diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/queue_context.cc | 452 | ||||
| -rw-r--r-- | src/queue_context.hh | 39 |
2 files changed, 177 insertions, 314 deletions
diff --git a/src/queue_context.cc b/src/queue_context.cc index 90039f2..7121898 100644 --- a/src/queue_context.cc +++ b/src/queue_context.cc @@ -6,6 +6,7 @@ #include <chrono> #include <iostream> #include <limits> +#include <ranges> #include <span> namespace low_latency { @@ -141,119 +142,41 @@ void QueueContext::notify_submit( void QueueContext::notify_present(const VkPresentInfoKHR& info) { - const auto waits = [&]() { - auto waits = std::unordered_set<VkSemaphore>{}; - std::ranges::copy_n(info.pWaitSemaphores, info.waitSemaphoreCount, - std::inserter(waits, std::end(waits))); - return waits; - }(); - - const auto collected_semaphores = [&info, this]() { - auto collected_semaphores = std::unordered_set<VkSemaphore>{}; - for (auto i = std::uint32_t{0}; i < info.swapchainCount; ++i) { - const auto& swapchain = info.pSwapchains[i]; - const auto& index = info.pImageIndices[i]; - - // Shouldn't be possible to present to a swapchain that wasn't - // waited in - - const auto& signals = this->device_context.swapchain_signals; - const auto swapchain_it = signals.find(swapchain); - assert(swapchain_it != std::end(signals)); - const auto index_it = swapchain_it->second.find(index); - assert(index_it != std::end(swapchain_it->second)); - - const auto& semaphore = index_it->second; - collected_semaphores.emplace(index_it->second); - } - return collected_semaphores; - }(); - - const auto acquire_iter = std::ranges::find_if( - std::rbegin(this->submissions), std::rend(this->submissions), - [&](const auto& submission) { - return std::ranges::any_of( - submission->waits, [&](const auto& wait) { - return collected_semaphores.contains(wait); - }); - }); - - if (acquire_iter == std::rend(this->submissions)) { - std::cerr << "couldn't find starting submission!\n"; - return; - } - const auto& acquire = *acquire_iter; - (*acquire_iter)->debug += "acquire "; - - const auto present_iter = std::ranges::find_if( - std::rbegin(this->submissions), std::rend(this->submissions), - [&](const auto& submission) { - return std::ranges::any_of( - submission->signals, [&](const auto& signal) { - if (waits.contains(signal)) { - std::cerr - << "queue with signal matching present iter found: " - << signal << '\n'; - } - - return waits.contains(signal); - }); - }); - - if (present_iter == std::rend(this->submissions)) { - std::cerr << "couldn't find ending submission!\n"; + // We are going to assume that all queue submissions before and on the same + // queue contribute to the frame. + + // This used to be more complicated where we found the first submission that + // was signalled by acquire, then we walked forwards until we found the + // submission before it that marked the end of frame (which was the last + // submission in the previous frame that called notify submit). This seemed + // completely redundant, in all cases it was exactly what we have here. But + // I could be wrong. + + // that wa + const auto start_iter = std::begin(this->submissions); + // no op submit? + if (start_iter == std::end(this->submissions)) { + std::cerr << "ignored no op submit\n"; return; } - (*present_iter)->debug += "present "; - const auto& end = *present_iter; - - std::cerr << "present for queue: " << queue << ", our waits:\n"; - for (const auto& wait : waits) { - std::cerr << " " << wait << '\n'; - } - - // The work including and between acquire -> present is effectively - // guaranteed to contribute to our frame. We are going to mark this point - // for future queues to read the 'start of frame' from. - (*present_iter)->end_of_frame_marker = true; - - // Now we read backwards to try to find our true start, starting at our - // acquire. - const auto after_previous_present_iter = std::prev(std::ranges::find_if( - std::next(acquire_iter), std::rend(this->submissions), - [](const auto& submission) { - return submission->end_of_frame_marker; - })); - - (*after_previous_present_iter)->debug += "after_previous_present "; - - this->submissions.back()->debug += "last_in_host_present "; - - std::cerr << "the present iter was dist from back of: " - << std::distance(std::rbegin(this->submissions), present_iter) - << '\n'; - std::cerr << "the acquire iter was dist from back of: " - << std::distance(std::rbegin(this->submissions), acquire_iter) - << '\n'; - std::cerr << "the before_previous_present iter was dist from back of: " - << std::distance(std::rbegin(this->submissions), - after_previous_present_iter) - << '\n'; - - auto frame = - Frame{.start = - Frame::Timepoint{ - .context = *this, - .handle = (*after_previous_present_iter)->start_handle, - .sequence = (*after_previous_present_iter)->sequence, - }, - .end = Frame::Timepoint{ - .context = *this, - .handle = (*present_iter)->end_handle, - .sequence = (*present_iter)->sequence, - }}; - this->in_flight_frames.emplace_back( - std::make_unique<Frame>(std::move(frame))); + const auto last_iter = std::prev(std::end(this->submissions)); + + (*start_iter)->debug += "first_during_present "; + (*last_iter)->debug += "last_during_present "; + + // TODO We probably want to walk through other graphics queues and get + // the time that they submitted at some stage. Somewhat difficult to do, + // we probably need to create a graph of dependencies but it seems like + // overengineering. I will just get this working well before I deal + // with complicated edge cases. + + this->in_flight_frames.emplace_back(Frame{ + .submissions = std::move(this->submissions), + .sequence = (*last_iter)->sequence, + }); + assert(std::size(this->in_flight_frames.back().submissions)); + // *valid but unspecified state after move, so clear!* + this->submissions.clear(); } const auto debug_log_time = [](const auto& diff) { @@ -269,74 +192,106 @@ void QueueContext::process_frames() { return; } - // Collect all devices and call calibrate. - [this]() -> void { - using context_ref_t = std::reference_wrapper<DeviceContext>; - auto target_devices = std::unordered_map<VkDevice, context_ref_t>{}; - for (const auto& frame : this->in_flight_frames) { - auto& start = frame->start.context.device_context; - auto& end = frame->end.context.device_context; - - target_devices.try_emplace(start.device, std::ref(start)); - target_devices.try_emplace(end.device, std::ref(end)); - } - for (const auto& pair : target_devices) { - auto& device = pair.second.get(); - device.clock.calibrate(); - } + // We used to collect all devices that were pointed to by all potential + // submissions, put them in a set and then call.calibrate() on each once. + // This is unnecessary now - we can assume all submissions come from the + // same queue (this one!). + auto& device_context = this->device_context; + auto& clock = device_context.clock; + clock.calibrate(); + + // Get the queue's sequence number so we can quickly check + // frames are finished without calling getCalibratedTimestamps. + // This is somewhat a premature optimization but it's elegant. + const auto seq = [&, this]() -> auto { + auto seq = std::uint64_t{0}; + device_context.vtable.GetSemaphoreCounterValueKHR( + device_context.device, this->semaphore, &seq); + return seq; }(); - const auto get_tick_time = [](const auto& timepoint) - -> std::optional<DeviceContext::Clock::time_point_t> { - const auto& handle = timepoint.handle; - const auto& context = timepoint.context; - - const auto ticks = handle->get_ticks(*context.timestamp_pool); - if (!ticks.has_value()) { - return std::nullopt; - } - const auto& clock = context.device_context.clock; - return clock.ticks_to_time(*ticks); - }; - - std::cerr << "starting frame readout\n"; while (std::size(this->in_flight_frames)) { const auto& frame = this->in_flight_frames.front(); - assert(frame); - const auto a = get_tick_time(frame->start); - if (!a.has_value()) { - break; - } + // There should be at least one submission, we guarantee it in + // notify_present. + assert(std::size(frame.submissions)); + + const auto& last_submission = frame.submissions.back(); - const auto b = get_tick_time(frame->end); - if (!b.has_value()) { + // Not completed (so future frames definitely aren't) - stop early. + if (seq < last_submission->sequence) { break; } - // We are committed to removing the frame at this stage and promoting it - // to a 'timing' struct because it's completed.. + // We are committed to removing the frame at this stage and + // promoting it to a 'timing' struct because it's completed. + // We can guarantee that we can extract timing information from + // all start/end handles now. + + // Using leetcode merge intervals in the wild lol + struct Interval { + DeviceContext::Clock::time_point_t start, end; + }; + + const auto sorted_intervals = [&, this]() -> auto { + auto intervals = std::vector<Interval>{}; + std::ranges::transform( + frame.submissions, std::back_inserter(intervals), + [&, this](const auto& submission) { + const auto get_time = [&, this](const auto& handle) { + const auto t = handle->get_ticks(*this->timestamp_pool); + assert(t.has_value()); // guaranteed from seq. + return clock.ticks_to_time(*t); + }; + + return Interval{ + .start = get_time(submission->start_handle), + .end = get_time(submission->end_handle), + }; + }); - const auto frametime = *b - *a; + std::ranges::sort(intervals, [](const auto& a, const auto& b) { + return a.start < b.start; + }); + return intervals; + }(); - const auto cpu_time = - [&]() -> DeviceContext::Clock::time_point_t::duration { - const auto latest_iter = std::rbegin(this->timings); - if (latest_iter == std::rend(this->timings)) { - return DeviceContext::Clock::time_point_t::duration::zero(); + const auto merged = [&]() -> auto { + auto merged = std::vector<Interval>{}; + auto last = sorted_intervals[0]; + + for (const auto& [s, e] : sorted_intervals | std::views::drop(1)) { + if (s <= last.end) { + last.end = std::max(last.end, e); + } else { + merged.push_back(last); + last = {s, e}; + } } - return *a - (*latest_iter)->gpu_end; + merged.push_back(last); + return merged; }(); - std::cerr - << " calculated total time from last frame (frametime): "; - debug_log_time(*b - *a); + const auto gputime = std::ranges::fold_left( + merged, DeviceContext::Clock::time_point_t::duration{}, + [](auto gputime, const auto& interval) { + const auto& [start, end] = interval; + return gputime + (end - start); + }); + + const auto start = merged.front().start; + const auto end = merged.back().end; + const auto not_gputime = + (merged.back().end - merged.front().start) - gputime; + + auto timing = Timing{ + .gputime = gputime, + .not_gputime = not_gputime, + .frame = frame, + }; + this->timings.emplace_back(std::make_unique<Timing>(timing)); - this->timings.emplace_back(std::make_unique<Timing>( - Timing{.gpu_end = *b, - .gpu_time = frametime, - .cpu_time = cpu_time, - .frame = std::move(this->in_flight_frames.front())})); this->in_flight_frames.pop_front(); } @@ -355,22 +310,23 @@ void QueueContext::sleep_in_present() { const auto& vtable = device.vtable; // Call this to push all in flight frames into our timings structure, - // but only if they're completed. So now they are truly *in flight frames*. + // but only if they're completed. So now they are truly *in flight + // frames*. this->process_frames(); if (const auto F = std::size(this->in_flight_frames); F > 1) { - // In this case, we are so far ahead that there are multiple frames in - // flight. Either that, or our bookkeeping has gone horribly wrong! Wait - // on the 2nd last frame in flight to complete. This shunts us to F=1. + // In this case, we are so far ahead that there are multiple frames + // in flight. Either that, or our bookkeeping has gone horribly + // wrong! Wait on the 2nd last frame in flight to complete. This + // shunts us to F=1. const auto second_iter = std::next(std::rbegin(this->in_flight_frames)); assert(second_iter != std::rend(this->in_flight_frames)); - const auto& frame = (*second_iter)->end.sequence; const auto swi = VkSemaphoreWaitInfo{ .sType = VK_STRUCTURE_TYPE_SEMAPHORE_WAIT_INFO, .semaphoreCount = 1, .pSemaphores = &this->semaphore, - .pValues = &(*second_iter)->end.sequence, + .pValues = &second_iter->sequence, }; vtable.WaitSemaphoresKHR(device.device, &swi, std::numeric_limits<std::uint64_t>::max()); @@ -382,6 +338,8 @@ void QueueContext::sleep_in_present() { return; } + // We are checking size again because process_frames might have drained + // it to zero. if (!std::size(this->in_flight_frames)) { return; } @@ -397,34 +355,34 @@ void QueueContext::sleep_in_present() { std::ranges::transform(this->timings, std::back_inserter(vect), [](const auto& timing) { return timing.get(); }); std::ranges::sort(vect, [](const auto& a, const auto& b) { - return a->gpu_time < b->gpu_time; + return a->gputime < b->gputime; }); // return vect[0]->frametime; - return vect[std::size(vect) / 2]->gpu_time; + return vect[std::size(vect) / 2]->gputime; }(); - const auto expected_cputime = [&, this]() { + const auto expected_not_gputime = [&, this]() { auto vect = std::vector<Timing*>{}; std::ranges::transform(this->timings, std::back_inserter(vect), [](const auto& timing) { return timing.get(); }); std::ranges::sort(vect, [](const