Parallel runtime thread pinning.
|
View:
New views
5 Messages
—
Rating Filter:
Alert me
|
 |
Parallel runtime thread pinning.
by Paul Bone
::
Rate this Message:
For post-commit review. A later changeset will address Julien's review comments on my other recent changes to the parallel runtime. Parallel runtime thread pinning. This change introduces two new features in the mercury runtime; pinning of threads to CPU cores/threads and runtime detection of the number of CPU cores/threads available. If MR_num_threads has not been specified in the runtime options with the -P flag we use the sysconf(_SC_NPROCESSORS_ONLN) call if available to detect the number of CPUs online and set MR_num_threads available. As before this defaults to 1. Thread pinning is enabled if the runtime was able to detect the number of CPUs on the machine or the user specifically requests thread pinning with the --thread-pinning runtime option. The sched_setaffinity() call is used to pin each thread to a specific CPU. I believe that in some cases thread pinning can achieve better performance, this is yet to be determined and it may depend on the machine's architecture. It does make profiling of the runtime system more reliable where the RDTSCP instruction is not available. It ensuring that a thread is not migrated to a different CPU between sampling of the CPU's TSC. configure.in: runtime/mercury_conf.h.in: Detect the presence of sched.h sysconf() sched_setaffinity() and _SC_NPROCESSORS_ONLN. doc/user_guide.texi: Document the new --thread-pinning runtime option. Adjust the documentation of -P to reflect the new behaviour. runtime/mercury_context.c: Add the MR_pin_thread() function. Create a new global MR_bool MR_pin_threads; Add the calculation of the number of threads to use to MR_init_thread_stuff() Correct a bug in a format string in my previous patch. runtime/mercury_context.h: Export the new MR_pin_thread() function. Export the new MR_pin_threads global. Correct a previous spelling mistake. Adjust the documentation of MR_init_thread_stuff to reflect the new behaviour. runtime/mercury_wrapper.c: Pin the primordial thread to a CPU after it spawns the other threads. Add the --thread-pinning runtime configuration option. Move the calculation of MR_max_outstanding_contexts until after MR_init_thread_stuff() so that it is calculated after the number of CPUs available has been determined. Add a pause instruction to a spinloop for better behaviour on later i386/x86_64 processors. See the documentation for MR_ATOMIC_PAUSE. runtime/mercury_thread.c: After a thread is spawned call MR_pin_thread() to pin a thread to a CPU if the thread has been created to pickup work from the global work queue. Index: configure.in =================================================================== RCS file: /home/mercury1/repository/mercury/configure.in,v retrieving revision 1.541 diff -u -p -b -r1.541 configure.in --- configure.in 18 Aug 2009 05:10:39 -0000 1.541 +++ configure.in 19 Aug 2009 06:43:37 -0000 @@ -1107,7 +1107,7 @@ mercury_check_for_functions \ getpid setpgid fork execlp wait kill \ grantpt unlockpt ptsname tcgetattr tcsetattr ioctl \ access sleep opendir readdir closedir mkdir symlink readlink \ - gettimeofday setenv putenv _putenv posix_spawn + gettimeofday setenv putenv _putenv posix_spawn sched_setaffinity #-----------------------------------------------------------------------------# @@ -1116,7 +1116,8 @@ MERCURY_CHECK_FOR_HEADERS( \ asm/sigcontext.h sys/param.h sys/time.h sys/times.h \ sys/types.h sys/stat.h fcntl.h termios.h sys/ioctl.h \ sys/stropts.h windows.h dirent.h getopt.h malloc.h \ - semaphore.h pthread.h time.h spawn.h fenv.h sys/mman.h sys/sem.h) + semaphore.h pthread.h time.h spawn.h fenv.h sys/mman.h sys/sem.h \ + sched.h) if test "$MR_HAVE_GETOPT_H" = 1; then GETOPT_H_AVAILABLE=yes @@ -1138,6 +1139,24 @@ MERCURY_CHECK_FOR_FENV_FUNC([fesetround] #-----------------------------------------------------------------------------# # +# Check for declarations. +# + +mercury_check_for_declarations () { + for mercury_cv_decl in "$@" + do + mercury_cv_decl_define="MR_HAVE_`echo $mercury_cv_decl | \ + tr abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ`" + AC_CHECK_DECL($mercury_cv_decl, [ + AC_DEFINE_UNQUOTED($mercury_cv_decl_define) + ]) + done +} + +mercury_check_for_declarations "_SC_NPROCESSORS_ONLN" + +#-----------------------------------------------------------------------------# +# # Check the basics of sigaction # Index: doc/user_guide.texi =================================================================== RCS file: /home/mercury1/repository/mercury/doc/user_guide.texi,v retrieving revision 1.590 diff -u -p -b -r1.590 user_guide.texi --- doc/user_guide.texi 19 Aug 2009 07:45:11 -0000 1.590 +++ doc/user_guide.texi 23 Aug 2009 22:27:28 -0000 @@ -9587,7 +9587,10 @@ This only has an effect if the executabl @item -P @var{num} @findex -P (runtime option) Tells the runtime system to use @var{num} threads -if the program was built in a parallel grade. +if the program was built in