http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57024
--- Comment #1 from Marc Glisse <glisse at gcc dot gnu.org> 2013-04-21 17:33:46
UTC ---
Maybe the processor is just badly detected or modeled? I noticed on the same
machine another surprising tuning issue. Adding -march=native changes for some
other code:
.cfi_startproc
- movsd %xmm0, -8(%rsp)
- movq -8(%rsp), %rax
+ movq %xmm0, %rax
shrq $32, %rax
andl $2146435072, %eax
subl $2146435072, %eax
shrl $31, %eax
ret
.cfi_endproc
and slows the function down by 10% (I assume that the difference is due to the
instructions, not some random code alignment difference, I could be wrong).
This is a well known tuning parameter, so I am surprised -march=native makes a
worse decision than generic. For reference, the output with -v:
/usr/lib/gcc-snapshot/libexec/gcc/x86_64-linux-gnu/4.9.0/cc1 -quiet -v
-imultilib . -imultiarch x86_64-linux-gnu -D finite=finite1 test.c -march=core2
-mcx16 -msahf -mno-movbe -mno-aes -mno-pclmul -mno-popcnt -mno-abm -mno-lwp
-mno-fma -mno-fma4 -mno-xop -mno-bmi -mno-bmi2 -mno-tbm -mno-avx -mno-avx2
-mno-sse4.2 -msse4.1 -mno-lzcnt -mno-rtm -mno-hle -mno-rdrnd -mno-f16c
-mno-fsgsbase -mno-rdseed -mno-prfchw -mno-adx -mfxsr -mno-xsave -mno-xsaveopt
--param l1-cache-size=32 --param l1-cache-line-size=64 --param
l2-cache-size=6144 -mtune=core2 -quiet -dumpbase test.c -auxbase-strip y.s -O3
-Wall -Wextra -version -fno-builtin-finite -o y.s
And full /proc/cpuinfo (for the first core):
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 23
model name : Intel(R) Core(TM)2 Duo CPU T9600 @ 2.80GHz
stepping : 10
microcode : 0xa07
cpu MHz : 2800.000
cache size : 6144 KB
physical id : 0
siblings : 2
core id : 0
cpu cores : 2
apicid : 0
initial apicid : 0
fpu : yes
fpu_exception : yes
cpuid level : 13
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov
pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm
constant_tsc arch_perfmon pebs bts rep_good nopl aperfmperf pni dtes64 monitor
ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 xsave lahf_lm ida dtherm
tpr_shadow vnmi flexpriority
bogomips : 5584.90
clflush size : 64
cache_alignment : 64
address sizes : 36 bits physical, 48 bits virtual
power management:
