debuggerx/eden
1
0
Fork 0
mirror of https://git.eden-emu.dev/eden-emu/eden.git synced 2026-08-18 22:23:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

[video_core/engines/maxwell3d] memory inline DrawState to reduce indirection on hot paths (#3758)

Browse Source 
usual indirection removal
helps very slightly to codegen

the idea is basically to reduce the amount of pointer deference overall in the code, and use idiomatic std::variant<>-isms to not rely on vtables/unique_ptr overhead
this should allow the compiler to emit better code
of course it's a tiny optimisation and only CPU side, but allows us to reduce indirection which is almost always a good thing

"but youre passing more parameters to the function!!!" its literally memoized into a register my friend

Signed-off-by: lizzie <lizzie@eden-emu.dev>
Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/3758
Reviewed-by: crueter <crueter@eden-emu.dev>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
This commit is contained in:
lizzie
2026-04-28 17:44:51 +02:00
committed by crueter
parent ed225f8a8b
commit 77decca678
15 changed files with 231 additions and 351 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/video_core/buffer_cache/buffer_cache.h
+13 -19
View File
@@ -356,7 +356,7 @@ void BufferCache<P>::BindHostGeometryBuffers(bool is_indexed) {
if (is_indexed) {
BindHostIndexBuffer();
} else if constexpr (!HAS_FULL_INDEX_AND_PRIMITIVE_SUPPORT) {
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
const auto& draw_state = maxwell3d->draw_manager.draw_state;
if (draw_state.topology == Maxwell::PrimitiveTopology::Quads ||
draw_state.topology == Maxwell::PrimitiveTopology::QuadStrip) {
runtime.BindQuadIndexBuffer(draw_state.topology, draw_state.vertex_buffer.first,
@@ -740,30 +740,25 @@ void BufferCache<P>::BindHostIndexBuffer() {
TouchBuffer(buffer, channel_state->index_buffer.buffer_id);
const u32 offset = buffer.Offset(channel_state->index_buffer.device_addr);
const u32 size = channel_state->index_buffer.size;
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
if (!draw_state.inline_index_draw_indexes.empty()) [[unlikely]] {
const auto& draw_state = maxwell3d->draw_manager.draw_state;
if (draw_state.inline_index_draw_indexes.empty()) {
SynchronizeBuffer(buffer, channel_state->index_buffer.device_addr, size);
} else {
if constexpr (USE_MEMORY_MAPS_FOR_UPLOADS) {
auto upload_staging = runtime.UploadStagingBuffer(size);
std::array<BufferCopy, 1> copies{
{BufferCopy{.src_offset = upload_staging.offset, .dst_offset = 0, .size = size}}};
std::memcpy(upload_staging.mapped_span.data(),
draw_state.inline_index_draw_indexes.data(), size);
std::array<BufferCopy, 1> copies{{BufferCopy{.src_offset = upload_staging.offset, .dst_offset = 0, .size = size}}};
std::memcpy(upload_staging.mapped_span.data(), draw_state.inline_index_draw_indexes.data(), size);
runtime.CopyBuffer(buffer, upload_staging.buffer, copies, true);
} else {
buffer.ImmediateUpload(0, draw_state.inline_index_draw_indexes);
}
} else {
SynchronizeBuffer(buffer, channel_state->index_buffer.device_addr, size);
}
if constexpr (HAS_FULL_INDEX_AND_PRIMITIVE_SUPPORT) {
const u32 new_offset =
offset + draw_state.index_buffer.first * draw_state.index_buffer.FormatSizeInBytes();
const u32 new_offset = offset + draw_state.index_buffer.first * draw_state.index_buffer.FormatSizeInBytes();
runtime.BindIndexBuffer(buffer, new_offset, size);
} else {
buffer.MarkUsage(offset, size);
runtime.BindIndexBuffer(draw_state.topology, draw_state.index_buffer.format,
draw_state.index_buffer.first, draw_state.index_buffer.count,
buffer, offset, size);
runtime.BindIndexBuffer(draw_state.topology, draw_state.index_buffer.format, draw_state.index_buffer.first, draw_state.index_buffer.count, buffer, offset, size);
}
}
@@ -945,10 +940,9 @@ void BufferCache<P>::BindHostGraphicsUniformBuffer(size_t stage, u32 index, u32
return alignment > 1 && (offset % alignment) != 0;
}
}();
const bool use_fast_buffer = needs_alignment_stream ||
(has_host_buffer &&
size <= channel_state->uniform_buffer_skip_cache_size &&
!memory_tracker.IsRegionGpuModified(device_addr, size));
const bool use_fast_buffer = needs_alignment_stream
|| (has_host_buffer && size <= channel_state->uniform_buffer_skip_cache_size
&& !memory_tracker.IsRegionGpuModified(device_addr, size));
if (use_fast_buffer) {
if constexpr (IS_OPENGL) {
if (runtime.HasFastBufferSubData()) {
@@ -1229,7 +1223,7 @@ template <class P>
void BufferCache<P>::UpdateIndexBuffer() {
// We have to check for the dirty flags and index count
// The index count is currently changed without updating the dirty flags
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
const auto& draw_state = maxwell3d->draw_manager.draw_state;
const auto& index_buffer_ref = draw_state.index_buffer;
auto& flags = maxwell3d->dirty.flags;
if (!flags[Dirty::IndexBuffer]) {
src/video_core/buffer_cache/buffer_cache_base.h
+3 -3
View File
@@ -32,7 +32,7 @@
#include "video_core/control/channel_state_cache.h"
#include "video_core/delayed_destruction_ring.h"
#include "video_core/dirty_flags.h"
#include "video_core/engines/draw_manager.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/engines/kepler_compute.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/memory_manager.h"
@@ -305,7 +305,7 @@ public:
[[nodiscard]] bool IsRegionCpuModified(DAddr addr, size_t size);
void SetDrawIndirect(
const Tegra::Engines::DrawManager::IndirectParams* current_draw_indirect_) {
const Tegra::Engines::Maxwell3D::DrawManager::IndirectParams* current_draw_indirect_) {
current_draw_indirect = current_draw_indirect_;
}
@@ -480,7 +480,7 @@ private:
#endif
DelayedDestructionRing<Buffer, TICKS_TO_DESTROY> delayed_destruction_ring;
const Tegra::Engines::DrawManager::IndirectParams* current_draw_indirect{};
const Tegra::Engines::Maxwell3D::DrawManager::IndirectParams* current_draw_indirect{};
u32 last_index_count = 0;