Merge pull request #2793 from ReinUsesLisp/bgr565

renderer_opengl: Implement RGB565 framebuffer format
This commit is contained in:
bunnei 2019-09-03 22:36:32 -04:00 committed by GitHub
commit 19af91434e
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
15 changed files with 72 additions and 205 deletions

View file

@ -37,7 +37,7 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
transform, crop_rect}; transform, crop_rect};
system.GetPerfStats().EndGameFrame(); system.GetPerfStats().EndGameFrame();
system.GPU().SwapBuffers(framebuffer); system.GPU().SwapBuffers(&framebuffer);
} }
} // namespace Service::Nvidia::Devices } // namespace Service::Nvidia::Devices

View file

@ -17,18 +17,6 @@
namespace Tegra { namespace Tegra {
u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
switch (format) {
case PixelFormat::ABGR8:
case PixelFormat::BGRA8:
return 4;
default:
return 4;
}
UNREACHABLE();
}
GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer, bool is_async) GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer, bool is_async)
: system{system}, renderer{renderer}, is_async{is_async} { : system{system}, renderer{renderer}, is_async{is_async} {
auto& rasterizer{renderer.Rasterizer()}; auto& rasterizer{renderer.Rasterizer()};

View file

@ -95,14 +95,10 @@ class DebugContext;
struct FramebufferConfig { struct FramebufferConfig {
enum class PixelFormat : u32 { enum class PixelFormat : u32 {
ABGR8 = 1, ABGR8 = 1,
RGB565 = 4,
BGRA8 = 5, BGRA8 = 5,
}; };
/**
* Returns the number of bytes per pixel.
*/
static u32 BytesPerPixel(PixelFormat format);
VAddr address; VAddr address;
u32 offset; u32 offset;
u32 width; u32 width;
@ -253,8 +249,7 @@ public:
virtual void PushGPUEntries(Tegra::CommandList&& entries) = 0; virtual void PushGPUEntries(Tegra::CommandList&& entries) = 0;
/// Swap buffers (render frame) /// Swap buffers (render frame)
virtual void SwapBuffers( virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) = 0;
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
virtual void FlushRegion(CacheAddr addr, u64 size) = 0; virtual void FlushRegion(CacheAddr addr, u64 size) = 0;

View file

@ -23,9 +23,8 @@ void GPUAsynch::PushGPUEntries(Tegra::CommandList&& entries) {
gpu_thread.SubmitList(std::move(entries)); gpu_thread.SubmitList(std::move(entries));
} }
void GPUAsynch::SwapBuffers( void GPUAsynch::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) { gpu_thread.SwapBuffers(framebuffer);
gpu_thread.SwapBuffers(std::move(framebuffer));
} }
void GPUAsynch::FlushRegion(CacheAddr addr, u64 size) { void GPUAsynch::FlushRegion(CacheAddr addr, u64 size) {

View file

@ -14,15 +14,14 @@ class RendererBase;
namespace VideoCommon { namespace VideoCommon {
/// Implementation of GPU interface that runs the GPU asynchronously /// Implementation of GPU interface that runs the GPU asynchronously
class GPUAsynch : public Tegra::GPU { class GPUAsynch final : public Tegra::GPU {
public: public:
explicit GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer); explicit GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer);
~GPUAsynch() override; ~GPUAsynch() override;
void Start() override; void Start() override;
void PushGPUEntries(Tegra::CommandList&& entries) override; void PushGPUEntries(Tegra::CommandList&& entries) override;
void SwapBuffers( void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;
void FlushRegion(CacheAddr addr, u64 size) override; void FlushRegion(CacheAddr addr, u64 size) override;
void InvalidateRegion(CacheAddr addr, u64 size) override; void InvalidateRegion(CacheAddr addr, u64 size) override;
void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override; void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;

View file

@ -19,9 +19,8 @@ void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) {
dma_pusher->DispatchCalls(); dma_pusher->DispatchCalls();
} }
void GPUSynch::SwapBuffers( void GPUSynch::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) { renderer.SwapBuffers(framebuffer);
renderer.SwapBuffers(std::move(framebuffer));
} }
void GPUSynch::FlushRegion(CacheAddr addr, u64 size) { void GPUSynch::FlushRegion(CacheAddr addr, u64 size) {