a parallel low-level C grade. +The mercury runtime attempts to automatically determine this value if support +is available from the operating system. +If it cannot or support is unavailable it defaults to @samp{1}. @sp 1 @item --max-contexts-per-thread @var{num} @@ -9608,6 +9611,17 @@ grade. @c and the @samp{MR_PROFILE_PARALLEL_EXECUTION_SUPPORT} C macro was @c defined when the runtime system was compiled. +@sp 1 +@item --thread-pinning +@findex --thread-pinning +Request that the runtime system attempts to pin mercury engines (POSIX threads) +to CPU cores/hardware threads. +This only has an effect if the executable was build in a parallel low-level C +grade. +It is disabled by default unless @samp{-P @var{num}} is not specified and the +runtime system is able to detect the number of processors enabled by the +operating system. + @c @item -r @var{num} @c @findex -r (runtime option) @c Repeats execution of the entry point procedure @var{num} times, Index: runtime/mercury_atomic_ops.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_atomic_ops.c,v retrieving revision 1.4 diff -u -p -b -r1.4 mercury_atomic_ops.c --- runtime/mercury_atomic_ops.c 16 Aug 2009 10:18:36 -0000 1.4 +++ runtime/mercury_atomic_ops.c 21 Aug 2009 02:14:29 -0000 @@ -71,18 +71,18 @@ MR_OUTLINE_DEFN( #if defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) static MR_bool MR_rdtscp_is_available = MR_FALSE; static MR_bool MR_rdtsc_is_available = MR_FALSE; #endif -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) /* Set this to 1 to enable some printfs below */ #define MR_DEBUG_CPU_FEATURE_DETECTION 0 /* -** cpuid, rdtscp and rdtsc are i386/amd64 instructions. +** cpuid, rdtscp and rdtsc are i386/x86_64 instructions. */ static __inline__ void MR_cpuid(MR_Unsigned code, MR_Unsigned sub_code, @@ -98,14 +98,14 @@ MR_rdtsc(MR_uint_least64_t *tsc); extern void MR_configure_profiling_timers(void) { -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) MR_Unsigned a, b, c, d; MR_Unsigned eflags, old_eflags; /* ** Check for the CPUID instruction. CPUID is supported if we can flip bit ** 21 in the CPU's EFLAGS register. The assembly below is written in a - ** subset of i386 and amd64 assembly. To read and write EFLAGS we have + ** subset of i386 and x86_64 assembly. To read and write EFLAGS we have ** to go via the C stack. */ __asm__ ("pushf; pop %0" @@ -256,7 +256,7 @@ MR_configure_profiling_timers(void) { extern void MR_profiling_start_timer(MR_Timer *timer) { -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) /* ** If we don't have enough data to fill in all the fields of this structure ** we leave them alone, we won't check them later without checking @@ -275,7 +275,7 @@ MR_profiling_start_timer(MR_Timer *timer extern void MR_profiling_stop_timer(MR_Timer *timer, MR_Stats *stats) { -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) MR_Timer now; MR_int_least64_t duration; MR_uint_least64_t duration_squared; @@ -310,9 +310,9 @@ MR_profiling_stop_timer(MR_Timer *timer, } /* -** It's convenient that this instruction is the same on both i386 and amd64 +** It's convenient that this instruction is the same on both i386 and x86_64 */ -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) static __inline__ void MR_cpuid(MR_Unsigned code, MR_Unsigned sub_code, Index: runtime/mercury_atomic_ops.h =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_atomic_ops.h,v retrieving revision 1.5 diff -u -p -b -r1.5 mercury_atomic_ops.h --- runtime/mercury_atomic_ops.h 16 Aug 2009 10:47:55 -0000 1.5 +++ runtime/mercury_atomic_ops.h 21 Aug 2009 02:14:52 -0000 @@ -16,16 +16,6 @@ #include "mercury_std.h" -/* -** AMD say that __amd64__ is defined by the compiler for 64bit platforms, -** Intel say that __x86_64__ is the correct macro. Really these refer to -** the same thing that is simply branded differently, we use __amd64__ below -** and define it if necessary ourselves. -*/ -#if defined(__x86_64__) && !defined(__amd64__) -#define __amd64__ -#endif - /*---------------------------------------------------------------------------*/ #if defined(MR_LL_PARALLEL_CONJ) @@ -47,7 +37,7 @@ MR_compare_and_swap_word(volatile MR_Int return __sync_bool_compare_and_swap(addr, old, new_val); \ } while (0) -#elif defined(__GNUC__) && defined(__amd64__) +#elif defined(__GNUC__) && defined(__x86_64__) #define MR_COMPARE_AND_SWAP_WORD_BODY \ do { \ @@ -95,7 +85,7 @@ MR_compare_and_swap_word(volatile MR_Int MR_EXTERN_INLINE void MR_atomic_inc_int(volatile MR_Integer *addr); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_INC_INT_BODY \ do { \ @@ -148,7 +138,7 @@ MR_atomic_inc_int(volatile MR_Integer *a MR_EXTERN_INLINE void MR_atomic_dec_int(volatile MR_Integer *addr); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_DEC_INT_BODY \ do { \ @@ -191,7 +181,7 @@ MR_atomic_dec_int(volatile MR_Integer *a MR_EXTERN_INLINE void MR_atomic_add_int(volatile MR_Integer *addr, MR_Integer addend); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_ADD_INT_BODY \ do { \ @@ -233,7 +223,7 @@ MR_atomic_add_int(volatile MR_Integer *a MR_EXTERN_INLINE void MR_atomic_sub_int(volatile MR_Integer *addr, MR_Integer x); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_SUB_INT_BODY \ do { \ @@ -294,7 +284,7 @@ MR_atomic_sub_int(volatile MR_Integer *a * References: Intel and AMD documentation for PAUSE, Intel optimisation * guide. */ -#if defined(__GNUC__) && ( defined(__i386__) || defined(__amd64__) ) +#if defined(__GNUC__) && ( defined(__i386__) || defined(__x86_64__) ) #define MR_ATOMIC_PAUSE \ do { \ @@ -344,7 +334,7 @@ typedef struct { } MR_Timer; /* -** Configure the profiling stats code. On i386 and amd64 machines this uses +** Configure the profiling stats code. On i386 and x86_64 machines this uses ** CPUID to determine if the RDTSCP instruction is available and not prohibited ** by the OS. */ Index: runtime/mercury_conf.h.in =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_conf.h.in,v retrieving revision 1.64 diff -u -p -b -r1.64 mercury_conf.h.in --- runtime/mercury_conf.h.in 30 Jul 2009 04:02:56 -0000 1.64 +++ runtime/mercury_conf.h.in 19 Aug 2009 07:13:41 -0000 @@ -136,6 +136,7 @@ ** MR_HAVE_FENV_H we have <fenv.h> ** MR_HAVE_SYS_MMAN_H we have <sys/mman.h> ** MR_HAVE_SYS_SEM_H we have <sys/sem.h> +** MR_HAVE_SCHED_H we have <sched.h> */ #undef MR_HAVE_SYS_SIGINFO_H #undef MR_HAVE_SYS_SIGNAL_H @@ -165,6 +166,7 @@ #undef MR_HAVE_FENV_H #undef MR_HAVE_SYS_MMAN_H #undef MR_HAVE_SYS_SEM_H +#undef MR_HAVE_SCHED_H /* ** MR_HAVE_POSIX_TIMES is defined if we have the POSIX @@ -263,6 +265,7 @@ ** MR_HAVE__PUTENV we have the _putenv() function. ** MR_HAVE_POSIX_SPAWN we have the posix_spawn() function. ** MR_HAVE_FESETROUND we have the fesetround() function. +** MR_HAVE_SCHED_SETAFFINITY if we have the sched_setaffinity() function. */ #undef MR_HAVE_GETPID #undef MR_HAVE_SETPGID @@ -324,6 +327,7 @@ #undef MR_HAVE__PUTENV #undef MR_HAVE_POSIX_SPAWN #undef MR_HAVE_FESETROUND +#undef MR_HAVE_SCHED_SETAFFINITY /* ** We use mprotect() and signals to catch stack and heap overflows. @@ -358,6 +362,14 @@ #undef MR_HAVE_SIGCONTEXT_STRUCT_2ARG /* +** These specify weather the given C macros are defined. +** +** MR_HAVE__SC_NPROCESSORS_ONLN, This is defined as a parameter for sysconf to +** determine the number of processors online. +*/ +#undef MR_HAVE__SC_NPROCESSORS_ONLN + +/* ** For debugging purposes, if we get a fatal signal, we print out the ** program counter (PC) at which the signal occurred. ** Index: runtime/mercury_context.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_context.c,v retrieving revision 1.67 diff -u -p -b -r1.67 mercury_context.c --- runtime/mercury_context.c 17 Aug 2009 08:12:17 -0000 1.67 +++ runtime/mercury_context.c 20 Aug 2009 02:31:36 -0000 @@ -31,6 +31,10 @@ ENDINIT #include <math.h> /* for sqrt and pow */ #endif +#ifdef MR_HAVE_SCHED_H +#include <sched.h> +#endif + #include "mercury_memory_handlers.h" #include "mercury_context.h" #include "mercury_engine.h" /* for `MR_memdebug' */ @@ -62,6 +66,9 @@ MR_Context *MR_runqueue_tai #ifdef MR_LL_PARALLEL_CONJ MR_SparkDeque MR_spark_queue; MercuryLock MR_sync_term_lock; + MR_bool MR_thread_pinning = MR_FALSE; + static MercuryLock MR_next_cpu_lock; + static MR_Unsigned MR_next_cpu = 0; #endif MR_PendingContext *MR_pending_contexts; @@ -134,6 +141,7 @@ MR_init_thread_stuff(void) #ifdef MR_LL_PARALLEL_CONJ MR_init_wsdeque(&MR_spark_queue, MR_INITIAL_GLOBAL_SPARK_QUEUE_SIZE); pthread_mutex_init(&MR_sync_term_lock, MR_MUTEX_ATTR); + pthread_mutex_init(&MR_next_cpu_lock, MR_MUTEX_ATTR); #ifdef MR_DEBUG_RUNTIME_GRANULARITY_CONTROL pthread_mutex_init(&MR_par_cond_stats_lock, MR_MUTEX_ATTR); #endif @@ -155,6 +163,66 @@ MR_init_thread_stuff(void) pthread_cond_init(&MR_thread_barrier_cond, MR_COND_ATTR); #endif + /* + * Configure MR_num_threads if unset to match number of processors on the + * system, If we do this then we prepare to set processor affinities later + * on. + */ + if (MR_num_threads == 0) + { +#if defined(MR_HAVE_SYSCONF) && defined(MR_HAVE__SC_NPROCESSORS_ONLN) + long result; + + result = sysconf(_SC_NPROCESSORS_ONLN); + if (result < 1) { + /* We couldn't determine the number of processors. */ + MR_num_threads = 1; + } else { + MR_num_threads = result; + /* On systems that don't support sched_setaffinity we don't try to + ** automatically enable thread pinning. This prevents a runtime + ** warning that could unnecessarily confuse the user. + **/ +#if defined(MR_LL_PARALLEL_CONJ) && defined(MR_HAVE_SCHED_SETAFFINITY) + MR_thread_pinning = MR_TRUE; +#endif + } +#else + MR_num_threads = 1; +#endif + } +#endif +} + +void +MR_pin_thread(void) +{ +#if defined(MR_THREAD_SAFE) && defined(MR_LL_PARALLEL_CONJ) && \ + defined(MR_HAVE_SCHED_SETAFFINITY) + MR_LOCK(&MR_next_cpu_lock, "MR_pin_thread"); + if (MR_thread_pinning == MR_TRUE) { + cpu_set_t cpus; + + if (MR_next_cpu < CPU_SETSIZE) { + CPU_ZERO(&cpus); + CPU_SET(MR_next_cpu, &cpus); + if (sched_setaffinity(0, sizeof(cpu_set_t), &cpus) == 0) + { + MR_next_cpu++; + } else { + perror("Warning: Couldn't set CPU affinity"); + /* if this