fabi/Xenia-Canary
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Drastically reduce cpu time wasted by XMADecoderThread spinning, went from 13% of all cpu time to about 0.6% in my tests

Browse Source 
Commented out lock in WatchMemoryRange, lock is always held by caller
properly set the value/check the irql for spinlocks in xboxkrnl_threading
This commit is contained in:
chss95cs@gmail.com
2022-10-15 03:07:07 -07:00
parent ecf6bfbbdf
commit efbeae660c
6 changed files with 62 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/xenia/gpu/d3d12/d3d12_command_processor.cc
View File

@@ -4357,7 +4357,7 @@ bool D3D12CommandProcessor::UpdateBindings(
uint32_t float_constant_index;
while (xe::bit_scan_forward(float_constant_map_entry,
&float_constant_index)) {
float_constant_map_entry &= ~(1ull << float_constant_index);
float_constant_map_entry = xe::clear_lowest_bit(float_constant_map_entry);
std::memcpy(float_constants,
&regs[XE_GPU_REG_SHADER_CONSTANT_000_X + (i << 8) +
(float_constant_index << 2)]
@@ -4388,7 +4388,7 @@ bool D3D12CommandProcessor::UpdateBindings(
uint32_t float_constant_index;
while (xe::bit_scan_forward(float_constant_map_entry,
&float_constant_index)) {
float_constant_map_entry &= ~(1ull << float_constant_index);
float_constant_map_entry = xe::clear_lowest_bit(float_constant_map_entry);
std::memcpy(float_constants,
&regs[XE_GPU_REG_SHADER_CONSTANT_256_X + (i << 8) +
(float_constant_index << 2)]

6
src/xenia/gpu/d3d12/d3d12_command_processor.h
View File

@@ -680,9 +680,6 @@ class D3D12CommandProcessor final : public CommandProcessor {
ID3D12Resource* readback_buffer_ = nullptr;
uint32_t readback_buffer_size_ = 0;
std::atomic<bool> pix_capture_requested_ = false;
bool pix_capturing_;
// The current fixed-function drawing state.
D3D12_VIEWPORT ff_viewport_;
D3D12_RECT ff_scissor_;
@@ -776,6 +773,9 @@ class D3D12CommandProcessor final : public CommandProcessor {
// scratch memexport data
MemExportRange memexport_ranges_[512];
uint32_t memexport_range_count_ = 0;
std::atomic<bool> pix_capture_requested_ = false;
bool pix_capturing_;
};
} // namespace d3d12

4
src/xenia/gpu/shared_memory.cc
View File

@@ -150,8 +150,8 @@ SharedMemory::WatchHandle SharedMemory::WatchMemoryRange(
watch_page_first << page_size_log2_ >> kWatchBucketSizeLog2;
uint32_t bucket_last =
watch_page_last << page_size_log2_ >> kWatchBucketSizeLog2;
auto global_lock = global_critical_region_.Acquire();
//chrispy: Not required the global lock is always held by the caller
// auto global_lock = global_critical_region_.Acquire();
// Allocate the range.
WatchRange* range = watch_range_first_free_;