| View previous topic :: View next topic |
| Author |
Message |
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
 Posted: Tue Sep 12, 2006 11:43 pm Post subject: post your new GCC compile times |
 |
|
| Code: | Mon Sep 11 14:11:43 2006 >>> sys-devel/gcc-4.1.1
merge time: 31 minutes and 33 seconds.
|
| Code: | Tue Sep 12 19:42:22 2006 >>> sys-devel/gcc-4.1.1-r1
merge time: 31 minutes and 31 seconds.
|
I would just like to add that i beat an Intel core Duo 2.8Ghz by 10 mins with mine 
Athlon64 3700 (overclocked)
1Gb RAM
CFLAGS="-march=i686 -O2 -pipe -fomit-frame-pointer"
250GB Maxtor SATA2
| Code: | [ebuild R ] sys-devel/gcc-4.1.1-r1 USE="fortran gtk nls (-altivec) -bootstrap -build -doc -gcj (-hardened) -ip28 -ip32r10k -mudflap (-multilib) -multislot (-n32) (-n64) -nocxx -objc -objc++ -objc-gc -test -vanilla" 0 kB
|
Last edited by cokehabit on Wed Sep 13, 2006 11:31 pm; edited 4 times in total |
|
| Back to top |
|
 |
Phenax
l33t
Joined: 10 Mar 2006
Posts: 972
|
| Posted: Tue Sep 12, 2006 11:49 pm Post subject: |
|
|
| Code: | Mon Sep 11 05:07:30 2006 >>> sys-devel/gcc-4.1.1-r1
merge time: 36 minutes and 41 seconds.
Fri Sep 8 18:51:45 2006 >>> sys-devel/gcc-4.2.0_alpha20060902
merge time: 40 minutes and 52 seconds.
|
| Code: |
AMD Athlon(tm) 64 Processor 3500+ @ 2.9ghz
2GB of DDR2 667 RAM
CFLAGS="-march=athlon64 -O2 -msse3 -pipe -mfpmath=sse"
|
Last edited by Phenax on Wed Sep 13, 2006 12:58 am; edited 3 times in total |
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Tue Sep 12, 2006 11:57 pm Post subject: |
|
|
| i think we should also post our computer specs and CFLAGS |
|
| Back to top |
|
|
xanas3712
Guru
Joined: 15 Oct 2004
Posts: 455
|
| Posted: Wed Sep 13, 2006 1:48 am Post subject: |
|
|
CPU: AMD 3800+ X2 (obviously not overclocked considering my compile time  )
RAM: 2G PC3200 DDR
CFLAGS="-march=athlon64 -pipe -O2"
| Code: |
Sun Sep 3 08:07:25 2006 >>> sys-devel/gcc-4.1.1
merge time: 44 minutes and 36 seconds.
|
|
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Wed Sep 13, 2006 2:29 am Post subject: |
|
|
| why am i beating people's dual core times by 10 mins? |
|
| Back to top |
|
|
Phenax
l33t
Joined: 10 Mar 2006
Posts: 972
|
| Posted: Wed Sep 13, 2006 2:30 am Post subject: |
|
|
| cokehabit wrote: | | why am i beating people's dual core times by 10 mins? |
Could be ccache, or maybe x86_64 vs x86-based compilations have a speed difference with GCC, USE flags possibly. |
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Wed Sep 13, 2006 2:53 am Post subject: |
|
|
| Phenax wrote: | | cokehabit wrote: | | why am i beating people's dual core times by 10 mins? |
Could be ccache, or maybe x86_64 vs x86-based compilations have a speed difference with GCC, USE flags possibly. |
no ccache but i am running x86 gentoo on an amd64 |
|
| Back to top |
|
|
Moloch
Apprentice
Joined: 17 Mar 2003
Posts: 293
Location: Albuquerque, NM, US
|
| Posted: Wed Sep 13, 2006 5:15 am Post subject: |
|
|
| Code: | Sun Sep 3 14:56:33 2006 >>> sys-devel/gcc-4.1.1
merge time: 50 minutes and 26 seconds. |
Intel Pentium M processor 1.73GHz/512MB RAM
CFLAGS="-O2 -march=pentium-m -pipe"
I am using ccache. Guess that helped a lot.
_________________
Understanding is a three-edged sword: your side, their side, and the truth. --Kosh
1010011010 |
|
| Back to top |
|
|
John-Boy
Guru
Joined: 23 Jun 2004
Posts: 436
Location: Desperately seeking Moksha in all the wrong places
|
| Posted: Wed Sep 13, 2006 12:54 pm Post subject: |
|
|
Doesn't ccache get cleared when emerging in a new gcc build ?
