sgilissen/llamaplayer
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

GPU/HW: Simplify 24-bit scanout and interlacing shader

Browse Source 
Has the added bonus of being faster (no downscale copy) and fixes the
edge case where the image is cropped in VRAM.
This commit is contained in:
Connor McLaughlin
2020-04-02 00:17:25 +10:00
parent 1d948a53e0
commit 2a6e04988f
5 changed files with 75 additions and 185 deletions
Download Patch File Download Diff File Expand all files Collapse all files

83
src/core/gpu_hw_opengl_es.cpp
View File

@ -371,75 +371,46 @@ void GPU_HW_OpenGL_ES::UpdateDisplay()
}
else
{
const u32 flipped_vram_offset_y = VRAM_HEIGHT - vram_offset_y - display_height;
const u32 scaled_flipped_vram_offset_y =
m_vram_texture.GetHeight() - scaled_vram_offset_y - scaled_display_height;
const u32 field_offset = BoolToUInt8(interlaced && m_GPUSTAT.interlaced_field);
glDisable(GL_BLEND);
glDisable(GL_SCISSOR_TEST);
const GL::Program& prog =
GL::Program& prog =
m_display_programs[BoolToUInt8(m_GPUSTAT.display_area_color_depth_24)][BoolToUInt8(interlaced)];
prog.Bind();
m_display_texture.BindFramebuffer(GL_DRAW_FRAMEBUFFER);
m_vram_texture.Bind();
// Because of how the reinterpret shader works, we need to use the downscaled version.
if (m_GPUSTAT.display_area_color_depth_24 && m_resolution_scale > 1)
{
const u32 copy_width = std::min<u32>((display_width * 3) / 2, VRAM_WIDTH - vram_offset_x);
const u32 scaled_copy_width = copy_width * m_resolution_scale;
m_vram_encoding_texture.BindFramebuffer(GL_DRAW_FRAMEBUFFER);
m_vram_texture.BindFramebuffer(GL_READ_FRAMEBUFFER);
glBlitFramebuffer(scaled_vram_offset_x, scaled_flipped_vram_offset_y, scaled_vram_offset_x + scaled_copy_width,
scaled_flipped_vram_offset_y + scaled_display_height, vram_offset_x, flipped_vram_offset_y,
vram_offset_x + copy_width, flipped_vram_offset_y + display_height, GL_COLOR_BUFFER_BIT,
GL_NEAREST);
const u32 flipped_vram_offset_y = VRAM_HEIGHT - vram_offset_y - display_height;
const u32 scaled_flipped_vram_offset_y =
m_vram_texture.GetHeight() - scaled_vram_offset_y - scaled_display_height;
m_display_texture.BindFramebuffer(GL_DRAW_FRAMEBUFFER);
m_vram_encoding_texture.Bind();
const u32 reinterpret_field_offset =
(m_crtc_state.regs.Y + BoolToUInt8(interlaced && m_GPUSTAT.interlaced_field)) & 1u;
const u32 reinterpret_start_x = m_crtc_state.regs.X * m_resolution_scale;
const u32 reinterpret_width = scaled_display_width + (m_crtc_state.display_vram_left - m_crtc_state.regs.X);
glViewport(0, field_offset, display_width, display_height);
prog.Uniform2i(0, reinterpret_field_offset, reinterpret_start_x);
m_batch_ubo_dirty = true;
prog.Uniform3i(0, static_cast<s32>(vram_offset_x), static_cast<s32>(flipped_vram_offset_y),
static_cast<s32>(field_offset));
m_batch_ubo_dirty = true;
glViewport(reinterpret_start_x, scaled_flipped_vram_offset_y, reinterpret_width, scaled_display_height);
glDrawArrays(GL_TRIANGLES, 0, 3);
glDrawArrays(GL_TRIANGLES, 0, 3);
m_host_display->SetDisplayTexture(reinterpret_cast<void*>(static_cast<uintptr_t>(m_display_texture.GetGLId())),
m_display_texture.GetWidth(), m_display_texture.GetHeight(), 0,
display_height, display_width, -static_cast<s32>(display_height));
}
else
{
m_display_texture.BindFramebuffer(GL_DRAW_FRAMEBUFFER);
m_vram_texture.Bind();
glViewport(0, field_offset, scaled_display_width, scaled_display_height);
prog.Uniform3i(0, static_cast<s32>(scaled_vram_offset_x), static_cast<s32>(scaled_flipped_vram_offset_y),
static_cast<s32>(field_offset));
m_batch_ubo_dirty = true;
glDrawArrays(GL_TRIANGLES, 0, 3);
m_host_display->SetDisplayTexture(reinterpret_cast<void*>(static_cast<uintptr_t>(m_display_texture.GetGLId())),
m_display_texture.GetWidth(), m_display_texture.GetHeight(), 0,
scaled_display_height, scaled_display_width,
-static_cast<s32>(scaled_display_height));
}
// restore state
m_vram_texture.BindFramebuffer(GL_DRAW_FRAMEBUFFER);
glViewport(0, 0, m_vram_texture.GetWidth(), m_vram_texture.GetHeight());
glEnable(GL_SCISSOR_TEST);
m_host_display->SetDisplayTexture(reinterpret_cast<void*>(static_cast<uintptr_t>(m_display_texture.GetGLId())),
m_display_texture.GetWidth(), m_display_texture.GetHeight(),
scaled_vram_offset_x, m_vram_texture.GetHeight() - scaled_vram_offset_y,
scaled_display_width, -static_cast<s32>(scaled_display_height));
}
m_host_display->SetDisplayParameters(m_crtc_state.display_width, m_crtc_state.display_height,
m_crtc_state.display_origin_left, m_crtc_state.display_origin_top,
m_crtc_state.display_vram_width, m_crtc_state.display_vram_height,
m_crtc_state.display_aspect_ratio);
// restore state
m_vram_texture.BindFramebuffer(GL_DRAW_FRAMEBUFFER);
glViewport(0, 0, m_vram_texture.GetWidth(), m_vram_texture.GetHeight());
glEnable(GL_SCISSOR_TEST);
}
m_host_display->SetDisplayParameters(m_crtc_state.display_width, m_crtc_state.display_height,
m_crtc_state.display_origin_left, m_crtc_state.display_origin_top,
m_crtc_state.display_vram_width, m_crtc_state.display_vram_height,
m_crtc_state.display_aspect_ratio);
}
void GPU_HW_OpenGL_ES::ReadVRAM(u32 x, u32 y, u32 width, u32 height)