failed once it will probably fail again so disable + ** it. + */ + MR_thread_pinning = MR_FALSE; + } + } else { + perror("Warning: Couldn't set CPU affinity due to a static " \ + "system limit"); + MR_thread_pinning = MR_FALSE; + } + } + MR_UNLOCK(&MR_next_cpu_lock, "MR_pin_thread"); #endif } @@ -260,7 +328,7 @@ MR_write_out_profiling_parallel_executio MR_INTEGER_LENGTH_MODIFIER "ur, %" MR_INTEGER_LENGTH_MODIFIER \ "unr), sample %ul\n") #define MR_FPRINT_STATS_FORMAT_STRING_NONE \ - ("s: count %" MR_INTEGER_LENGTH_MODIFIER "u (%" \ + ("%s: count %" MR_INTEGER_LENGTH_MODIFIER "u (%" \ MR_INTEGER_LENGTH_MODIFIER "ur, %" MR_INTEGER_LENGTH_MODIFIER "unr)\n") static int Index: runtime/mercury_context.h =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_context.h,v retrieving revision 1.52 diff -u -p -b -r1.52 mercury_context.h --- runtime/mercury_context.h 16 Aug 2009 10:47:55 -0000 1.52 +++ runtime/mercury_context.h 20 Aug 2009 02:31:43 -0000 @@ -344,6 +344,7 @@ extern MR_Context *MR_runqueue_tai #ifdef MR_LL_PARALLEL_CONJ extern MR_SparkDeque MR_spark_queue; extern MercuryLock MR_sync_term_lock; + extern MR_bool MR_thread_pinning; #endif #if defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) @@ -458,13 +459,20 @@ extern MR_Context *MR_create_context(c extern void MR_destroy_context(MR_Context *context); /* -** MR_init_thread_stuff() initializes the lock structures for the runqueue. +** MR_init_thread_stuff() initializes the lock structures for the runqueue, +** and detect the number of threads to use on the LLC backend. */ extern void MR_init_thread_stuff(void); /* +** MR_pin_thread() pins the current thread to the next available processor ID, +** if thread pinning is enabled. +*/ +extern void MR_pin_thread(void); + +/* ** MR_finalize_thread_stuff() finalizes the lock structures for the runqueue -** amoung other things setup by MR_init_thread_stuff(). +** among other things setup by MR_init_thread_stuff(). */ extern void MR_finalize_thread_stuff(void); Index: runtime/mercury_thread.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_thread.c,v retrieving revision 1.34 diff -u -p -b -r1.34 mercury_thread.c --- runtime/mercury_thread.c 4 May 2009 01:50:41 -0000 1.34 +++ runtime/mercury_thread.c 19 Aug 2009 03:05:31 -0000 @@ -90,6 +90,7 @@ MR_create_thread_2(void *goal0) MR_init_thread(MR_use_now); (goal->func)(goal->arg); } else { + MR_pin_thread(); MR_init_thread(MR_use_later); } Index: runtime/mercury_wrapper.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_wrapper.c,v retrieving revision 1.198 diff -u -p -b -r1.198 mercury_wrapper.c --- runtime/mercury_wrapper.c 16 Aug 2009 10:18:36 -0000 1.198 +++ runtime/mercury_wrapper.c 19 Aug 2009 07:11:18 -0000 @@ -289,7 +289,13 @@ static char *MR_mem_usage_report static int MR_num_output_args = 0; -MR_Unsigned MR_num_threads = 1; +/* +** This is initialized to zero, if it is still zero after configuration of the +** runtime but before threads are started then the number of processors on the +** system is detected and used if support is available. Otherwise we fall back +** to 1 +*/ +MR_Unsigned MR_num_threads = 0; static MR_bool MR_print_table_statistics = MR_FALSE; @@ -582,6 +588,7 @@ mercury_runtime_init(int argc, char **ar #ifdef MR_THREAD_SAFE /* MR_init_thread_stuff() must be called prior to MR_init_memory() */ MR_init_thread_stuff(); + MR_max_outstanding_contexts = MR_max_contexts_per_thread * MR_num_threads; MR_primordial_thread = pthread_self(); #endif @@ -623,9 +630,10 @@ mercury_runtime_init(int argc, char **ar for (i = 1 ; i < MR_num_threads ; i++) { MR_create_thread(NULL); } - + MR_pin_thread(); while (MR_num_idle_engines < MR_num_threads-1) { /* busy wait until the worker threads are ready */ + MR_ATOMIC_PAUSE; } } #endif /* ! MR_LL_PARALLEL_CONJ */ @@ -1208,6 +1216,7 @@ enum MR_long_option { MR_GEN_NONDETSTACK_REDZONE_SIZE, MR_GEN_NONDETSTACK_REDZONE_SIZE_KWORDS, MR_MAX_CONTEXTS_PER_THREAD, + MR_THREAD_PINNING, MR_PROFILE_PARALLEL_EXECUTION, MR_MDB_TTY, MR_MDB_IN, @@ -1305,6 +1314,7 @@ struct MR_option MR_long_opts[] = { { "gen-nondetstack-zone-size-kwords", 1, 0, MR_GEN_NONDETSTACK_REDZONE_SIZE_KWORDS }, { "max-contexts-per-thread", 1, 0, MR_MAX_CONTEXTS_PER_THREAD }, + { "thread-pinning", 0, 0, MR_THREAD_PINNING }, { "profile-parallel-execution", 0, 0, MR_PROFILE_PARALLEL_EXECUTION }, { "mdb-tty", 1, 0, MR_MDB_TTY }, { "mdb-in", 1, 0, MR_MDB_IN }, @@ -1718,6 +1728,11 @@ MR_process_options(int argc, char **argv MR_max_contexts_per_thread = size; break; + case MR_THREAD_PINNING: +#if defined(MR_THREAD_SAFE) && defined(MR_LL_PARALLEL_CONJ) + MR_thread_pinning = MR_TRUE; +#endif + case MR_PROFILE_PARALLEL_EXECUTION: #if defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) MR_profile_parallel_execution = MR_TRUE; @@ -2201,8 +2216,6 @@ MR_process_options(int argc, char **argv } } - MR_max_outstanding_contexts = MR_max_contexts_per_thread * MR_num_threads; - if (MR_lld_print_min > 0 || MR_lld_start_name != NULL) { MR_lld_print_enabled = 0; } |
|
|
Re: Parallel runtime thread pinning.
