4c3b11445c
Phase 0 of design/pyrowave-codec-plan.md — the opt-in wired-LAN ultra-low- latency codec. Vendored at upstream 509e4f88 (API 0.4.0, Granite 44362775, volk + vulkan-headers pins in PUNKTFUNK-VENDOR.txt), pruned to the 6.6 MB the standalone no-renderer build needs; scripts/vendor-pyrowave.sh reproduces the tree (a pin bump is protocol-affecting, plan §4.2). build.rs drives the wrapper CMakeLists (static archives incl. a static C-API lib upstream only ships shared) + bindgen over pyrowave.h; Linux and Windows only, empty stub elsewhere (Apple gets a native Metal port, §4.7). Offline-safe by construction: no network, no system lib, vendored Vulkan headers — same model as the opus dep (flatpak builder has no network). Phase-0 validation on .21 (RTX 5070 Ti, driver 610.43.03): - upstream pyrowave-c-test + interop test (incl. dmabuf/DRM-modifier Vulkan<->Vulkan) pass, from the pristine AND the pruned tree - GPU kernel times at ~1.6 bpp noise: encode/decode 0.090/0.042 ms @800p, 0.146/0.067 @1080p, 0.226/0.103 @1440p, 0.477/0.201 @4K — order of magnitude under NVENC's 1-2 ms retrieve, CBR lands within ~100 B of target - cargo test -p pyrowave-sys green (static link + API-version pin check) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
447 lines
10 KiB
C++
447 lines
10 KiB
C++
/* Copyright (c) 2017-2026 Hans-Kristian Arntzen
|
|
*
|
|
* Permission is hereby granted, free of charge, to any person obtaining
|
|
* a copy of this software and associated documentation files (the
|
|
* "Software"), to deal in the Software without restriction, including
|
|
* without limitation the rights to use, copy, modify, merge, publish,
|
|
* distribute, sublicense, and/or sell copies of the Software, and to
|
|
* permit persons to whom the Software is furnished to do so, subject to
|
|
* the following conditions:
|
|
*
|
|
* The above copyright notice and this permission notice shall be
|
|
* included in all copies or substantial portions of the Software.
|
|
*
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
|
|
* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
|
|
* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
|
|
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
|
|
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
|
*/
|
|
|
|
#include "thread_group.hpp"
|
|
#include <assert.h>
|
|
#include <stdexcept>
|
|
#include <type_traits>
|
|
#include "logging.hpp"
|
|
#include "thread_id.hpp"
|
|
#include "thread_priority.hpp"
|
|
#include "string_helpers.hpp"
|
|
#include "timeline_trace_file.hpp"
|
|
#include "thread_name.hpp"
|
|
#include "environment.hpp"
|
|
|
|
namespace Granite
|
|
{
|
|
namespace Internal
|
|
{
|
|
void TaskDeps::notify_dependees()
|
|
{
|
|
if (signal)
|
|
signal->signal_increment();
|
|
|
|
for (auto &dep : pending)
|
|
dep->dependency_satisfied();
|
|
pending.clear();
|
|
|
|
{
|
|
std::lock_guard<std::mutex> holder{cond_lock};
|
|
done = true;
|
|
cond.notify_all();
|
|
}
|
|
}
|
|
|
|
void TaskDeps::task_completed()
|
|
{
|
|
auto old_tasks = count.fetch_sub(1, std::memory_order_acq_rel);
|
|
assert(old_tasks > 0);
|
|
if (old_tasks == 1)
|
|
notify_dependees();
|
|
}
|
|
|
|
void TaskDeps::dependency_satisfied()
|
|
{
|
|
auto old_deps = dependency_count.fetch_sub(1, std::memory_order_acq_rel);
|
|
assert(old_deps > 0);
|
|
|
|
if (old_deps == 1)
|
|
{
|
|
if (pending_tasks.empty())
|
|
notify_dependees();
|
|
else
|
|
{
|
|
group->move_to_ready_tasks(pending_tasks);
|
|
pending_tasks.clear();
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
TaskGroup::TaskGroup(ThreadGroup *group_)
|
|
: group(group_)
|
|
{
|
|
}
|
|
|
|
void TaskGroup::flush()
|
|
{
|
|
if (flushed)
|
|
throw std::logic_error("Cannot flush more than once.");
|
|
|
|
flushed = true;
|
|
deps->dependency_satisfied();
|
|
}
|
|
|
|
void TaskGroup::wait()
|
|
{
|
|
if (!flushed)
|
|
flush();
|
|
|
|
std::unique_lock<std::mutex> holder{deps->cond_lock};
|
|
deps->cond.wait(holder, [this]() {
|
|
return deps->done;
|
|
});
|
|
}
|
|
|
|
bool TaskGroup::poll()
|
|
{
|
|
if (!flushed)
|
|
flush();
|
|
return deps->count.load(std::memory_order_acquire) == 0;
|
|
}
|
|
|
|
TaskGroup::~TaskGroup()
|
|
{
|
|
if (!flushed)
|
|
flush();
|
|
}
|
|
|
|
void ThreadGroup::set_async_main_thread()
|
|
{
|
|
Util::set_current_thread_name("MainAsyncThread");
|
|
Util::TimelineTraceFile::set_tid("main-async");
|
|
// Seems reasonable to make sure main thread is making forward progress when it has something useful to do.
