wangtao
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

OpenMP RTL cleanup: two PAUSEs per spin loop iteration replaced with single one 

Differential Revision: https://reviews.llvm.org/D35490

llvm-svn: 308423
This commit is contained in:
Andrey Churbanov 2017-07-19 09:26:13 +00:00
parent b05a55787a
commit c7476ed0be
2 changed files with 13 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
openmp/runtime/src/kmp.h
View File

@ -1040,7 +1040,11 @@ extern void __kmp_x86_cpuid(int mode, int mode2, struct kmp_cpuid *p);
#if KMP_ARCH_X86
extern void __kmp_x86_pause(void);
#elif KMP_MIC
static void __kmp_x86_pause(void) { _mm_delay_32(100); }
// Performance testing on KNC (C0QS-7120 P/A/X/D, 61-core, 16 GB Memory) showed
// regression after removal of extra PAUSE from KMP_YIELD_SPIN(). Changing
// the delay from 100 to 300 showed even better performance than double PAUSE
// on Spec OMP2001 and LCPC tasking tests, no regressions on EPCC.
static void __kmp_x86_pause(void) { _mm_delay_32(300); }
#else
static void __kmp_x86_pause(void) { _mm_pause(); }
#endif
@ -1076,7 +1080,7 @@ static void __kmp_x86_pause(void) { _mm_pause(); }
KMP_CPU_PAUSE(); \
(count) -= 2; \
if (!(count)) { \
KMP_YIELD(cond); \
__kmp_yield(cond); \
(count) = __kmp_yield_next; \
} \
}
@ -1085,7 +1089,7 @@ static void __kmp_x86_pause(void) { _mm_pause(); }
KMP_CPU_PAUSE(); \
(count) -= 2; \
if (!(count)) { \
KMP_YIELD(1); \
__kmp_yield(1); \
(count) = __kmp_yield_next; \
} \
}

9
openmp/runtime/src/kmp_wait_release.h
View File

@ -47,7 +47,7 @@ enum flag_type {
*/
template <typename P> class kmp_flag {
volatile P
*loc; /**< Pointer to the flag storage that is modified by another thread
*loc; /**< Pointer to the flag storage that is modified by another thread
*/
flag_type t; /**< "Type" of the flag in loc */
public:
@ -225,11 +225,14 @@ __kmp_wait_template(kmp_info_t *this_thr, C *flag,
// If we are oversubscribed, or have waited a bit (and
// KMP_LIBRARY=throughput), then yield
KMP_YIELD(oversubscribed);
// TODO: Should it be number of cores instead of thread contexts? Like:
// KMP_YIELD(TCR_4(__kmp_nth) > __kmp_ncores);
// Need performance improvement data to make the change...
KMP_YIELD_SPIN(spins);
if (oversubscribed) {
KMP_YIELD(1);
} else {
KMP_YIELD_SPIN(spins);
}
// Check if this thread was transferred from a team
// to the thread pool (or vice-versa) while spinning.
in_pool = !!TCR_4(this_thr->th.th_in_pool);