by Peter Wang-6
::
Rate this Message:
|
|
|
Re: Parallel runtime thread pinning.
by Paul Bone
::
Rate this Message:
On Mon, Aug 24, 2009 at 09:33:27AM +1000, Peter Wang wrote:
> On 2009-08-24, Paul Bone <pbone@...> wrote: > > > > For post-commit review. > > > > A later changeset will address Julien's review comments on my other > > recent changes to the parallel runtime. > > > > > > Parallel runtime thread pinning. > > > > This change introduces two new features in the mercury runtime; > > pinning of threads to CPU cores/threads and runtime detection of the number of > > CPU cores/threads available. > > > > If MR_num_threads has not been specified in the runtime options with the -P > > flag we use the sysconf(_SC_NPROCESSORS_ONLN) call if available to detect the > > number of CPUs online and set MR_num_threads available. As before this > > defaults to 1. > > Excellent. > > Index: configure.in > > =================================================================== > > RCS file: /home/mercury1/repository/mercury/configure.in,v > > retrieving revision 1.541 > > diff -u -p -b -r1.541 configure.in > > --- configure.in 18 Aug 2009 05:10:39 -0000 1.541 > > +++ configure.in 19 Aug 2009 06:43:37 -0000 > > > @@ -1138,6 +1139,24 @@ MERCURY_CHECK_FOR_FENV_FUNC([fesetround] > > > > #-----------------------------------------------------------------------------# > > # > > +# Check for declarations. > > +# > > + > > +mercury_check_for_declarations () { > > + for mercury_cv_decl in "$@" > > + do > > + mercury_cv_decl_define="MR_HAVE_`echo $mercury_cv_decl | \ > > + tr abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ`" > > + AC_CHECK_DECL($mercury_cv_decl, [ > > + AC_DEFINE_UNQUOTED($mercury_cv_decl_define) > > + ]) > > + done > > +} > > + > > +mercury_check_for_declarations "_SC_NPROCESSORS_ONLN" > > Can't you just use #ifdef _SC_NPROCESSORS_ONLN in the source code? constant. Using an #ifdef in the source code won't work for emums. I'll add a comment describing this rationale. I'd be surprised if these weren't macros. If anyone knows for sure if such an assumption is safe please let me know. > > Index: doc/user_guide.texi > > =================================================================== > > RCS file: /home/mercury1/repository/mercury/doc/user_guide.texi,v > > retrieving revision 1.590 > > diff -u -p -b -r1.590 user_guide.texi > > --- doc/user_guide.texi 19 Aug 2009 07:45:11 -0000 1.590 > > +++ doc/user_guide.texi 23 Aug 2009 22:27:28 -0000 > > @@ -9608,6 +9611,17 @@ grade. > > @c and the @samp{MR_PROFILE_PARALLEL_EXECUTION_SUPPORT} C macro was > > @c defined when the runtime system was compiled. > > > > +@sp 1 > > +@item --thread-pinning > > +@findex --thread-pinning > > +Request that the runtime system attempts to pin mercury engines (POSIX threads) > > +to CPU cores/hardware threads. > > Mercury > > +This only has an effect if the executable was build in a parallel low-level C > > +grade. > > built Fixed, > > +It is disabled by default unless @samp{-P @var{num}} is not specified and the > > +runtime system is able to detect the number of processors enabled by the > > +operating system. > > I think it should just be disabled unless explicitly enabled. You can't assume > that the machine is dedicated to your process. That's up to the operating system's scheduler. Pinning doesn't ask for exclusive access to a CPU, just that the thread can only be ran on that particular CPU. I'll disable this for now, at least until we know that it is a net benefit in most situations. I'm concerned that if two parallel Mercury programs are scheduled and both pin their main thread to the same CPU (very probable) and the programs don't parallelize well the operating system could run them more optimally if it migrated one of their master threads to an otherwise idle CPU. Thanks. > > Index: runtime/mercury_conf.h.in > > =================================================================== > > RCS file: /home/mercury1/repository/mercury/runtime/mercury_conf.h.in,v > > retrieving revision 1.64 > > diff -u -p -b -r1.64 mercury_conf.h.in > > --- runtime/mercury_conf.h.in 30 Jul 2009 04:02:56 -0000 1.64 > > +++ runtime/mercury_conf.h.in 19 Aug 2009 07:13:41 -0000 > > @@ -136,6 +136,7 @@ > > ** MR_HAVE_FENV_H we have <fenv.h> > > ** MR_HAVE_SYS_MMAN_H we have <sys/mman.h> > > ** MR_HAVE_SYS_SEM_H we have <sys/sem.h> > > +** MR_HAVE_SCHED_H we have <sched.h> > > Indentation. sw=8 ts=8 noexpandtab I've made the other corrections you recommended. Thanks again. |
|
|
Re: Parallel runtime thread pinning.
by Julien Fischer-2
::
Rate this Message:
|
|
|
Re: Parallel runtime thread pinning.