|
|
Util::set_current_thread_priority(Util::ThreadPriority::High);
|
|
}
|
|
|
|
static void set_main_thread_name()
|
|
{
|
|
Util::set_current_thread_name("MainThread");
|
|
Util::TimelineTraceFile::set_tid("main");
|
|
// Seems reasonable to make sure main thread is making forward progress when it has something useful to do.
|
|
Util::set_current_thread_priority(Util::ThreadPriority::High);
|
|
}
|
|
|
|
static void set_worker_thread_name_and_prio(unsigned index, TaskClass task_class)
|
|
{
|
|
auto name = Util::join(task_class == TaskClass::Foreground ? "FG-" : "BG-", index);
|
|
Util::set_current_thread_name(name.c_str());
|
|
Util::TimelineTraceFile::set_tid(name.c_str());
|
|
Util::set_current_thread_priority(task_class == TaskClass::Foreground ?
|
|
Util::ThreadPriority::Default : Util::ThreadPriority::Low);
|
|
}
|
|
|
|
void ThreadGroup::refresh_global_timeline_trace_file()
|
|
{
|
|
Util::TimelineTraceFile::set_per_thread(timeline_trace_file.get());
|
|
}
|
|
|
|
void ThreadGroup::set_thread_context()
|
|
{
|
|
refresh_global_timeline_trace_file();
|
|
}
|
|
|
|
Util::TimelineTraceFile *ThreadGroup::get_timeline_trace_file()
|
|
{
|
|
return timeline_trace_file.get();
|
|
}
|
|
|
|
void ThreadGroup::start(unsigned num_threads_foreground,
|
|
unsigned num_threads_background,
|
|
const std::function<void ()> &on_thread_begin)
|
|
{
|
|
if (active)
|
|
throw std::logic_error("Cannot start a thread group which has already started.");
|
|
|
|
dead = false;
|
|
active = true;
|
|
|
|
fg.thread_group.resize(num_threads_foreground);
|
|
bg.thread_group.resize(num_threads_background);
|
|
|
|
#ifndef GRANITE_SHIPPING
|
|
std::string path;
|
|
if (Util::get_environment("GRANITE_TIMELINE_TRACE", path))
|
|
{
|
|
LOGI("Enabling JSON timeline tracing to %s.\n", path.c_str());
|
|
timeline_trace_file = std::make_unique<Util::TimelineTraceFile>(path);
|
|
}
|
|
#endif
|
|
|
|
refresh_global_timeline_trace_file();
|
|
set_main_thread_name();
|
|
|
|
unsigned self_index = 1;
|
|
for (auto &t : fg.thread_group)
|
|
{
|
|
t = std::make_unique<std::thread>([this, on_thread_begin, self_index]() {
|
|
refresh_global_timeline_trace_file();
|
|
set_worker_thread_name_and_prio(self_index - 1, TaskClass::Foreground);
|
|
if (on_thread_begin)
|
|
on_thread_begin();
|
|
thread_looper(self_index, TaskClass::Foreground);
|
|
});
|
|
self_index++;
|
|
}
|
|
|
|
for (auto &t : bg.thread_group)
|
|
{
|
|
t = std::make_unique<std::thread>([this, on_thread_begin, self_index]() {
|
|
refresh_global_timeline_trace_file();
|
|
set_worker_thread_name_and_prio(self_index - 1, TaskClass::Background);
|
|
if (on_thread_begin)
|
|
on_thread_begin();
|
|
thread_looper(self_index, TaskClass::Background);
|
|
});
|
|
self_index++;
|
|
}
|
|
}
|