auto& a, const auto& b) { - return a->cpu_time < b->cpu_time; + return a->not_gputime < b->not_gputime; }); // return vect[0]->frametime; - return vect[std::size(vect) / 2]->cpu_time; + return vect[std::size(vect) / 2]->not_gputime; }(); + std::cerr << " expected gputime: "; debug_log_time(expected_gputime); - std::cerr << " expected cputime: "; - debug_log_time(expected_cputime); + std::cerr << " expected not_gputime: "; + debug_log_time(expected_not_gputime); // PRESENT CALL - // |--------------|-------------------|----------------| - // a swap_acquire b c + // |----------------------------------|----------------| + // first b c // - // Us, the CPU on the host, is approximately at 'b'. We have a good guess - // for the distance between a and b as gputime. Our educated guess for the - // distance between b and c is cputime. The GPU is between a and b. + // Us, the CPU on the host, is approximately at 'b'. We have a good + // guess for the distance between a and b as gputime. const auto& frame = this->in_flight_frames.back(); @@ -432,137 +390,49 @@ void QueueContext::sleep_in_present() { // It's impossible to make a decision until we know a. // Doing this is fine because it won't affect throughput at all. // (ie, there's more work queued after regardless). + // FIXME: If a == b, then we're waiting for the entire queue + // to finish because the semaphore only says if it has finished. + // The fix is to check the start timestamp instead of the query + // in the case that it's... + // Honestly it might be better to signal two semaphores because + // we need to wait for when the submission starts work and + // right now, we only signal when the submission finishes work. + // Ideally we have both, so we can elegantly wait on the start + // semaphore of A, then get A's start timestamp. This is BROKEN. + [&]() -> void { const auto swi = VkSemaphoreWaitInfo{ .sType = VK_STRUCTURE_TYPE_SEMAPHORE_WAIT_INFO, .semaphoreCount = 1, .pSemaphores = &this->semaphore, - .pValues = &frame->start.sequence, + .pValues = &frame.submissions.front()->sequence, }; vtable.WaitSemaphoresKHR(device.device, &swi, std::numeric_limits<std::uint64_t>::max()); }(); - // We now know that A is available because its semaphore has been signalled. - const auto a_handle = frame->start.handle; - const auto& a_context = frame->start.context; + // We now know that A is available because its semaphore has been + // signalled. - const auto a_ticks_opt = a_handle->get_ticks(*a_context.timestamp_pool); - assert(a_ticks_opt.has_value()); - const auto& a_clock = a_context.device_context.clock; - const auto a = a_clock.ticks_to_time(*a_ticks_opt); + const auto a = [&, this]() { + const auto& handle = frame.submissions.front()->start_handle; + const auto ticks = handle->get_ticks(*this->timestamp_pool); + assert(ticks.has_value()); + return device_context.clock.ticks_to_time(*ticks); + }(); const auto now = std::chrono::steady_clock::now(); const auto dist = now - a; - const auto expected = expected_gputime - dist - expected_cputime; + const auto expected = expected_gputime - dist; const auto swi = VkSemaphoreWaitInfo{ .sType = VK_STRUCTURE_TYPE_SEMAPHORE_WAIT_INFO, .semaphoreCount = 1, .pSemaphores = &this->semaphore, - .pValues = &frame->end.sequence, + .pValues = &frame.sequence, }; vtable.WaitSemaphoresKHR(device.device, &swi, std::max(expected.count(), 0l)); - - /* - - // 2% of average gpu time for dealing with variance. - // This could be calculated more precisely with the - // numbers we have (like we could construct a high% confidence - // interval? not big on maths). - const auto slack = gputime / 50; - - // All educated guesses: - // dist_to_b = gputime - dist_to_last_b; - // dist_to_last_b = now - last_b - // sleep_until = now + extra_delay + slack + dist_to_b - // = now + extra_delay + slack + (frametime - - dist_to_last_b) - // = now + extra_delay + slack + frametime - (now - last_b) - - const auto now = std::chrono::steady_clock::now(); - assert(last_b <= now); - const auto dist = now - last_b; - // Even if this is negative, it's a no-op to sleep backwards. - const auto sleep_target = - now + extra_delay + slack + median_frametime - dist; - */ - - /* - std::cerr << " SLEEPING FOR: "; - debug_log_time(sleep_target - now); - std::this_thread::sleep_until(sleep_target); - */ - - /* - const auto get_tick_time = [](const auto& timepoint) - -> std::optional<DeviceContext::Clock::time_point_t> { - const auto& handle = timepoint.handle; - const auto& context = timepoint.context; - - const auto ticks = handle->get_ticks(*context.timestamp_pool); - if (!ticks.has_value()) { - return std::nullopt; - } - const auto& clock = context.device_context.clock; - return clock.ticks_to_time(*ticks); - }; - - const auto finished_time = - get_tick_time(this->in_flight_frames.back()->end); - const auto after = std::chrono::steady_clock::now(); - if (!finished_time.has_value()) { - std::cerr << "didnt finish late!