View file

@ -13,15 +13,14 @@ class RendererBase;
namespace VideoCommon { namespace VideoCommon {
/// Implementation of GPU interface that runs the GPU synchronously /// Implementation of GPU interface that runs the GPU synchronously
class GPUSynch : public Tegra::GPU { class GPUSynch final : public Tegra::GPU {
public: public:
explicit GPUSynch(Core::System& system, VideoCore::RendererBase& renderer); explicit GPUSynch(Core::System& system, VideoCore::RendererBase& renderer);
~GPUSynch() override; ~GPUSynch() override;
void Start() override; void Start() override;
void PushGPUEntries(Tegra::CommandList&& entries) override; void PushGPUEntries(Tegra::CommandList&& entries) override;
void SwapBuffers( void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;
void FlushRegion(CacheAddr addr, u64 size) override; void FlushRegion(CacheAddr addr, u64 size) override;
void InvalidateRegion(CacheAddr addr, u64 size) override; void InvalidateRegion(CacheAddr addr, u64 size) override;
void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override; void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;

View file

@ -39,7 +39,7 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
dma_pusher.Push(std::move(submit_list->entries)); dma_pusher.Push(std::move(submit_list->entries));
dma_pusher.DispatchCalls(); dma_pusher.DispatchCalls();
} else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) { } else if (const auto data = std::get_if<SwapBuffersCommand>(&next.data)) {
renderer.SwapBuffers(std::move(data->framebuffer)); renderer.SwapBuffers(data->framebuffer ? &*data->framebuffer : nullptr);
} else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) { } else if (const auto data = std::get_if<FlushRegionCommand>(&next.data)) {
renderer.Rasterizer().FlushRegion(data->addr, data->size); renderer.Rasterizer().FlushRegion(data->addr, data->size);
} else if (const auto data = std::get_if<InvalidateRegionCommand>(&next.data)) { } else if (const auto data = std::get_if<InvalidateRegionCommand>(&next.data)) {
@ -78,9 +78,9 @@ void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
system.CoreTiming().ScheduleEvent(synchronization_ticks, synchronization_event, fence); system.CoreTiming().ScheduleEvent(synchronization_ticks, synchronization_event, fence);
} }
void ThreadManager::SwapBuffers( void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) { PushCommand(SwapBuffersCommand(framebuffer ? *framebuffer
PushCommand(SwapBuffersCommand(std::move(framebuffer))); : std::optional<const Tegra::FramebufferConfig>{}));
} }
void ThreadManager::FlushRegion(CacheAddr addr, u64 size) { void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {

View file

@ -110,8 +110,7 @@ public:
void SubmitList(Tegra::CommandList&& entries); void SubmitList(Tegra::CommandList&& entries);
/// Swap buffers (render frame) /// Swap buffers (render frame)
void SwapBuffers( void SwapBuffers(const Tegra::FramebufferConfig* framebuffer);
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer);
/// Notify rasterizer that any caches of the specified region should be flushed to Switch memory /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
void FlushRegion(CacheAddr addr, u64 size); void FlushRegion(CacheAddr addr, u64 size);