_________________
When you break rules, break 'em good and hard |
|
| Back to top |
|
|
FGA
Apprentice
Joined: 07 Apr 2004
Posts: 179
|
| Posted: Wed Sep 13, 2006 1:41 pm Post subject: |
|
|
| Code: | Thu Sep 7 12:08:33 2006 >>> sys-devel/gcc-4.1.1
merge time: 50 minutes and 28 seconds. |
Athlon64 3400+ 2,4Ghz, not overclocked. Using DDR266 memory (1GB). Not using ccache. |
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Wed Sep 13, 2006 1:56 pm Post subject: |
|
|
| FGA wrote: | | Code: | Thu Sep 7 12:08:33 2006 >>> sys-devel/gcc-4.1.1
merge time: 50 minutes and 28 seconds. |
Athlon64 3400+ 2,4Ghz, not overclocked. Using DDR266 memory (1GB). Not using ccache. |
Y could have put that you have a AMD Athlon 64 3400+ | ASUS K8V SE Deluxe | VIA K8T800 | 1GB DDR400 | nVidia fx5700VE | Creative Sound Blaster Live! + Live!Drive | 2x 80GB ATA133 RAID-0 | WANADOO ADSL 1024/320 WIFI |
|
| Back to top |
|
|
FGA
Apprentice
Joined: 07 Apr 2004
Posts: 179
|
| Posted: Wed Sep 13, 2006 2:07 pm Post subject: |
|
|
| cokehabit wrote: | | FGA wrote: | | Code: | Thu Sep 7 12:08:33 2006 >>> sys-devel/gcc-4.1.1
merge time: 50 minutes and 28 seconds. |
Athlon64 3400+ 2,4Ghz, not overclocked. Using DDR266 memory (1GB). Not using ccache. |
Y could have put that you have a AMD Athlon 64 3400+ | ASUS K8V SE Deluxe | VIA K8T800 | 1GB DDR400 | nVidia fx5700VE | Creative Sound Blaster Live! + Live!Drive | 2x 80GB ATA133 RAID-0 | WANADOO ADSL 1024/320 WIFI |
Yeah that's my rig. Except the DDR400 bit, a 512MB module was incompatable with my motherboard and I had to replace it with a DDR266 one.
What's the mean of your post?
EDIT: I have noticed that on iexplorer and firefox, my sig is too tiny, is that? I thought it was big enought because konqueror renders it at normal size...
Last edited by FGA on Wed Sep 13, 2006 2:14 pm; edited 2 times in total |
|
| Back to top |
|
|
slithy
Guru
Joined: 26 Nov 2005
Posts: 321
Location: Kansas
|
| Posted: Wed Sep 13, 2006 2:11 pm Post subject: |
|
|
| Code: | Sat Sep 9 14:26:44 2006 >>> sys-devel/gcc-4.1.1-r1
merge time: 49 minutes and 7 seconds. |
| Code: | CPU: Athlon64 3500+ Venice @ stock
Memory: 2x512 PC3200 DDR
CFLAGS="-march=k8 -O2 -pipe -msse3"
CHOST="x86_64-pc-linux-gnu" |
One reason for dual cores not having faster times is that they might not have the MAKE_OPTS set for threading their compiles.
Last edited by slithy on Wed Sep 13, 2006 4:42 pm; edited 1 time in total |
|
| Back to top |
|
|
R.Smith
Tux's lil' helper
Joined: 20 Nov 2005
Posts: 131
Location: Caerdydd, Cymru.
|
| Posted: Wed Sep 13, 2006 2:38 pm Post subject: |
|
|
| genlop wrote: |
Sat Aug 26 13:19:14 2006 >>> sys-devel/gcc-4.1.1
merge time: 1 hour, 2 minutes and 49 seconds.
Sat Aug 26 16:39:40 2006 >>> sys-devel/gcc-4.1.1
merge time: 59 minutes and 22 seconds.
Sat Aug 26 20:42:18 2006 >>> sys-devel/gcc-4.1.1
merge time: 1 hour, 3 minutes and 47 seconds.
|
amd64 Gentoo
Athlon 64 3200+ @ 2 GHz, not overclocked
1 GB RAM, DDR400
2 x 80 GB SATA Samsung Spin Point HD's, ext3 formatted with full journalling (probably impacted performance) and directory indexing
CFLAGS="-march=athlon64 -O2 -pipe"
No ccache
Last edited by R.Smith on Wed Sep 13, 2006 2:50 pm; edited 2 times in total |
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Wed Sep 13, 2006 2:39 pm Post subject: |
|
|
| FGA wrote: | What's the mean of your post?
EDIT: I have noticed that on iexplorer and firefox, my sig is too tiny, is that? I thought it was big enought because konqueror renders it at normal size... |
well it looks so tiny i thought you weren't expecting people to notice it |
|
| Back to top |
|
|
Uppi
n00b
Joined: 06 Jan 2004
Posts: 17
|
| Posted: Wed Sep 13, 2006 3:55 pm Post subject: |
|
|
| slithy wrote: |
One reason for dual cores not having faster times is that they might not have the MAKE_OPTS set for threading their compiles. |
I think that the gcc-4.1.1 ebuild somehow ignores the MAKE_OPTS setting, contrary to the gcc-3.4 ebuild:
| genlop wrote: |
Tue Jun 20 18:44:31 2006 >>> sys-devel/gcc-3.4.6-r1
merge time: 29 minutes and 14 seconds.