|
|
|
void ThreadGroup::submit(TaskGroupHandle &group)
|
|
{
|
|
group->flush();
|
|
group.reset();
|
|
}
|
|
|
|
void ThreadGroup::add_dependency(TaskGroup &dependee, TaskGroup &dependency)
|
|
{
|
|
if (dependency.flushed)
|
|
throw std::logic_error("Cannot wait for task group which has been flushed.");
|
|
if (dependee.flushed)
|
|
throw std::logic_error("Cannot add dependency to task group which has been flushed.");
|
|
|
|
dependency.deps->pending.push_back(dependee.deps);
|
|
dependee.deps->dependency_count.fetch_add(1, std::memory_order_relaxed);
|
|
}
|
|
|
|
void ThreadGroup::move_to_ready_tasks(const Util::SmallVector<Internal::Task *> &list)
|
|
{
|
|
unsigned fg_task_count = 0;
|
|
unsigned bg_task_count = 0;
|
|
for (auto *t : list)
|
|
{
|
|
if (t->deps->task_class == TaskClass::Foreground)
|
|
fg_task_count++;
|
|
else
|
|
bg_task_count++;
|
|
}
|
|
|
|
total_tasks.fetch_add(list.size(), std::memory_order_relaxed);
|
|
|
|
if (fg_task_count)
|
|
{
|
|
std::lock_guard<std::mutex> holder{fg.cond_lock};
|
|
|
|
for (auto &t : list)
|
|
fg.ready_tasks.push(t);
|
|
|
|
if (fg_task_count >= fg.thread_group.size())
|
|
fg.cond.notify_all();
|
|
else
|
|
{
|
|
for (unsigned i = 0; i < fg_task_count; i++)
|
|
fg.cond.notify_one();
|
|
}
|
|
}
|
|
|
|
if (bg_task_count)
|
|
{
|
|
std::lock_guard<std::mutex> holder{bg.cond_lock};
|
|
|
|
for (auto &t : list)
|
|
bg.ready_tasks.push(t);
|
|
|
|
if (bg_task_count >= bg.thread_group.size())
|
|
bg.cond.notify_all();
|
|
else
|
|
{
|
|
for (unsigned i = 0; i < bg_task_count; i++)
|
|
bg.cond.notify_one();
|
|
}
|
|
}
|
|
}
|
|
|
|
void Internal::TaskGroupDeleter::operator()(TaskGroup *group)
|
|
{
|
|
group->group->free_task_group(group);
|
|
}
|
|
|
|
void Internal::TaskDepsDeleter::operator()(Internal::TaskDeps *deps)
|
|
{
|
|
deps->group->free_task_deps(deps);
|
|
}
|
|
|
|
void ThreadGroup::free_task_group(TaskGroup *group)
|
|
{
|
|
task_group_pool.free(group);
|
|
}
|
|
|
|
void ThreadGroup::free_task_deps(Internal::TaskDeps *deps)
|
|
{
|
|
task_deps_pool.free(deps);
|
|
}
|
|
|
|
void TaskSignal::signal_increment()
|
|
{
|
|
std::lock_guard<std::mutex> holder{lock};
|
|
counter++;
|
|
cond.notify_all();
|
|
}
|
|
|
|
void TaskSignal::wait_until_at_least(uint64_t count)
|
|
{
|
|
std::unique_lock<std::mutex> holder{lock};
|
|
cond.wait(holder, [&]() -> bool {
|
|
return counter >= count;
|
|
});
|
|
}
|
|
|
|
uint64_t TaskSignal::get_count()
|
|
{
|
|
std::lock_guard<std::mutex> holder{lock};
|
|
return counter;
|
|
}
|
|
|
|
TaskGroupHandle ThreadGroup::create_task()
|
|
{
|
|
TaskGroupHandle group(task_group_pool.allocate(this));
|
|
group->deps = Internal::TaskDepsHandle(task_deps_pool.allocate(this));
|
|
group->deps->count.store(0, std::memory_order_relaxed);