View file

@ -25,8 +25,8 @@ static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth
// With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
// pixel values. // pixel values.
const u32 tile_size_x{GetDefaultBlockWidth(format)}; constexpr u32 tile_size_x{GetDefaultBlockWidth(format)};
const u32 tile_size_y{GetDefaultBlockHeight(format)}; constexpr u32 tile_size_y{GetDefaultBlockHeight(format)};
if constexpr (morton_to_linear) { if constexpr (morton_to_linear) {
Tegra::Texture::UnswizzleTexture(buffer, addr, tile_size_x, tile_size_y, bytes_per_pixel, Tegra::Texture::UnswizzleTexture(buffer, addr, tile_size_x, tile_size_y, bytes_per_pixel,
@ -186,99 +186,6 @@ static MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFor
return morton_to_linear_fns[static_cast<std::size_t>(format)]; return morton_to_linear_fns[static_cast<std::size_t>(format)];
} }
static u32 MortonInterleave128(u32 x, u32 y) {
// 128x128 Z-Order coordinate from 2D coordinates
static constexpr u32 xlut[] = {
0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042,
0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809,
0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000,
0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043,
0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a,
0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001,
0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048,
0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b,
0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002,
0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049,
0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840,
0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003,
0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a,
0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841,
0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008,
0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b,
0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842,
0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009,
0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800,
0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843,
0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a,
0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801,
0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848,
0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b,
0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802,
0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849,
0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040,
0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803,
0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a,
0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041,
0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808,
0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b,
0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042,
0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809,
0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b,
};
static constexpr u32 ylut[] = {
0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090,
0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124,
0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200,
0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294,
0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330,
0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404,
0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0,
0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534,
0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610,
0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4,
0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780,
0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014,
0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0,
0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184,
0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220,
0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4,
0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390,
0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424,
0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500,
0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594,
0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630,
0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704,
0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0,
0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034,
0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110,
0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4,
0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280,
0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314,
0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0,
0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484,
0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520,
0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4,
0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690,
0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724,
0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4,
};
return xlut[x % 128] + ylut[y % 128];
}
static u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
// Calculates the offset of the position of the pixel in Morton order
// Framebuffer images are split into 128x128 tiles.
constexpr u32 block_height = 128;
const u32 coarse_x = x & ~127;
const u32 i = MortonInterleave128(x, y);
const u32 offset = coarse_x * block_height;
return (i + offset) * bytes_per_pixel;
}
void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stride, void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stride,
u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing, u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
u8* buffer, u8* addr) { u8* buffer, u8* addr) {
@ -286,23 +193,4 @@ void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stri
tile_width_spacing, buffer, addr); tile_width_spacing, buffer, addr);
} }
void MortonCopyPixels128(MortonSwizzleMode mode, u32 width, u32 height, u32 bytes_per_pixel,
u32 linear_bytes_per_pixel, u8* morton_data, u8* linear_data) {
const bool morton_to_linear = mode == MortonSwizzleMode::MortonToLinear;
u8* data_ptrs[2];
for (u32 y = 0; y < height; ++y) {
for (u32 x = 0; x < width; ++x) {
const u32 coarse_y = y & ~127;
const u32 morton_offset =
GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
const u32 linear_pixel_index = (x + y * width) * linear_bytes_per_pixel;
data_ptrs[morton_to_linear ? 1 : 0] = morton_data + morton_offset;
data_ptrs[morton_to_linear ? 0 : 1] = &linear_data[linear_pixel_index];
std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
}
}
}
} // namespace VideoCore } // namespace VideoCore

View file

@ -15,7 +15,4 @@ void MortonSwizzle(MortonSwizzleMode mode, VideoCore::Surface::PixelFormat forma
u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing, u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
u8* buffer, u8* addr); u8* buffer, u8* addr);
void MortonCopyPixels128(MortonSwizzleMode mode, u32 width, u32 height, u32 bytes_per_pixel,
u32 linear_bytes_per_pixel, u8* morton_data, u8* linear_data);
} // namespace VideoCore } // namespace VideoCore

View file

@ -36,8 +36,7 @@ public:
virtual ~RendererBase(); virtual ~RendererBase();
/// Swap buffers (render frame) /// Swap buffers (render frame)
virtual void SwapBuffers( virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) = 0;
/// Initialize the renderer /// Initialize the renderer
virtual bool Init() = 0; virtual bool Init() = 0;