by Paul Bone
::
Rate this Message:
On Tue, Aug 25, 2009 at 02:15:52AM +1000, Julien Fischer wrote:
> > sysconf() is posix and the parameters for it that are a part of posix > *must* be macros. It is extremely improbable that any non-standard > parameters like this one would be defined using an enum. > > It will be sufficient to use the #ifdef here - it will also avoid > yet more clutter in the configure script. > The following patch makes these corrections and makes some runtime modules tidier and more conformant to our standards. --- For post commit review. Corrections in response to review comments on recent parallel runtime changes. Made thread pinning off by default, the operating system should handle this unless we have a good reason to. configure.in: Removed newly added declaration checking code. doc/user_guide.texi: Documentation corrections. Adjusted the --pin-threads runtime option default. runtime/mercury_atomic_ops.c runtime/mercury_atomic_ops.h Use __x86_64__ instead of __amd64__ Altered comments at the beginning of sections of the file to better describe the contents of that section. Placed comments at the end of long conditional compilation blocks that match the condition at the beginning of the block. runtime/mercury_conf.h.in: Added editor hint for vim at the top of the file. Remove newly added declarations section. runtime/mercury_context.c: Adjusted default behaviour of --pin-threads Fixed some style issues. runtime/mercury_context.h: Fixed grammatical error. runtime/mercury_wrapper.c: Fixed grammatical error. Fixed a missing break statement in a switch statement. Index: configure.in =================================================================== RCS file: /home/mercury1/repository/mercury/configure.in,v retrieving revision 1.551 diff -u -p -b -r1.551 configure.in --- configure.in 26 Oct 2009 13:32:47 -0000 1.551 +++ configure.in 5 Nov 2009 04:09:46 -0000 @@ -1173,24 +1173,6 @@ MERCURY_CHECK_FOR_FENV_FUNC([fesetround] #-----------------------------------------------------------------------------# # -# Check for declarations. -# - -mercury_check_for_declarations () { - for mercury_cv_decl in "$@" - do - mercury_cv_decl_define="MR_HAVE_`echo $mercury_cv_decl | \ - tr abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ`" - AC_CHECK_DECL($mercury_cv_decl, [ - AC_DEFINE_UNQUOTED($mercury_cv_decl_define) - ]) - done -} - -mercury_check_for_declarations "_SC_NPROCESSORS_ONLN" - -#-----------------------------------------------------------------------------# -# # Check the basics of sigaction # Index: doc/user_guide.texi =================================================================== RCS file: /home/mercury1/repository/mercury/doc/user_guide.texi,v retrieving revision 1.597 diff -u -p -b -r1.597 user_guide.texi --- doc/user_guide.texi 30 Oct 2009 03:33:26 -0000 1.597 +++ doc/user_guide.texi 5 Nov 2009 04:03:39 -0000 @@ -9711,7 +9711,7 @@ This only has an effect if the executabl @findex -P (runtime option) Tells the runtime system to use @var{num} threads if the program was built in a parallel low-level C grade. -The mercury runtime attempts to automatically determine this value if support +The Mercury runtime attempts to automatically determine this value if support is available from the operating system. If it cannot or support is unavailable it defaults to @samp{1}. @@ -9737,13 +9737,15 @@ grade. @sp 1 @item --thread-pinning @findex --thread-pinning -Request that the runtime system attempts to pin mercury engines (POSIX threads) +Request that the runtime system attempts to pin Mercury engines (POSIX threads) to CPU cores/hardware threads. -This only has an effect if the executable was build in a parallel low-level C +This only has an effect if the executable was built in a parallel low-level C grade. -It is disabled by default unless @samp{-P @var{num}} is not specified and the -runtime system is able to detect the number of processors enabled by the -operating system. +This is disabled by default but may be enabled by default in the future. +@c In case this is enabled by default the following comment is relevant. +@c This is disabled by default unless @samp{-P @var{num}} is not specified and the +@c runtime system is able to detect the number of processors enabled by the +@c operating system. @c @item -r @var{num} @c @findex -r (runtime option) Index: runtime/mercury_atomic_ops.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_atomic_ops.c,v retrieving revision 1.4 diff -u -p -b -r1.4 mercury_atomic_ops.c --- runtime/mercury_atomic_ops.c 16 Aug 2009 10:18:36 -0000 1.4 +++ runtime/mercury_atomic_ops.c 24 Aug 2009 04:49:48 -0000 @@ -14,11 +14,12 @@ #include "mercury_imp.h" #include "mercury_atomic_ops.h" +/*---------------------------------------------------------------------------*/ + #if defined(MR_LL_PARALLEL_CONJ) -/*---------------------------------------------------------------------------*/ /* -** Provide definitions for functions declared `extern inline'. +** Definitions for the atomic functions declared `extern inline'. */ MR_OUTLINE_DEFN( @@ -69,20 +70,26 @@ MR_OUTLINE_DEFN( #endif /* MR_LL_PARALLEL_CONJ */ +/*---------------------------------------------------------------------------*/ + #if defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +/* +** Profiling of the parallel runtime. +*/ + +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) static MR_bool MR_rdtscp_is_available = MR_FALSE; static MR_bool MR_rdtsc_is_available = MR_FALSE; #endif -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) /* Set this to 1 to enable some printfs below */ #define MR_DEBUG_CPU_FEATURE_DETECTION 0 /* -** cpuid, rdtscp and rdtsc are i386/amd64 instructions. +** cpuid, rdtscp