Thu Aug 31 18:49:28 2006 >>> sys-devel/gcc-4.1.1
merge time: 59 minutes and 41 seconds.
|
(Athlon64 X2 3800+)
I have -j3 in my MAKE_OPTS and I noticed that the average load is around 1 instead of the usual 2 |
|
| Back to top |
|
|
chunderbunny
Veteran
Joined: 31 May 2004
Posts: 1281
Location: 51°24'27" N, 0°57'15" W
|
| Posted: Wed Sep 13, 2006 5:10 pm Post subject: |
|
|
| Yeah, gcc-4.1 forces -j1 for some reason. I've noticed a few packages don't use both cores when compiling. |
|
| Back to top |
|
|
xanas3712
Guru
Joined: 15 Oct 2004
Posts: 455
|
| Posted: Wed Sep 13, 2006 6:38 pm Post subject: |
|
|
This is for stability reasons. A lot of applications will not compile due to issues with multi-threaded compilation, or they will compile with bugs.
I had a package like this (kdevelop) that I had to specifically compile forcing -j1 to get it to emerge. I try not to mess with that to prevent issues.
But from what I can tell of the results, besides the fact the other core isn't used it seems overclocking is indeed a huge benefit to compile times. I don't OC primarily due to the fact that my house is hot enough even on the athlon x2's I don't feel comfortable doing it. I could probably get away with it as the chip idles at around 33-35 and it maxes out at about 46, but I don't like pushing stability (not to mention my psu is a little under spec for my hardware). |
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Wed Sep 13, 2006 9:01 pm Post subject: |
|
|
| i wonder how many ways there is to push the compile times down... CFLAGS, portage niceness... |
|
| Back to top |
|
|
revenger
n00b
Joined: 23 Sep 2005
Posts: 23
Location: The Netherlands
|
| Posted: Wed Sep 13, 2006 10:07 pm Post subject: |
|
|
| Code: | Wed Sep 6 23:16:02 2006 >>> sys-devel/gcc-4.1.1-r1
merge time: 1 hour, 51 seconds. |
specs:
Athlon64 3500+
1GB (2x512 MB) PC3200 DDR SDRAM
Hitachi 250GB SATA HDD
running ~amd64 (gentoo-sources-2.6.17-r7)
CFLAGS="-march=athlon64 -O2 -msse3 -pipe -fomit-frame-pointer" |
|
| Back to top |
|
|
cokehabit
Advocate
Joined: 23 Apr 2004
Posts: 3302
|
| Posted: Wed Sep 13, 2006 10:27 pm Post subject: |
|
|
| revenger wrote: | | Code: | Wed Sep 6 23:16:02 2006 >>> sys-devel/gcc-4.1.1-r1
merge time: 1 hour, 51 seconds. |
specs:
Athlon64 3500+
1GB (2x512 MB) PC3200 DDR SDRAM
Hitachi 250GB SATA HDD
running ~amd64 (gentoo-sources-2.6.17-r7)
CFLAGS="-march=athlon64 -O2 -msse3 -pipe -fomit-frame-pointer" |
jesus, why is that so slow? |
|
| Back to top |
|
|
Phenax
l33t
Joined: 10 Mar 2006
Posts: 972
|
| Posted: Wed Sep 13, 2006 10:31 pm Post subject: |
|
|
| He was probably doing something/low niceness/etc.. |
|
| Back to top |
|
|
anachranerd
n00b
Joined: 28 Feb 2006
Posts: 11
Location: Missouri
|
| Posted: Wed Sep 13, 2006 10:33 pm Post subject: |
|
|
| Sorry for the noobish question, but how do you guys obtain that merge time? Is it a flag to portage? |
|
| Back to top |
|
|
think4urs11
Administrator
Joined: 25 Jun 2003
Posts: 6659
Location: above the cloud
|
| Posted: Wed Sep 13, 2006 10:33 pm Post subject: |
|
|
from the lower end of 'performant' machines
| Code: | Sat Sep 2 22:10:11 2006 >>> sys-devel/gcc-4.1.1
merge time: 5 hours, 1 minute and 9 seconds. |
Via C3/600Mhz/1GB/'-march=c3 -Os -pipe -fomit-frame-pointer'; no other processor intensive tasks during compile time
_________________
Nothing is secure / Security is always a trade-off with usability / Do not assume anything / Trust no-one, nothing / Paranoia is your friend / Think for yourself |
|
| Back to top |
|
|
revenger
n00b
Joined: 23 Sep 2005
Posts: 23
Location: The Netherlands
|
| Posted: Wed Sep 13, 2006 10:35 pm Post subject: |
|
|
| cokehabit wrote: | | jesus, why is that so slow? |
it could be that i was doing some other stuff while compiling, don't remember exactly. and because i did a recent fresh install i have no other history to compare with.
USE flags could also impact the compile time
| Code: | # emerge -pv gcc
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild R ] sys-devel/gcc-4.1.1-r1 USE="fortran gtk nls (-altivec) -bootstrap -build -doc -gcj (-hardened) -ip28 -ip32r10k -mudflap (-multilib) -multislot (-n32) (-n64) -nocxx -objc -objc++ -objc-gc -test -vanilla" 0 kB
Total size of downloads: 0 kB
|
|
|
| Back to top |
|
|
|
Off the Wall