\n"; - } else { - std::cerr << "finished late by: "; - const auto late_time = after - *finished_time; - } - - const auto get_tick_time2 = [this](const auto& handle) - -> std::optional<DeviceContext::Clock::time_point_t> { - const auto& context = *this; - - const auto ticks = handle->get_ticks(*context.timestamp_pool); - if (!ticks.has_value()) { - return std::nullopt; - } - const auto& clock = context.device_context.clock; - return clock.ticks_to_time(*ticks); - }; - - this->submissions.back()->debug += " last_in_wait"; - - std::cerr << "---------------SUBMISSION READOUT POST FRAME SUBMIT AND " - "WAIT--------------------\n"; - auto i = std::size_t{0}; - for (auto it = std::rbegin(this->submissions); - i < 20 && it != std::rend(this->submissions); ++it, ++i) { - - std::cerr << " submission -" << i << '\n'; - const auto& submission = **it; - - std::cerr << " sequence target: " << submission.sequence << '\n'; - - std::cerr << " debug tags: " << submission.debug << '\n'; - - const auto end_time = get_tick_time2(submission.end_handle); - std::cerr << " end_time: "; - if (end_time.has_value()) { - debug_log_time(after - *end_time); - } else { - std::cerr << "not finished yet!\n"; - } - - const auto start_time = get_tick_time2(submission.start_handle); - std::cerr << " start_time: "; - if (start_time.has_value()) { - debug_log_time(after - *start_time); - } else { - std::cerr << "not finished yet!\n"; - } - } - */ } } // namespace low_latency
\ No newline at end of file diff --git a/src/queue_context.hh b/src/queue_context.hh index 172683d..ed13465 100644 --- a/src/queue_context.hh +++ b/src/queue_context.hh @@ -31,7 +31,8 @@ class QueueContext final : public Context { private: static constexpr auto MAX_TRACKED_TIMINGS = 50; - // Potentially in flight queue submissions + + // Potentially in flight queue submissions that come from this queue. struct Submission { const std::unordered_set<VkSemaphore> signals; const std::unordered_set<VkSemaphore> waits; @@ -41,32 +42,26 @@ class QueueContext final : public Context { std::uint64_t sequence; - bool end_of_frame_marker = false; std::string debug; }; - std::deque<std::shared_ptr<Submission>> submissions; + using submission_ptr_t = std::shared_ptr<Submission>; + std::deque<submission_ptr_t> submissions; - // In flight frames! - // These might come from different contexts. + // In flight frame submissions grouped together. + // The first element in the vector refers to the first submission that + // contributed to that frame. The last element is the last submission before + // present was called. struct Frame { - - struct Timepoint { - const QueueContext& context; - const std::shared_ptr<TimestampPool::Handle> handle; - std::uint64_t sequence; - }; - - const Timepoint start; - const Timepoint end; + std::deque<submission_ptr_t> submissions; + std::uint64_t sequence; }; - std::deque<std::unique_ptr<Frame>> in_flight_frames; + std::deque<Frame> in_flight_frames; + // Completed frames. struct Timing { - DeviceContext::Clock::time_point_t gpu_end; + DeviceContext::Clock::time_point_t::duration gputime, not_gputime; - DeviceContext::Clock::time_point_t::duration gpu_time, cpu_time; - - std::unique_ptr<Frame> frame; + Frame frame; }; std::deque<std::unique_ptr<Timing>> timings; @@ -80,14 +75,12 @@ class QueueContext final : public Context { public: void - notify_submit(const VkSubmitInfo& info, - const std::uint64_t& sequence, + notify_submit(const VkSubmitInfo& info, const std::uint64_t& sequence, const std::shared_ptr<TimestampPool::Handle> head_handle, const std::shared_ptr<TimestampPool::Handle> tail_handle); void - notify_submit(const VkSubmitInfo2& info, - const std::uint64_t& sequence, + notify_submit(const VkSubmitInfo2& info, const std::uint64_t& sequence, const std::shared_ptr<TimestampPool::Handle> head_handle, const std::shared_ptr<TimestampPool::Handle> tail_handle); |