12#ifndef SST_CORE_THREADSAFE_H
13#define SST_CORE_THREADSAFE_H
15#if ( defined(__amd64) || defined(__amd64__) || defined(__x86_64) || defined(__x86_64__) )
17#define sst_pause() _mm_pause()
18#elif (defined(__arm__) || defined(__arm) || defined(__aarch64__))
19#define sst_pause() __asm__ __volatile__("yield")
20#elif defined(__PPC64__)
21#define sst_pause() __asm__ __volatile__("or 27, 27, 27" ::: "memory");
25#include <condition_variable>
31#include "sst/core/profile.h"
39#if defined(__GNUC__) && ((__GNUC__ == 4 && __GNUC_MINOR__ < 8))
40#define CACHE_ALIGNED(type, name) type name __attribute__((aligned(64)))
41#define CACHE_ALIGNED_T
43#define CACHE_ALIGNED(type, name) alignas(64) type name
44#define CACHE_ALIGNED_T alignas(64)
50 std::atomic<bool> enabled;
51 std::atomic<size_t> count, generation;
54 Barrier(
size_t count) : origCount(count), enabled(
true), count(count), generation(0) {}
57 Barrier() : origCount(0), enabled(
false), count(0), generation(0) {}
62 count = origCount = newCount;
75 auto startTime = SST::Core::Profile::now();
77 size_t gen = generation.load(std::memory_order_acquire);
79 size_t c = count.fetch_sub(1) - 1;
82 count.store(origCount);
85 generation.fetch_add(1, std::memory_order_release);
93 if ( count < 1024 ) { sst_pause(); }
94 else if ( count < (1024 * 1024) ) {
95 std::this_thread::yield();
101 nanosleep(&ts,
nullptr);
103 }
while ( gen == generation.load(std::memory_order_acquire) );
105 elapsed = SST::Core::Profile::getElapsed(startTime);
112 enabled.store(
false);
123 std::atomic_flag latch = ATOMIC_FLAG_INIT;
130 while ( latch.test_and_set(std::memory_order_acquire) ) {
132#if defined(__PPC64__)
133 __sync_synchronize();
138 inline void unlock() { latch.clear(std::memory_order_release); }
146 inline void lock() {}
148 inline void unlock() {}
157 std::atomic<size_t> sequence;
164 CACHE_ALIGNED(std::atomic<size_t>, rPtr);
165 CACHE_ALIGNED(std::atomic<size_t>, wPtr);
169 BoundedQueue(
size_t maxSize) : initialized(
false) { initialize(maxSize); }
173 void initialize(
size_t maxSize)
175 if ( initialized )
return;
177 data =
new cell_t[dsize];
178 for (
size_t i = 0; i < maxSize; i++ )
179 data[i].sequence.store(i);
188 if ( initialized )
delete[] data;
191 size_t size()
const {
return (wPtr.load() - rPtr.load()); }
193 bool empty()
const {
return (rPtr.load() == wPtr.load()); }
195 bool try_insert(
const T& arg)
197 cell_t* cell =
nullptr;
198 size_t pos = wPtr.load(std::memory_order_relaxed);
200 cell = &data[pos % dsize];
201 size_t seq = cell->sequence.load(std::memory_order_acquire);
202 intptr_t diff = (intptr_t)seq - (intptr_t)pos;
204 if ( wPtr.compare_exchange_weak(pos, pos + 1, std::memory_order_relaxed) )
break;
206 else if ( 0 > diff ) {
211 pos = wPtr.load(std::memory_order_relaxed);
215 cell->sequence.store(pos + 1, std::memory_order_release);
219 bool try_remove(T& res)
221 cell_t* cell =
nullptr;
222 size_t pos = rPtr.load(std::memory_order_relaxed);
224 cell = &data[pos % dsize];
225 size_t seq = cell->sequence.load(std::memory_order_acquire);
226 intptr_t diff = (intptr_t)seq - (intptr_t)(pos + 1);
228 if ( rPtr.compare_exchange_weak(pos, pos + 1, std::memory_order_relaxed) )
break;
230 else if ( 0 > diff ) {
234 pos = rPtr.load(std::memory_order_relaxed);
238 cell->sequence.store(pos + dsize, std::memory_order_release);
246 if ( try_remove(res) ) {
return res; }
255 struct CACHE_ALIGNED_T Node
257 std::atomic<Node*> next;
260 Node() : next(
nullptr) {}
263 CACHE_ALIGNED(Node*, first);
264 CACHE_ALIGNED(Node*, last);
265 CACHE_ALIGNED(
Spinlock, consumerLock);
266 CACHE_ALIGNED(
Spinlock, producerLock);
272 first = last =
new Node();
277 while ( first !=
nullptr ) {
284 void insert(
const T& t)
286 Node* tmp =
new Node();
288 std::lock_guard<Spinlock> lock(producerLock);
293 bool try_remove(T& result)
295 std::lock_guard<Spinlock> lock(consumerLock);
296 Node* theFirst = first;
297 Node* theNext = first->next;
298 if ( theNext !=
nullptr ) {
299 result = theNext->data;
311 if ( try_remove(res) ) {
return res; }
Definition: threadsafe.h:48
double wait()
Wait for all threads to reach this point.
Definition: threadsafe.h:71
void resize(size_t newCount)
ONLY call this while nobody is in wait()
Definition: threadsafe.h:60
Definition: threadsafe.h:153
Definition: threadsafe.h:143
Definition: threadsafe.h:122
Definition: threadsafe.h:254