and rdtsc are i386/x86_64 instructions. */ static __inline__ void MR_cpuid(MR_Unsigned code, MR_Unsigned sub_code, @@ -94,18 +101,18 @@ MR_rdtscp(MR_uint_least64_t *tsc, MR_Uns static __inline__ void MR_rdtsc(MR_uint_least64_t *tsc); -#endif +#endif /* __GNUC__ && (__i386__ || __x86_64__) */ extern void MR_configure_profiling_timers(void) { -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) MR_Unsigned a, b, c, d; MR_Unsigned eflags, old_eflags; /* ** Check for the CPUID instruction. CPUID is supported if we can flip bit ** 21 in the CPU's EFLAGS register. The assembly below is written in a - ** subset of i386 and amd64 assembly. To read and write EFLAGS we have + ** subset of i386 and x86_64 assembly. To read and write EFLAGS we have ** to go via the C stack. */ __asm__ ("pushf; pop %0" @@ -251,12 +258,12 @@ MR_configure_profiling_timers(void) { #endif MR_rdtscp_is_available = MR_TRUE; -#endif +#endif /* __GNUC__ && (__i386__ || __x86_64__) */ } extern void MR_profiling_start_timer(MR_Timer *timer) { -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) /* ** If we don't have enough data to fill in all the fields of this structure ** we leave them alone, we won't check them later without checking @@ -275,7 +282,7 @@ MR_profiling_start_timer(MR_Timer *timer extern void MR_profiling_stop_timer(MR_Timer *timer, MR_Stats *stats) { -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) MR_Timer now; MR_int_least64_t duration; MR_uint_least64_t duration_squared; @@ -303,16 +310,16 @@ MR_profiling_stop_timer(MR_Timer *timer, MR_atomic_add_int(&(stats->MR_stat_sum), duration); MR_atomic_add_int(&(stats->MR_stat_sum_squares), duration_squared); } -#elif +#elif /* not __GNUC__ && (__i386__ || __x86_64__) */ /* No TSC support on this architecture or with this C compiler */ MR_atomic_inc_int(&(stats->MR_stat_count_recorded)); -#endif +#endif /* not __GNUC__ && (__i386__ || __x86_64__) */ } /* -** It's convenient that this instruction is the same on both i386 and amd64 +** It's convenient that this instruction is the same on both i386 and x86_64 */ -#if defined(__GNUC__) && (defined(__i386__) || defined(__amd64__)) +#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__)) static __inline__ void MR_cpuid(MR_Unsigned code, MR_Unsigned sub_code, @@ -348,7 +355,7 @@ MR_rdtsc(MR_uint_least64_t *tsc) { *tsc |= tsc_high; } -#endif +#endif /* __GNUC__ && (__i386__ || __x86_64__) */ -#endif +#endif /* MR_THREAD_SAFE && MR_PROFILE_PARALLEL_EXECUTION_SUPPORT */ Index: runtime/mercury_atomic_ops.h =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_atomic_ops.h,v retrieving revision 1.5 diff -u -p -b -r1.5 mercury_atomic_ops.h --- runtime/mercury_atomic_ops.h 16 Aug 2009 10:47:55 -0000 1.5 +++ runtime/mercury_atomic_ops.h 24 Aug 2009 05:01:53 -0000 @@ -8,7 +8,8 @@ */ /* -** mercury_atomic.h - defines atomic operations. +** mercury_atomic.h - defines atomic operations and other primitives used by +** the parallel runtime. */ #ifndef MERCURY_ATOMIC_OPS_H @@ -16,20 +17,15 @@ #include "mercury_std.h" -/* -** AMD say that __amd64__ is defined by the compiler for 64bit platforms, -** Intel say that __x86_64__ is the correct macro. Really these refer to -** the same thing that is simply branded differently, we use __amd64__ below -** and define it if necessary ourselves. -*/ -#if defined(__x86_64__) && !defined(__amd64__) -#define __amd64__ -#endif - /*---------------------------------------------------------------------------*/ + #if defined(MR_LL_PARALLEL_CONJ) /* +** Declarations for inline atomic operations. +*/ + +/* ** If the value at addr is equal to old, assign new to addr and return true. ** Otherwise return false. */ @@ -47,7 +43,7 @@ MR_compare_and_swap_word(volatile MR_Int return __sync_bool_compare_and_swap(addr, old, new_val); \ } while (0) -#elif defined(__GNUC__) && defined(__amd64__) +#elif defined(__GNUC__) && defined(__x86_64__) #define MR_COMPARE_AND_SWAP_WORD_BODY \ do { \ @@ -95,7 +91,7 @@ MR_compare_and_swap_word(volatile MR_Int MR_EXTERN_INLINE void MR_atomic_inc_int(volatile MR_Integer *addr); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_INC_INT_BODY \ do { \ @@ -148,7 +144,7 @@ MR_atomic_inc_int(volatile MR_Integer *a MR_EXTERN_INLINE void MR_atomic_dec_int(volatile MR_Integer *addr); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_DEC_INT_BODY \ do { \ @@ -191,7 +187,7 @@ MR_atomic_dec_int(volatile MR_Integer *a MR_EXTERN_INLINE void MR_atomic_add_int(volatile MR_Integer *addr, MR_Integer addend); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_ADD_INT_BODY \ do { \ @@ -233,7 +229,7 @@ MR_atomic_add_int(volatile MR_Integer *a MR_EXTERN_INLINE void MR_atomic_sub_int(volatile MR_Integer *addr, MR_Integer x); -#if defined(__GNUC__) && defined(__amd64__) +#if defined(__GNUC__) && defined(__x86_64__) #define MR_ATOMIC_SUB_INT_BODY \ do { \ @@ -294,7 +290,7 @@ MR_atomic_sub_int(volatile MR_Integer *a * References: Intel and AMD documentation for PAUSE, Intel optimisation * guide. */ -#if defined(__GNUC__) && ( defined(__i386__) || defined(__amd64__) ) +#if defined(__GNUC__) && ( defined(__i386__) || defined(__x86_64__) ) #define MR_ATOMIC_PAUSE \ do { \ @@ -323,6 +319,10 @@ MR_atomic_sub_int(volatile MR_Integer *a #if defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) +/* +** Declarations for profiling the parallel runtime. +*/ + typedef struct { MR_Unsigned MR_stat_count_recorded; MR_Unsigned MR_stat_count_not_recorded; @@ -344,7 +344,7 @@ typedef struct { } MR_Timer; /* -** Configure the profiling stats code. On i386 and amd64 machines this uses +** Configure the profiling stats code. On i386 and x86_64 machines this uses ** CPUID to determine if the RDTSCP