406654be7a
SkTaskGroup is like SkThreadPool except the threads stay in one global pool. Each SkTaskGroup itself is tiny (4 bytes) and its wait() method applies only to tasks add()ed to that instance, not the whole thread pool. This means we don't need to bring up new thread pools when tests themselves want to use multithreading (e.g. pathops, quilt). We just create a new SkTaskGroup and wait for that to complete. This should be more efficient, and allow us to expand where we use threads to really latency sensitive places. E.g. we can probably now use these in nanobench for CPU .skp rendering. Now that all threads are sharing the same pool, I think we can remove most of the custom mechanism pathops tests use to control threading. They'll just ride on the global pool with all other tests now. This (temporarily?) removes the GPU multithreading feature from DM, which we don't use. On my desktop, DM runs a little faster (57s -> 55s) in Debug, and a lot faster in Release (36s -> 24s). The bots show speedups of similar proportions, cutting more than a minute off the N4/Release and Win7/Debug runtimes. BUG=skia: Committed: https://skia.googlesource.com/skia/+/9c7207b5dc71dc5a96a2eb107d401133333d5b6f R=caryclark@google.com, bsalomon@google.com, bungeman@google.com, mtklein@google.com, reed@google.com Author: mtklein@chromium.org Review URL: https://codereview.chromium.org/531653002
88 lines
2.3 KiB
C++
88 lines
2.3 KiB
C++
#include "DMTask.h"
|
|
#include "DMTaskRunner.h"
|
|
#include "SkCommonFlags.h"
|
|
|
|
namespace DM {
|
|
|
|
Task::Task(Reporter* reporter, TaskRunner* taskRunner)
|
|
: fReporter(reporter)
|
|
, fTaskRunner(taskRunner)
|
|
, fDepth(0) {
|
|
fReporter->taskCreated();
|
|
}
|
|
|
|
Task::Task(const Task& parent)
|
|
: fReporter(parent.fReporter)
|
|
, fTaskRunner(parent.fTaskRunner)
|
|
, fDepth(parent.depth() + 1) {
|
|
fReporter->taskCreated();
|
|
}
|
|
|
|
Task::~Task() {
|
|
fReporter->taskDestroyed();
|
|
}
|
|
|
|
void Task::fail(const char* msg) {
|
|
SkString failure(this->name());
|
|
if (msg) {
|
|
failure.appendf(": %s", msg);
|
|
}
|
|
fReporter->fail(failure);
|
|
}
|
|
|
|
void Task::start() {
|
|
fStart = SkTime::GetMSecs();
|
|
}
|
|
|
|
void Task::finish() {
|
|
fReporter->printStatus(this->name(), SkTime::GetMSecs() - fStart);
|
|
}
|
|
|
|
void Task::reallySpawnChild(CpuTask* task) {
|
|
fTaskRunner->add(task);
|
|
}
|
|
|
|
CpuTask::CpuTask(Reporter* reporter, TaskRunner* taskRunner) : Task(reporter, taskRunner) {}
|
|
CpuTask::CpuTask(const Task& parent) : Task(parent) {}
|
|
|
|
void CpuTask::run() {
|
|
if (FLAGS_cpu && !this->shouldSkip()) {
|
|
this->start();
|
|
if (!FLAGS_dryRun) this->draw();
|
|
this->finish();
|
|
}
|
|
SkDELETE(this);
|
|
}
|
|
|
|
void CpuTask::spawnChild(CpuTask* task) {
|
|
// Run children serially on this (CPU) thread. This tends to save RAM and is usually no slower.
|
|
// Calling reallySpawnChild() is nearly equivalent, but it'd pointlessly contend on the
|
|
// threadpool; reallySpawnChild() is most useful when you want to change threadpools.
|
|
task->run();
|
|
}
|
|
|
|
GpuTask::GpuTask(Reporter* reporter, TaskRunner* taskRunner) : Task(reporter, taskRunner) {}
|
|
|
|
void GpuTask::run(GrContextFactory* factory) {
|
|
if (FLAGS_gpu && !this->shouldSkip()) {
|
|
this->start();
|
|
if (!FLAGS_dryRun) this->draw(factory);
|
|
this->finish();
|
|
if (FLAGS_abandonGpuContext) {
|
|
factory->abandonContexts();
|
|
}
|
|
if (FLAGS_resetGpuContext || FLAGS_abandonGpuContext) {
|
|
factory->destroyContexts();
|
|
}
|
|
}
|
|
SkDELETE(this);
|
|
}
|
|
|
|
void GpuTask::spawnChild(CpuTask* task) {
|
|
// Spawn a new task so it runs on the CPU threadpool instead of the GPU one we're on now.
|
|
// It goes on the front of the queue to minimize the time we must hold reference bitmaps in RAM.
|
|
this->reallySpawnChild(task);
|
|
}
|
|
|
|
} // namespace DM
|