|
|
return group;
|
|
}
|
|
|
|
void TaskGroup::set_fence_counter_signal(TaskSignal *signal)
|
|
{
|
|
deps->signal = signal;
|
|
}
|
|
|
|
ThreadGroup *TaskGroup::get_thread_group() const
|
|
{
|
|
return group;
|
|
}
|
|
|
|
void TaskGroup::set_desc(const char *desc)
|
|
{
|
|
snprintf(deps->desc, sizeof(deps->desc), "%s", desc);
|
|
}
|
|
|
|
void TaskGroup::set_task_class(TaskClass task_class)
|
|
{
|
|
deps->task_class = task_class;
|
|
}
|
|
|
|
void ThreadGroup::wait_idle()
|
|
{
|
|
std::unique_lock<std::mutex> holder{wait_cond_lock};
|
|
wait_cond.wait(holder, [&]() {
|
|
return total_tasks.load(std::memory_order_relaxed) == completed_tasks.load(std::memory_order_relaxed);
|
|
});
|
|
}
|
|
|
|
bool ThreadGroup::is_idle()
|
|
{
|
|
return total_tasks.load(std::memory_order_acquire) == completed_tasks.load(std::memory_order_acquire);
|
|
}
|
|
|
|
void ThreadGroup::thread_looper(unsigned index, TaskClass task_class)
|
|
{
|
|
Util::register_thread_index(index);
|
|
auto &ctx = task_class == TaskClass::Foreground ? fg : bg;
|
|
|
|
for (;;)
|
|
{
|
|
Internal::Task *task = nullptr;
|
|
|
|
{
|
|
std::unique_lock<std::mutex> holder{ctx.cond_lock};
|
|
ctx.cond.wait(holder, [&]() {
|
|
return dead || !ctx.ready_tasks.empty();
|
|
});
|
|
|
|
if (dead && ctx.ready_tasks.empty())
|
|
break;
|
|
|
|
task = ctx.ready_tasks.front();
|
|
ctx.ready_tasks.pop();
|
|
}
|
|
|
|
if (task->callable)
|
|
{
|
|
GRANITE_SCOPED_TIMELINE_EVENT_FILE(timeline_trace_file.get(), task->deps->desc);
|
|
task->callable.call();
|
|
}
|
|
|
|
task->deps->task_completed();
|
|
task_pool.free(task);
|
|
|
|
{
|
|
auto completed = completed_tasks.fetch_add(1, std::memory_order_relaxed) + 1;
|
|
//LOGI("Task completed (%u / %u)!\n", completed, total_tasks.load(memory_order_relaxed));
|
|
|
|
if (completed == total_tasks.load(std::memory_order_relaxed))
|
|
{
|
|
std::lock_guard<std::mutex> holder{wait_cond_lock};
|
|
wait_cond.notify_all();
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
ThreadGroup::ThreadGroup()
|
|
{
|
|
total_tasks.store(0);
|
|
completed_tasks.store(0);
|
|
}
|
|
|
|
ThreadGroup::~ThreadGroup()
|
|
{
|
|
stop();
|
|
}
|
|
|
|
void ThreadGroup::stop()
|
|
{
|
|
if (!active)
|
|
return;
|
|
|
|
wait_idle();
|
|
|
|
{
|
|
std::lock_guard<std::mutex> holder_fg{fg.cond_lock};
|
|
std::lock_guard<std::mutex> holder_bg{bg.cond_lock};
|
|
dead = true;
|
|
fg.cond.notify_all();
|
|
bg.cond.notify_all();
|
|
}
|
|
|
|
for (auto &t : fg.thread_group)
|
|
{
|
|
if (t && t->joinable())
|
|
{
|
|
t->join();
|
|
t.reset();
|
|
}
|
|
}
|
|
|
|
for (auto &t : bg.thread_group)
|
|
{
|
|
if (t && t->joinable())
|
|
{
|
|
t->join();
|
|
t.reset();
|
|
}
|
|
}
|
|
|
|
active = false;
|
|
dead = false;
|
|
}
|
|
} |