Not as exciting an improvement as I'd hoped:
4 GB RAM Intel(R) Core(TM)2 Quad CPU Q9300 @ 2.50GHz make -j5 real 7m35.985s user 24m54.741s sys 3m15.628s
6 GB RAM model name : Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz make -j9 real 4m39.577s user 30m43.651s sys 2m30.109s
On Tue, Mar 23, 2010 at 4:11 PM, Dan Kegel dank@kegel.com wrote:
Not as exciting an improvement as I'd hoped:
4 GB RAM Intel(R) Core(TM)2 Quad CPU Q9300 @ 2.50GHz make -j5 real 7m35.985s user 24m54.741s sys 3m15.628s
6 GB RAM model name : Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz make -j9 real 4m39.577s user 30m43.651s sys 2m30.109s
Even though Nehalem is a better architecture, you don't get the 2x (or better) speed up because the extra 4 cores are from hyperthreading. With 8 real Nehalem cores and -j9, I got 3m24s. With 8 cores + hyperthreading, I get 2m49s.
And now to figure out how much electricity I just used... ;-)
On Wed, Mar 24, 2010 at 1:08 AM, Lei Zhang thestig@google.com wrote:
Even though Nehalem is a better architecture, you don't get the 2x (or better) speed up because the extra 4 cores are from hyperthreading. With 8 real Nehalem cores and -j9, I got 3m24s. With 8 cores + hyperthreading, I get 2m49s.
Is that with -j17 on the latter measurement? Also, what model number CPU is that again? - Dan
On Wed, Mar 24, 2010 at 6:44 AM, Dan Kegel dank@kegel.com wrote:
On Wed, Mar 24, 2010 at 1:08 AM, Lei Zhang thestig@google.com wrote:
Even though Nehalem is a better architecture, you don't get the 2x (or better) speed up because the extra 4 cores are from hyperthreading. With 8 real Nehalem cores and -j9, I got 3m24s. With 8 cores + hyperthreading, I get 2m49s.
Is that with -j17 on the latter measurement? Also, what model number CPU is that again?
- Dan
Yes, with -j17. This is on the dual xeon E5520. (2.26 Ghz core i7)
Also of interest, a new version of ccache after 6 years: http://samba.org/ftp/ccache/ccache-3.0pre0-NEWS