This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Performance report (350MHz i686 configured as i386-unknown-freebsd4.8)


A while back, I promised to enable performance checking on the FreeBSD
daily regression checkers.  I still haven't done that but I have
tweaked/tested the stuff such that it is ready to be so enabled (the
few patches are all committed).  For current mainline:

                === libstdc++-v3 Summary ===

# of expected passes            1499

                === libstdc++-v3 check-abi Summary ===

# of added symbols:              41
# of missing symbols:            123
# of incompatible symbols:       129

Run without -pthread:

allocator.cc             	default   	1241r 1234u    0s    0pf 
allocator.cc             	malloc    	2433r 2163u    2s    0pf 
allocator.cc             	mt        	1389r 1303u    0s    0pf 
complex_norm.cc          	norm      	 709r  653u    0s    0pf 
complex_norm.cc          	          	 400r  396u    0s    0pf 
cout_insert_int.cc       	          	3231r  139u  206s    0pf 
filebuf_sputc.cc         	C         	4064r 2574u  582s    0pf 
filebuf_sputc.cc         	C unlocked	2639r 1417u  528s    0pf 
filebuf_sputc.cc         	C++       	4463r 3243u  352s    0pf 
fstream_seek_write.cc    	          	2291r 1008u 1143s    0pf 
ifstream_getline.cc      	          	  66r   41u    8s    1pf 
map_create_fill.cc       	          	5260r 5085u    0s    0pf 
ofstream_insert_float.cc 	          	43553r 40257u  606s    0pf 
ofstream_insert_int.cc   	          	4509r 3674u  292s    0pf 
string_append.cc         	char      	   0r    0u    0s    0pf 
string_append.cc         	string    	   1r    0u    0s    0pf 
string_append.cc         	char      	   7r    6u    0s    0pf 
string_append.cc         	string    	   9r    9u    0s    0pf 
string_append.cc         	char      	  67r   66u    1s    0pf 
string_append.cc         	string    	  86r   83u    2s    0pf 

Run with -pthread (by manually hacking libstdc++-v3/scripts/check_performance):

allocator.cc             	default   	1190r 1141u    0s    0pf 
allocator.cc             	malloc    	 426r  399u    1s    0pf 
allocator.cc             	mt        	 480r  468u    0s    0pf 
complex_norm.cc          	norm      	 799r  660u    0s    0pf 
complex_norm.cc          	          	 415r  398u    0s    0pf 
cout_insert_int.cc       	          	3332r  156u  179s    0pf 
filebuf_sputc.cc         	C         	3996r 2669u  608s    0pf 
filebuf_sputc.cc         	C unlocked	2904r 1503u  608s    0pf 
filebuf_sputc.cc         	C++       	4552r 3250u  411s    0pf 
fstream_seek_write.cc    	          	2603r 1190u 1295s    0pf 
ifstream_getline.cc      	          	  78r   44u    8s    0pf 
map_create_fill.cc       	          	61441r 56855u   80s    0pf 
ofstream_insert_float.cc 	          	39284r 35586u  602s    0pf 
ofstream_insert_int.cc   	          	4683r 3681u  352s    0pf 
string_append.cc         	char      	   1r    0u    0s    0pf 
string_append.cc         	string    	   1r    1u    0s    0pf 
string_append.cc         	char      	   7r    7u    0s    0pf 
string_append.cc         	string    	   8r    8u    0s    0pf 
string_append.cc         	char      	  68r   64u    2s    0pf 
string_append.cc         	string    	  82r   79u    1s    0pf 

(BTW, yes, I think we should run the test cases with and without
e.g. -pthread esp. if we find it matters even for non-threaded cases.
I'll make it possible to run either way without having to change a
srcdir file perhaps by respecting the same RUNTESTFLAG format that
dejagnu honors.)

Thanks again Benjamin for translating mailing list traffic into a
workable scheme!

Regards,
Loren


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]