View file

@ -101,9 +101,7 @@ RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::Syst
RendererOpenGL::~RendererOpenGL() = default; RendererOpenGL::~RendererOpenGL() = default;
void RendererOpenGL::SwapBuffers( void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
system.GetPerfStats().EndSystemFrame(); system.GetPerfStats().EndSystemFrame();
// Maintain the rasterizer's state as a priority // Maintain the rasterizer's state as a priority
@ -113,9 +111,9 @@ void RendererOpenGL::SwapBuffers(
if (framebuffer) { if (framebuffer) {
// If framebuffer is provided, reload it from memory to a texture // If framebuffer is provided, reload it from memory to a texture
if (screen_info.texture.width != (GLsizei)framebuffer->get().width || if (screen_info.texture.width != static_cast<GLsizei>(framebuffer->width) ||
screen_info.texture.height != (GLsizei)framebuffer->get().height || screen_info.texture.height != static_cast<GLsizei>(framebuffer->height) ||
screen_info.texture.pixel_format != framebuffer->get().pixel_format) { screen_info.texture.pixel_format != framebuffer->pixel_format) {
// Reallocate texture if the framebuffer size has changed. // Reallocate texture if the framebuffer size has changed.
// This is expected to not happen very often and hence should not be a // This is expected to not happen very often and hence should not be a
// performance problem. // performance problem.
@ -149,43 +147,43 @@ void RendererOpenGL::SwapBuffers(
* Loads framebuffer from emulated memory into the active OpenGL texture. * Loads framebuffer from emulated memory into the active OpenGL texture.
*/ */
void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer) { void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuffer) {
const u32 bytes_per_pixel{Tegra::FramebufferConfig::BytesPerPixel(framebuffer.pixel_format)};
const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset};
// Framebuffer orientation handling // Framebuffer orientation handling
framebuffer_transform_flags = framebuffer.transform_flags; framebuffer_transform_flags = framebuffer.transform_flags;
framebuffer_crop_rect = framebuffer.crop_rect; framebuffer_crop_rect = framebuffer.crop_rect;
// Ensure no bad interactions with GL_UNPACK_ALIGNMENT, which by default const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset};
// only allows rows to have a memory alignement of 4. if (rasterizer->AccelerateDisplay(framebuffer, framebuffer_addr, framebuffer.stride)) {
ASSERT(framebuffer.stride % 4 == 0); return;
if (!rasterizer->AccelerateDisplay(framebuffer, framebuffer_addr, framebuffer.stride)) {
// Reset the screen info's display texture to its own permanent texture
screen_info.display_texture = screen_info.texture.resource.handle;
rasterizer->FlushRegion(ToCacheAddr(Memory::GetPointer(framebuffer_addr)), size_in_bytes);
constexpr u32 linear_bpp = 4;
VideoCore::MortonCopyPixels128(VideoCore::MortonSwizzleMode::MortonToLinear,
framebuffer.width, framebuffer.height, bytes_per_pixel,
linear_bpp, Memory::GetPointer(framebuffer_addr),
gl_framebuffer_data.data());
glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride));
// Update existing texture
// TODO: Test what happens on hardware when you change the framebuffer dimensions so that
// they differ from the LCD resolution.
// TODO: Applications could theoretically crash yuzu here by specifying too large
// framebuffer sizes. We should make sure that this cannot happen.
glTextureSubImage2D(screen_info.texture.resource.handle, 0, 0, 0, framebuffer.width,
framebuffer.height, screen_info.texture.gl_format,
screen_info.texture.gl_type, gl_framebuffer_data.data());
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
} }
// Reset the screen info's display texture to its own permanent texture
screen_info.display_texture = screen_info.texture.resource.handle;
const auto pixel_format{
VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)};
const u32 bytes_per_pixel{VideoCore::Surface::GetBytesPerPixel(pixel_format)};
const u64 size_in_bytes{framebuffer.stride * framebuffer.height * bytes_per_pixel};
const auto host_ptr{Memory::GetPointer(framebuffer_addr)};
rasterizer->FlushRegion(ToCacheAddr(host_ptr), size_in_bytes);
// TODO(Rodrigo): Read this from HLE
constexpr u32 block_height_log2 = 4;
VideoCore::MortonSwizzle(VideoCore::MortonSwizzleMode::MortonToLinear, pixel_format,
framebuffer.stride, block_height_log2, framebuffer.height, 0, 1, 1,
gl_framebuffer_data.data(), host_ptr);
glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride));
// Update existing texture
// TODO: Test what happens on hardware when you change the framebuffer dimensions so that
// they differ from the LCD resolution.
// TODO: Applications could theoretically crash yuzu here by specifying too large
// framebuffer sizes. We should make sure that this cannot happen.
glTextureSubImage2D(screen_info.texture.resource.handle, 0, 0, 0, framebuffer.width,
framebuffer.height, screen_info.texture.gl_format,
screen_info.texture.gl_type, gl_framebuffer_data.data());
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
} }
/** /**
@ -276,22 +274,29 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
texture.height = framebuffer.height; texture.height = framebuffer.height;
texture.pixel_format = framebuffer.pixel_format; texture.pixel_format = framebuffer.pixel_format;
const auto pixel_format{
VideoCore::Surface::PixelFormatFromGPUPixelFormat(framebuffer.pixel_format)};
const u32 bytes_per_pixel{VideoCore::Surface::GetBytesPerPixel(pixel_format)};
gl_framebuffer_data.resize(texture.width * texture.height * bytes_per_pixel);
GLint internal_format; GLint internal_format;
switch (framebuffer.pixel_format) { switch (framebuffer.pixel_format) {
case Tegra::FramebufferConfig::PixelFormat::ABGR8: case Tegra::FramebufferConfig::PixelFormat::ABGR8:
internal_format = GL_RGBA8; internal_format = GL_RGBA8;
texture.gl_format = GL_RGBA; texture.gl_format = GL_RGBA;
texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV; texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV;
gl_framebuffer_data.resize(texture.width * texture.height * 4); break;
case Tegra::FramebufferConfig::PixelFormat::RGB565:
internal_format = GL_RGB565;
texture.gl_format = GL_RGB;
texture.gl_type = GL_UNSIGNED_SHORT_5_6_5;
break; break;
default: default:
internal_format = GL_RGBA8; internal_format = GL_RGBA8;
texture.gl_format = GL_RGBA; texture.gl_format = GL_RGBA;
texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV; texture.gl_type = GL_UNSIGNED_INT_8_8_8_8_REV;
gl_framebuffer_data.resize(texture.width * texture.height * 4); UNIMPLEMENTED_MSG("Unknown framebuffer pixel format: {}",
LOG_CRITICAL(Render_OpenGL, "Unknown framebuffer pixel format: {}", static_cast<u32>(framebuffer.pixel_format));
static_cast<u32>(framebuffer.pixel_format));
UNREACHABLE();
} }
texture.resource.Release(); texture.resource.Release();

