatchFragmentShader(filter, render_mode, lookup_mode);
if (!fs_blob)
return {};
fs_bc.pShaderBytecode = fs_blob->GetBufferPointer();
fs_bc.BytecodeLength = fs_blob->GetBufferSize();
// Fifth and final pre-baked batch FS slice (e07ce04 foundation
// + this commit's activation). The remaining filter values are
// xBR / xBRBinAlpha by elimination. Same picker shape as
// Bilinear / JINC2; the body of the picked DXBC is xBR's
// 5x5 neighbourhood + 4-quadrant blend decision tree + per-
// quadrant line-blend special cases. binalpha drives the
// BINALPHA -D arm: xBRBinAlpha quantises the blend-weighted
// alpha to {0, 1} before the `ialpha < 0.5 ? discard : ...`
// test.
//
// GPUTextureFilter::Count is unreachable here (the filter enum
// is set from settings and validated at parse time), so this
// arm covers exactly {xBR, xBRBinAlpha}. The shadergen +
// D3DCompile fallback that used to live here is deleted - the
// tmp_shadergen path on D3D11 and the GetBatchFragmentShader
// call on D3D12 are no longer reached. GetBatchFragmentShader
// itself remains in place for any remaining non-batch callers
// (display-shader emission, etc.) but is no longer invoked
// from the batch FS path.
const bool binalpha = (filter == GPUTextureFilter::xBRBinAlpha);
const auto bc = D3DCommon::EmbeddedShaders::PickBatchTexturedXBRFS(
lookup_mode, binalpha, use_dual_source, m_multisamples,
m_per_sample_shading, m_disable_color_perspective);
fs_bc.pShaderBytecode = bc.data;
fs_bc.BytecodeLength = bc.size;
}
}
else