2011/10/8 Feng Wei feng.wei@linaro.org:
I test the dts with 'time avconv -i test.dts -f null -', i.mx53 still cost more than 20s. Following is mx53's cpuinfo root@linaro-desktop:/home/linaro# cat /proc/cpuinfo Processor : ARMv7 Processor rev 5 (v7l) BogoMIPS : 999.42 Features : swp half thumb fastmult vfp edsp thumbee neon vfpv3 CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x2 CPU part : 0xc08 CPU revision : 5
Hardware : Freescale MX53 LOCO Board Revision : 53020 Serial : 0000000000000000
Is there any difference in co-processor between beagle and mx53?
Although both are Cortex-A8, the Beagle-xm has core revision r3p2 while the mx53 has r2p5, but this should not cause much of a difference.
The times you've reported, are they "real" or "user" times? Did you also have a pandaboard?
2011/10/8 Mans Rullgard mans.rullgard@linaro.org:
2011/10/8 Feng Wei feng.wei@linaro.org:
I test the dts with 'time avconv -i test.dts -f null -', i.mx53 still cost more than 20s. Following is mx53's cpuinfo root@linaro-desktop:/home/linaro# cat /proc/cpuinfo Processor : ARMv7 Processor rev 5 (v7l) BogoMIPS : 999.42 Features : swp half thumb fastmult vfp edsp thumbee neon vfpv3 CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x2 CPU part : 0xc08 CPU revision : 5
Hardware : Freescale MX53 LOCO Board Revision : 53020 Serial : 0000000000000000
Is there any difference in co-processor between beagle and mx53?
Although both are Cortex-A8, the Beagle-xm has core revision r3p2 while the mx53 has r2p5, but this should not cause much of a difference.
The times you've reported, are they "real" or "user" times?
real time
Did you also have a pandaboard?
yes, the result is about 3.4s, maybe SD card read speed cause the difference from yours.
-- Mans Rullgard / mru
2011/10/10 Feng Wei feng.wei@linaro.org:
2011/10/8 Mans Rullgard mans.rullgard@linaro.org:
The times you've reported, are they "real" or "user" times?
real time
What numbers do you get for user time?
Did you also have a pandaboard?
yes, the result is about 3.4s, maybe SD card read speed cause the difference from yours.
I get very little difference between real and user time after the first run when everything is hot in memory cache.
I found a L2 cache disabled bug in linaro release on i.mx53, and after fixed it, the result is really good. "time avconv -i test.dts -f null -" cost about 3.12s on i.mx53loco
2011/10/10 Mans Rullgard mans.rullgard@linaro.org:
2011/10/10 Feng Wei feng.wei@linaro.org:
2011/10/8 Mans Rullgard mans.rullgard@linaro.org:
The times you've reported, are they "real" or "user" times?
real time
What numbers do you get for user time?
Did you also have a pandaboard?
yes, the result is about 3.4s, maybe SD card read speed cause the difference from yours.
I get very little difference between real and user time after the first run when everything is hot in memory cache.
-- Mans Rullgard / mru
2011/10/28 Feng Wei feng.wei@linaro.org:
I found a L2 cache disabled bug in linaro release on i.mx53, and after fixed it, the result is really good. "time avconv -i test.dts -f null -" cost about 3.12s on i.mx53loco
I guess that settles the matter.
linaro-multimedia@lists.linaro.org
-
Feng Wei -
Mans Rullgard