Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

#

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Handle-local read windows may be one ahead of global ones to alleviate contention on few remaining blocks in window #4

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Handle-local read windows may be one ahead of global ones to alleviate contention on few remaining blocks in window #4

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 29 additions & 1 deletion relaxed_concurrent_fifo/block_based_queue.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,6 +67,7 @@ class block_based_queue {
static_assert(sizeof(block_t) == CELLS_PER_BLOCK * sizeof(T) + sizeof(header_t));

struct window_t {
alignas(std::hardware_destructive_interference_size) atomic_bitset<BLOCKS_PER_WINDOW, BITSET_TYPE> touched_set;
alignas(std::hardware_destructive_interference_size) atomic_bitset<BLOCKS_PER_WINDOW, BITSET_TYPE> filled_set;
alignas(std::hardware_destructive_interference_size) block_t blocks[BLOCKS_PER_WINDOW];
};
Expand Down Expand Up @@ -140,6 +141,7 @@ class block_based_queue {
window_index = fifo.write_window.load(std::memory_order_relaxed);
window = &fifo.get_window(window_index);
free_bit = window->filled_set.template claim_bit<false, true>(std::memory_order_relaxed);
// The touched set can be missed, which might trigger a reader to attempt to move, but the filled set will prevent the move from occuring.
if (free_bit == std::numeric_limits<std::size_t>::max()) {
// No more free bits, we move.
if (window_index + 1 - fifo.read_window.load(std::memory_order_relaxed) == fifo.window_count) {
Expand All @@ -150,6 +152,7 @@ class block_based_queue {
std::cout << "Write move " << (window_index + 1) << std::endl;
#endif // LOG_WINDOW_MOVE
} else {
window->touched_set.set(free_bit, std::memory_order_relaxed);
break;
}
} while (true);
Expand All @@ -163,12 +166,36 @@ class block_based_queue {
std::size_t free_bit;
std::uint64_t window_index;
window_t* window;
bool dont_advance = false;
do {
bool is_ahead = false;
window_index = fifo.read_window.load(std::memory_order_relaxed);
if (!dont_advance && window_index + 1 == read_window) {
is_ahead = true;
window_index = read_window;
}
window = &fifo.get_window(window_index);
free_bit = window->filled_set.template claim_bit<true, false>(std::memory_order_relaxed);
free_bit = window->touched_set.template claim_bit<true, true>(std::memory_order_relaxed);
if (free_bit == std::numeric_limits<std::size_t>::max()) {
if (is_ahead) {
dont_advance = true;
continue;
}

std::uint64_t write_window = fifo.write_window.load(std::memory_order_relaxed);

// Don't go ahead if write_window is just ahead of us (so we don't have to force move).
if (!dont_advance && window_index + 1 != write_window) {
read_window = window_index + 1;
// A few redundant ops, fine for now.
continue;
}

free_bit = window->filled_set.template claim_bit<true, false>();
if (free_bit != std::numeric_limits<std::size_t>::max()) {
break;
}

if (write_window == window_index + 1) {
if (!fifo.get_window(write_window).filled_set.any(std::memory_order_relaxed)) {
return false;
Expand Down Expand Up @@ -221,6 +248,7 @@ class block_based_queue {
// We're abandoning an empty block!
window_t& window = fifo.get_window(write_window);
auto diff = write_block - window.blocks;
window.touched_set.reset(diff, std::memory_order_relaxed);
window.filled_set.reset(diff, std::memory_order_relaxed);
}
if (!claim_new_block_write()) {
Expand Down