instruction is available and not prohibited ** by the OS. */ @@ -363,7 +363,7 @@ MR_profiling_start_timer(MR_Timer *timer extern void MR_profiling_stop_timer(MR_Timer *timer, MR_Stats *stats); -#endif /* MR_THREAD_SAFE && MR_PROFILE_PARALLEL_EXECUTION */ +#endif /* MR_THREAD_SAFE && MR_PROFILE_PARALLEL_EXECUTION_SUPPORT */ /*---------------------------------------------------------------------------*/ Index: runtime/mercury_conf.h.in =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_conf.h.in,v retrieving revision 1.65 diff -u -p -b -r1.65 mercury_conf.h.in --- runtime/mercury_conf.h.in 23 Aug 2009 22:52:35 -0000 1.65 +++ runtime/mercury_conf.h.in 5 Nov 2009 04:12:08 -0000 @@ -1,4 +1,7 @@ /* +** vim:ts=8 sw=8 noexpandtab +*/ +/* ** Copyright (C) 1995-2003, 2005-2009 The University of Melbourne. ** This file may only be copied under the terms of the GNU Library General ** Public License - see the file COPYING.LIB in the Mercury distribution. @@ -362,14 +365,6 @@ #undef MR_HAVE_SIGCONTEXT_STRUCT_2ARG /* -** These specify weather the given C macros are defined. -** -** MR_HAVE__SC_NPROCESSORS_ONLN, This is defined as a parameter for sysconf to -** determine the number of processors online. -*/ -#undef MR_HAVE__SC_NPROCESSORS_ONLN - -/* ** For debugging purposes, if we get a fatal signal, we print out the ** program counter (PC) at which the signal occurred. ** Index: runtime/mercury_context.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_context.c,v retrieving revision 1.68 diff -u -p -b -r1.68 mercury_context.c --- runtime/mercury_context.c 23 Aug 2009 22:52:35 -0000 1.68 +++ runtime/mercury_context.c 5 Nov 2009 04:16:24 -0000 @@ -170,7 +170,7 @@ MR_init_thread_stuff(void) */ if (MR_num_threads == 0) { -#if defined(MR_HAVE_SYSCONF) && defined(MR_HAVE__SC_NPROCESSORS_ONLN) +#if defined(MR_HAVE_SYSCONF) && defined(_SC_NPROCESSORS_ONLN) long result; result = sysconf(_SC_NPROCESSORS_ONLN); @@ -179,19 +179,24 @@ MR_init_thread_stuff(void) MR_num_threads = 1; } else { MR_num_threads = result; - /* On systems that don't support sched_setaffinity we don't try to + /* + ** On systems that don't support sched_setaffinity we don't try to ** automatically enable thread pinning. This prevents a runtime ** warning that could unnecessarily confuse the user. **/ #if defined(MR_LL_PARALLEL_CONJ) && defined(MR_HAVE_SCHED_SETAFFINITY) - MR_thread_pinning = MR_TRUE; + /* + ** Comment this back in to enable thread pinning by default if we + ** autodetected the correct number of CPUs. + */ + /* MR_thread_pinning = MR_TRUE; */ #endif } -#else +#else /* ! defined(MR_HAVE_SYSCONF) && defined(_SC_NPROCESSORS_ONLN) */ MR_num_threads = 1; -#endif +#endif /* ! defined(MR_HAVE_SYSCONF) && defined(_SC_NPROCESSORS_ONLN) */ } -#endif +#endif /* MR_THREAD_SAFE */ } void @@ -200,7 +205,7 @@ MR_pin_thread(void) #if defined(MR_THREAD_SAFE) && defined(MR_LL_PARALLEL_CONJ) && \ defined(MR_HAVE_SCHED_SETAFFINITY) MR_LOCK(&MR_next_cpu_lock, "MR_pin_thread"); - if (MR_thread_pinning == MR_TRUE) { + if (MR_thread_pinning) { cpu_set_t cpus; if (MR_next_cpu < CPU_SETSIZE) { @@ -211,19 +216,21 @@ MR_pin_thread(void) MR_next_cpu++; } else { perror("Warning: Couldn't set CPU affinity"); - /* if this failed once it will probably fail again so disable - ** it. + /* + ** If this failed once it will probably fail again so we + ** disable it. */ MR_thread_pinning = MR_FALSE; } } else { - perror("Warning: Couldn't set CPU affinity due to a static " \ + perror("Warning: Couldn't set CPU affinity due to a static " "system limit"); MR_thread_pinning = MR_FALSE; } } MR_UNLOCK(&MR_next_cpu_lock, "MR_pin_thread"); -#endif +#endif /* defined(MR_THREAD_SAFE) && defined(MR_LL_PARALLEL_CONJ) && \ + defined(MR_HAVE_SCHED_SETAFFINITY) */ } void @@ -364,7 +371,7 @@ fprint_stats(FILE *stream, const char *m } }; -#endif +#endif /* defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) */ static void MR_init_context_maybe_generator(MR_Context *c, const char *id, Index: runtime/mercury_context.h =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_context.h,v retrieving revision 1.53 diff -u -p -b -r1.53 mercury_context.h --- runtime/mercury_context.h 23 Aug 2009 22:52:35 -0000 1.53 +++ runtime/mercury_context.h 24 Aug 2009 00:37:46 -0000 @@ -460,7 +460,7 @@ extern void MR_destroy_context(M /* ** MR_init_thread_stuff() initializes the lock structures for the runqueue, -** and detect the number of threads to use on the LLC backend. +** and detects the number of threads to use on the LLC backend. */ extern void MR_init_thread_stuff(void); Index: runtime/mercury_wrapper.c =================================================================== RCS file: /home/mercury1/repository/mercury/runtime/mercury_wrapper.c,v retrieving revision 1.200 diff -u -p -b -r1.200 mercury_wrapper.c --- runtime/mercury_wrapper.c 30 Oct 2009 03:33:30 -0000 1.200 +++ runtime/mercury_wrapper.c 5 Nov 2009 03:59:47 -0000 @@ -290,7 +290,7 @@ static char *MR_mem_usage_report static int MR_num_output_args = 0; /* -** This is initialized to zero, if it is still zero after configuration of the +** This is initialized to zero. If it is still zero after configuration of the ** runtime but before threads are started then the number of processors on the ** system is detected and used if support is available. Otherwise we fall back ** to 1 @@ -1736,6 +1736,7 @@ MR_process_options(int argc, char **argv #if defined(MR_THREAD_SAFE) && defined(MR_LL_PARALLEL_CONJ) MR_thread_pinning = MR_TRUE; #endif + break; case MR_PROFILE_PARALLEL_EXECUTION: #if defined(MR_THREAD_SAFE) && defined(MR_PROFILE_PARALLEL_EXECUTION_SUPPORT) |
signature.asc (492 bytes) | Free embeddable forum powered by Nabble | Forum Help |