View file

@ -43,14 +43,13 @@ struct ScreenInfo {
TextureInfo texture; TextureInfo texture;
}; };
class RendererOpenGL : public VideoCore::RendererBase { class RendererOpenGL final : public VideoCore::RendererBase {
public: public:
explicit RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system); explicit RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system);
~RendererOpenGL() override; ~RendererOpenGL() override;
/// Swap buffers (render frame) /// Swap buffers (render frame)
void SwapBuffers( void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;
/// Initialize the renderer /// Initialize the renderer
bool Init() override; bool Init() override;

View file

@ -445,11 +445,12 @@ PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat
switch (format) { switch (format) {
case Tegra::FramebufferConfig::PixelFormat::ABGR8: case Tegra::FramebufferConfig::PixelFormat::ABGR8:
return PixelFormat::ABGR8U; return PixelFormat::ABGR8U;
case Tegra::FramebufferConfig::PixelFormat::RGB565:
return PixelFormat::B5G6R5U;
case Tegra::FramebufferConfig::PixelFormat::BGRA8: case Tegra::FramebufferConfig::PixelFormat::BGRA8:
return PixelFormat::BGRA8; return PixelFormat::BGRA8;
default: default:
LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format)); UNIMPLEMENTED_MSG("Unimplemented format={}", static_cast<u32>(format));
UNREACHABLE();
return PixelFormat::ABGR8U; return PixelFormat::ABGR8U;
} }
} }