blender
/
blender
136
407
Fork 589
Code Issues 6.3k Pull Requests 779 Projects 19 Wiki Activity

WIP: Functions: new local allocator for better memory reuse and performance #104630

Draft
Jacques Lucke wants to merge 44 commits from JacquesLucke/blender:local-allocator into main
pull from: JacquesLucke/blender:local-allocator
merge into: blender:main

When changing the target branch, be careful to rebase the branch in your fork to match. See documentation.
Conversation 0 Commits 44 Files Changed 13 +5 -23
1 changed files with 5 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 630a44d18e - Show all commits

28
source/blender/blenlib/BLI_local_pool.hh
View File

@ -6,6 +6,7 @@
#include "BLI_allocator.hh"
#include "BLI_asan.h"
#include "BLI_enumerable_thread_specific.hh"
#include "BLI_linear_allocator.hh"
#include "BLI_map.hh"
#include "BLI_math_bits.h"
#include "BLI_stack.hh"
@ -17,8 +18,7 @@ namespace blender {
template<typename Allocator = GuardedAllocator> class LocalPool : NonCopyable, NonMovable {
private:
static constexpr int64_t s_alignment = 64;
Vector<MutableSpan<std::byte>> owned_buffers_;
LinearAllocator<> linear_allocator_;
struct BufferStack {
int64_t element_size = -1;
@ -28,8 +28,6 @@ template<typename Allocator = GuardedAllocator> class LocalPool : NonCopyable, N
std::array<BufferStack, 8> small_stacks_;
std::unique_ptr<Map<int, BufferStack>> large_stacks_;
BLI_NO_UNIQUE_ADDRESS Allocator allocator_;
public:
LocalPool()
{
@ -40,10 +38,6 @@ template<typename Allocator = GuardedAllocator> class LocalPool : NonCopyable, N
~LocalPool()
{
for (MutableSpan<std::byte> buffer : owned_buffers_) {
BLI_asan_unpoison(buffer.data(), buffer.size());
allocator_.deallocate(buffer.data());
}
}
void *allocate(const int64_t size, const int64_t alignment)
@ -56,22 +50,10 @@ template<typename Allocator = GuardedAllocator> class LocalPool : NonCopyable, N
return buffer;
}
if (size <= 4096) {
const int64_t allocation_size = std::clamp<int64_t>(
buffer_stack.element_size * 16, 512, 4096);
void *buffer = allocator_.allocate(allocation_size, s_alignment, __func__);
BLI_asan_poison(buffer, allocation_size);
const int64_t num = allocation_size / buffer_stack.element_size;
for (int64_t i = num - 1; i > 0; i--) {
buffer_stack.stack.push(POINTER_OFFSET(buffer, buffer_stack.element_size * i));
}
owned_buffers_.append({static_cast<std::byte *>(buffer), allocation_size});
BLI_asan_unpoison(buffer, size);
return buffer;
return linear_allocator_.allocate(size, alignment);
}
void *buffer = allocator_.allocate(
size_t(size), std::max<size_t>(s_alignment, size_t(alignment)), __func__);
owned_buffers_.append({static_cast<std::byte *>(buffer), size});
return buffer;
return linear_allocator_.allocate(size_t(size),
std::max<size_t>(s_alignment, size_t(alignment)));
}
void deallocate(const void *buffer, const int64_t size, const int64_t alignment)