Fw: !0.2-R is slow in some applications

[Jin Guojun]

I posted this in freebsd.question earlier and wish more people may see this and make comment, but got no answer. Now narrow to performance group.



 It looks like not just X server is slow on 10.2-R, other servers slow too and take a lot of CPU.forefox is play music from youtube, while thunderbird connected to imapd and doing nothing.
Those services take less than 10% total CPU when running 8.4 / 10.1.
Is something under 10.2-R needs to be tuned for performance? 
or this hardware combination cause 10.2 -R not working well.
Some hardware information is toward the end of this email (in the original part).

-Jin

8.4-Release top output for the same applications:
last pid:  1233;  load averages:  0.47,  0.46,  0.25    up 0+00:05:21  23:22:11
58 processes:  1 running, 57 sleeping
CPU:  4.5% user,  0.0% nice,  2.5% system,  0.4% interrupt, 92.6% idle
Mem: 317M Active, 303M Inact, 750M Wired, 1944K Cache, 651M Buf, 6485M Free
Swap: 12G Total, 12G Free

  PID USERNAME    THR PRI NICE   SIZE    RES STATE   C   TIME   WCPU COMMAND
 1104 jin          29  44    0   457M   323M ucond   3   1:11 26.27% firefox
 1066 jin           1  46    0   371M   328M select  3   0:17  3.08% Xorg
 1209 jin          26  44    0   289M   162M ucond   1   0:14  0.00% thunderbir
 1211 jin           1  44    0 18348K  4896K select  2   0:01  0.00% imapd
 1216 jin           1  44    0 18348K  4672K select  1   0:00  0.00% imapd
  864 root          1  44    0  8080K  1404K select  0   0:00  0.00% moused------------------------------------- vs ---------------------------------------

10.2-Release top output (2 in a couple of minutes away):
last pid:  1088;  load averages:  1.93,  1.07,  0.53    up 0+00:05:16  22:49:12
58 processes:  3 running, 55 sleeping
CPU: 19.9% user,  0.0% nice, 61.0% system,  0.3% interrupt, 18.8% idle
Mem: 402M Active, 513M Inact, 528M Wired, 192K Cache, 821M Buf, 6440M Free
Swap: 8192M Total, 8192M Free

  PID USERNAME    THR PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
  951 jin          75  20    0   711M   365M uwait   3   1:17 115.23% firefox
 1083 jin           1  89    0 36688K  5988K CPU0    0   0:47  55.37% imapd
 1081 jin           1  88    0 36688K  5988K CPU2    2   0:47  52.59% imapd
 1077 jin          44  21    0   555M   271M uwait   1   0:19  21.14% thunderbi
  923 jin           5  20    0   460M   363M uwait   0   0:09   2.20% Xorg
  777 root          1  20    0 16728K  2236K select  2   0:00   0.00% moused

last pid:  1090;  load averages:  3.67,  2.11,  1.04    up 0+00:07:51  22:51:47
58 processes:  3 running, 55 sleeping
CPU: 19.9% user,  0.0% nice, 63.8% system,  0.6% interrupt, 15.7% idle
Mem: 604M Active, 592M Inact, 576M Wired, 192K Cache, 820M Buf, 6110M Free
Swap: 8192M Total, 8192M Free

  PID USERNAME    THR PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
  951 jin          73  20    0   708M   364M uwait   2   4:22 112.79% firefox
 1083 jin           1 101    0 36688K  5988K CPU2    2   2:46  94.78% imapd
 1081 jin           1 101    0 36688K  5988K CPU0    0   2:46  94.48% imapd
 1077 jin          41  24    0   767M   481M uwait   3   1:02  34.62% thunderbi
  923 jin           5  20    0   452M   355M uwait   1   0:12   2.10% Xorg
 


     On Saturday, August 15, 2015 10:01 PM, Jin Guojun <jguojun at sbcglobal.net> wrote:
   

 A machine with AMD A8-6600K with Radeon + Nvidia GeForce 210 graphic card.It works with FreeBSD 8.4-R .. 10.1-R well; however, after upgraded to 10.2-R, 
X Window seems having problem in doing configuration and running server.
In X -configure, the xorg.conf.new generated correctly, but have error at the end says "number of screen s does not match number of devices. It found two screens and two devices (see log below), so not sure what is the problem.

When startx, moving window around and scrolling window (using sliding bar) are very sluggish.When exist X window, server generate a non-error error:
[  2700.740] (EE) Server terminated successfully (0). Closing log file.

Does this mean that a Xorg 7.7_2 has some defect?
Does anyone encounter the similar issue in 10.2-Release?
-Jin