1st vs 2nd pass on dual core CPUs

[Bindibadgi]

Im using AutoGK which uses vdub mod to encode xvid of a short clip. Im trying to analyse an Intel system with a 2.8d, where basically the first pass loads both CPUs and is exceptionally fast, but the second pass is significantly slower and only loads the CPUs 70-75%. Is the second pass more memory bandwidth dependent? Im trying to determine the bottleneck of the system, and im not out of memory, or am I loading the pagefile and it has CPU cycles to spare!?

For the same clip and same settings:
3.4EE - 24mins 8 secs
3.73EE - 20mins 33 secs
2.8d - 21mins 54 secs

The most part of the time is the first and second passes (over 2/3rds of the time), and with both the 2.8 CPUs using the same memory, is that why is isnt significantly faster?

[Buzz Lightyear]

I would say this is the wrong section for your question, but anyway.
See http://forum.doom9.org/showthread.php?t=87715

In short form: None of the programs GK/AutoGK uses to encode are made for SMP. The only reason why both CPUs get used, is that one CPU runs avisynth and the other XviD.
So it all depends on the XviD/avisynth settings you use. If (by luck) both programs use the same amount of CPU-time you get nearly 100% CPU usage, but if one of the programs runs much slower/faster than the other then you get between 55%-90% CPU usage.
That also explains the difference between first/second pass.

[Bindibadgi]

There's SOOO many sections here it's hard to know where to put things. I thought it was a vdub thing, because vdub was running at the time, but if what you say is true then ill look into avisynth and xvid further. Ta

[708145]

Im using AutoGK which uses vdub mod to encode xvid of a short clip. Im trying to analyse an Intel system with a 2.8d, where basically the first pass loads both CPUs and is exceptionally fast, but the second pass is significantly slower and only loads the CPUs 70-75%. Is the second pass more memory bandwidth dependent? Im trying to determine the bottleneck of the system, and im not out of memory, or am I loading the pagefile and it has CPU cycles to spare!?

For the same clip and same settings:
3.4EE - 24mins 8 secs
3.73EE - 20mins 33 secs
2.8d - 21mins 54 secs

The most part of the time is the first and second passes (over 2/3rds of the time), and with both the 2.8 CPUs using the same memory, is that why is isnt significantly faster?

I suggest you keep an eye on ELDER development as it uses dual core machines to the max!
But neither very user friendly nor complete yet I admit :rolleyes:

http://forum.doom9.org/showthread.php?t=100766

bis besser,
Tobias