6690928: Use spinning in combination with yields for workstealing termination.
Summary: Substitute a spin loop for most calls to yield() to reduce the stress on the system.
Reviewed-by: tonyp
/*
* Copyright 2001-2008 Sun Microsystems, Inc. All Rights Reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
* CA 95054 USA or visit www.sun.com if you need additional information or
* have any questions.
*
*/
# include "incls/_precompiled.incl"
# include "incls/_taskqueue.cpp.incl"
#ifdef TRACESPINNING
uint ParallelTaskTerminator::_total_yields = 0;
uint ParallelTaskTerminator::_total_spins = 0;
uint ParallelTaskTerminator::_total_peeks = 0;
#endif
bool TaskQueueSuper::peek() {
return _bottom != _age.top();
}
int TaskQueueSetSuper::randomParkAndMiller(int *seed0) {
const int a = 16807;
const int m = 2147483647;
const int q = 127773; /* m div a */
const int r = 2836; /* m mod a */
assert(sizeof(int) == 4, "I think this relies on that");
int seed = *seed0;
int hi = seed / q;
int lo = seed % q;
int test = a * lo - r * hi;
if (test > 0)
seed = test;
else
seed = test + m;
*seed0 = seed;
return seed;
}
ParallelTaskTerminator::
ParallelTaskTerminator(int n_threads, TaskQueueSetSuper* queue_set) :
_n_threads(n_threads),
_queue_set(queue_set),
_offered_termination(0) {}
bool ParallelTaskTerminator::peek_in_queue_set() {
return _queue_set->peek();
}
void ParallelTaskTerminator::yield() {
os::yield();
}
void ParallelTaskTerminator::sleep(uint millis) {
os::sleep(Thread::current(), millis, false);
}
bool
ParallelTaskTerminator::offer_termination(TerminatorTerminator* terminator) {
Atomic::inc(&_offered_termination);
uint yield_count = 0;
// Number of hard spin loops done since last yield
uint hard_spin_count = 0;
// Number of iterations in the hard spin loop.
uint hard_spin_limit = WorkStealingHardSpins;
// If WorkStealingSpinToYieldRatio is 0, no hard spinning is done.
// If it is greater than 0, then start with a small number
// of spins and increase number with each turn at spinning until
// the count of hard spins exceeds WorkStealingSpinToYieldRatio.
// Then do a yield() call and start spinning afresh.
if (WorkStealingSpinToYieldRatio > 0) {
hard_spin_limit = WorkStealingHardSpins >> WorkStealingSpinToYieldRatio;
hard_spin_limit = MAX2(hard_spin_limit, 1U);
}
// Remember the initial spin limit.
uint hard_spin_start = hard_spin_limit;
// Loop waiting for all threads to offer termination or
// more work.
while (true) {
// Are all threads offering termination?
if (_offered_termination == _n_threads) {
return true;
} else {
// Look for more work.
// Periodically sleep() instead of yield() to give threads
// waiting on the cores the chance to grab this code
if (yield_count <= WorkStealingYieldsBeforeSleep) {
// Do a yield or hardspin. For purposes of deciding whether
// to sleep, count this as a yield.
yield_count++;
// Periodically call yield() instead spinning
// After WorkStealingSpinToYieldRatio spins, do a yield() call
// and reset the counts and starting limit.
if (hard_spin_count > WorkStealingSpinToYieldRatio) {
yield();
hard_spin_count = 0;
hard_spin_limit = hard_spin_start;
#ifdef TRACESPINNING
_total_yields++;
#endif
} else {
// Hard spin this time
// Increase the hard spinning period but only up to a limit.
hard_spin_limit = MIN2(2*hard_spin_limit,
(uint) WorkStealingHardSpins);
for (uint j = 0; j < hard_spin_limit; j++) {
SpinPause();
}
hard_spin_count++;
#ifdef TRACESPINNING
_total_spins++;
#endif
}
} else {
if (PrintGCDetails && Verbose) {
gclog_or_tty->print_cr("ParallelTaskTerminator::offer_termination() "
"thread %d sleeps after %d yields",
Thread::current(), yield_count);
}
yield_count = 0;
// A sleep will cause this processor to seek work on another processor's
// runqueue, if it has nothing else to run (as opposed to the yield
// which may only move the thread to the end of the this processor's
// runqueue).
sleep(WorkStealingSleepMillis);
}
#ifdef TRACESPINNING
_total_peeks++;
#endif
if (peek_in_queue_set() ||
(terminator != NULL && terminator->should_exit_termination())) {
Atomic::dec(&_offered_termination);
return false;
}
}
}
}
#ifdef TRACESPINNING
void ParallelTaskTerminator::print_termination_counts() {
gclog_or_tty->print_cr("ParallelTaskTerminator Total yields: %lld "
"Total spins: %lld Total peeks: %lld",
total_yields(),
total_spins(),
total_peeks());
}
#endif
void ParallelTaskTerminator::reset_for_reuse() {
if (_offered_termination != 0) {
assert(_offered_termination == _n_threads,
"Terminator may still be in use");
_offered_termination = 0;
}
}
bool RegionTaskQueueWithOverflow::is_empty() {
return (_region_queue.size() == 0) &&
(_overflow_stack->length() == 0);
}
bool RegionTaskQueueWithOverflow::stealable_is_empty() {
return _region_queue.size() == 0;
}
bool RegionTaskQueueWithOverflow::overflow_is_empty() {
return _overflow_stack->length() == 0;
}
void RegionTaskQueueWithOverflow::initialize() {
_region_queue.initialize();
assert(_overflow_stack == 0, "Creating memory leak");
_overflow_stack =
new (ResourceObj::C_HEAP) GrowableArray<RegionTask>(10, true);
}
void RegionTaskQueueWithOverflow::save(RegionTask t) {
if (TraceRegionTasksQueuing && Verbose) {
gclog_or_tty->print_cr("CTQ: save " PTR_FORMAT, t);
}
if(!_region_queue.push(t)) {
_overflow_stack->push(t);
}
}
// Note that using this method will retrieve all regions
// that have been saved but that it will always check
// the overflow stack. It may be more efficient to
// check the stealable queue and the overflow stack
// separately.
bool RegionTaskQueueWithOverflow::retrieve(RegionTask& region_task) {
bool result = retrieve_from_overflow(region_task);
if (!result) {
result = retrieve_from_stealable_queue(region_task);
}
if (TraceRegionTasksQueuing && Verbose && result) {
gclog_or_tty->print_cr(" CTQ: retrieve " PTR_FORMAT, result);
}
return result;
}
bool RegionTaskQueueWithOverflow::retrieve_from_stealable_queue(
RegionTask& region_task) {
bool result = _region_queue.pop_local(region_task);
if (TraceRegionTasksQueuing && Verbose) {
gclog_or_tty->print_cr("CTQ: retrieve_stealable " PTR_FORMAT, region_task);
}
return result;
}
bool
RegionTaskQueueWithOverflow::retrieve_from_overflow(RegionTask& region_task) {
bool result;
if (!_overflow_stack->is_empty()) {
region_task = _overflow_stack->pop();
result = true;
} else {
region_task = (RegionTask) NULL;
result = false;
}
if (TraceRegionTasksQueuing && Verbose) {
gclog_or_tty->print_cr("CTQ: retrieve_stealable " PTR_FORMAT, region_task);
}
return result;
}