PDA

View Full Version : Question about Multi Core Processing reverting to one processor

rcubed
21st February 2008, 11:23
jdobbs,
I was encoding a movie tonight using DVD-RB 2.7.1 and noticed that at some point in the encoding process DVD-RB had reverted to using only a single instance of HC022 during the encode process. I am running an AMS X2 4800+ system with 2G of RAM.

Is this a normal operation? Or caused by some input data structure? I don't think I've ever seen this happen in the past. I'm including the log up to and past the point at which the switch back to a single instance of HC022 occurred. The video was edited to delete FBI warnings etc. using VobBlanker.

********* Start of Log ********

[03:21:10] One Click encoding activated...
-----------------
[03:21:10] Phase I, PREPARATION started.
- DVD-RB v1.27.0
- AVISYNTH 2.5.7.0
- HC v0.22.0.0 encoder selected
- "Steal Space from Extras" mode (10%) is enabled.
- Source: (title deleted by R**3))
- VTS_01: 1,973,655 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 143,585 frames.
-- Building .AVS and .ECL files
- VTS_02: 89,840 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 9,180 frames.
-- Building .AVS and .ECL files
- VTS_03: 71,141 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 6,565 frames.
-- Building .AVS and .ECL files
- VTS_04: 66,971 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 6,437 frames.
-- Building .AVS and .ECL files
- VTS_05: 112,385 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 12,115 frames.
-- Building .AVS and .ECL files
- VTS_06: 67,355 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 5,821 frames.
-- Building .AVS and .ECL files
- VTS_07: 227,159 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 23,705 frames.
-- Building .AVS and .ECL files
- VTS_08: 160,456 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 16,430 frames.
-- Building .AVS and .ECL files
- VTS_09: 167,616 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 17,160 frames.
-- Building .AVS and .ECL files
- VTS_17: 70,525 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 7,785 frames.
-- Building .AVS and .ECL files
- VTS_18: 52,277 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 6,090 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 61.9%
- Overall Bitrate : 3,284/2,627Kbs
- Space for Video : 3,408,664KB
- Redistributing using Base_Q: 35
- Movie improvement from extra reduction = 5.7%
- HIGH/LOW/TYPICAL Bitrates: 4,656/400/2,627 Kbs
[03:28:12] Phase I, PREPARATION completed in 7 minutes.
[03:28:12] Phase II ENCODING started
- Creating M2V for VTS_01 segment 0
- Creating M2V for VTS_01 segment 1
- Creating M2V for VTS_01 segment 2
- Creating M2V for VTS_01 segment 3
- Creating M2V for VTS_01 segment 4
- Creating M2V for VTS_01 segment 5
- Creating M2V for VTS_01 segment 6
- Creating M2V for VTS_01 segment 7
- Creating M2V for VTS_01 segment 8
- Creating M2V for VTS_01 segment 9
- Creating M2V for VTS_01 segment 10
- Creating M2V for VTS_01 segment 11
- Creating M2V for VTS_01 segment 12
- Creating M2V for VTS_01 segment 13
- Creating M2V for VTS_01 segment 14
- Creating M2V for VTS_01 segment 15
- Creating M2V for VTS_01 segment 16
- Creating M2V for VTS_01 segment 17
- Creating M2V for VTS_01 segment 18
- Creating M2V for VTS_01 segment 19
- Creating M2V for VTS_01 segment 20
- Creating M2V for VTS_01 segment 21
- Creating M2V for VTS_01 segment 22
- Creating M2V for VTS_01 segment 23
- Creating M2V for VTS_01 segment 24
- Creating M2V for VTS_01 segment 25
- Creating M2V for VTS_01 segment 26
- Creating M2V for VTS_01 segment 27
- Creating M2V for VTS_02 segment 0
- Extracting Video for VTS_02 segment 1
- Creating M2V for VTS_03 segment 0
- Extracting Video for VTS_03 segment 1
- Creating M2V for VTS_04 segment 0
- Extracting Video for VTS_04 segment 1
- Creating M2V for VTS_05 segment 0
- Extracting Video for VTS_05 segment 1
- Creating M2V for VTS_06 segment 0

********* End of log data *******

I don't know for sure when it happened, but I suspect it may have been at the end of VTS 01 processing. :confused:

At the current time I'm letting the re encode finish up and am going to bed (it's getting late) :) I'm just going to let it run and check it in the morning.

Any enlightenment about what is going on would be appreciated. :thanks:

I can supply additional data from the working file if that will help.

Cheers,

Cuber

PS I noticed that someone all ready pointed out the log shows 1.27.0, The title on the DVD-RB window I am running shows 1.27.1
zacoz
21st February 2008, 12:10
jdobbs reply here (http://forum.doom9.org/showthread.php?p=1100795#post1100795) may apply

If you see only one instance running that usually means the next segment is going to be extracted. I did some tests and found that doing extraction and encoding simultaneously slowed the system down -- so I modified this release to wait until encoding was complete before extracting.
blutach
21st February 2008, 14:23
1.27.1 has nosmp on by default. I believe the next version will have an option to have it off (and also have the extraction thingy off).

Regards
rcubed
21st February 2008, 23:27
jdobbs reply here (http://forum.doom9.org/showthread.php?p=1100795#post1100795) may apply

Zacos, Jdobbs,
Sorry I remember reading that post, but didn't remember it when I posted last night.

Just as an experiment I reran the re encode of the movie I was doing last night and used 1.266, with the same options selected as far as I can tell. The time for 1.266 was 125 min where as the time for 1.271 as 157 min. At least in my case keeping both instances of HC022 active reduced the overall time by more than 1/2 hour. Both are the Pro version of DVD-RB

Jdobbs,
Under what conditions (system) were you running to determine that when extraction is involved, running only one instance was faster? Was that in a true multi core system, or some other configuration. Or are there other factors involved? I guess I'm still confused based on these test results :confused:

I've attached the logs from the two runs. The only difference was is the 1.271 case I had disabled the firewall, the internet connection was disabled, virus checker on demand was disabled, and some other background programs had been terminated, so if anything the 1.266 case might be a little faster had I done the same for that run.

Again any help or info would be appreciated :thanks:

BTW I really like the results from the program and it is my primary method to reduce DL to SL movies.

Thanks in advance,

rcubed
jdobbs
23rd February 2008, 13:51
I think you're right. I've reversed the default for this option in the next release. It seems to speed up some discs/computers, and slow down others.
rcubed
23rd February 2008, 20:23
I think you're right. I've reversed the default for this option in the next release. It seems to speed up some discs/computers, and slow down others.

Thanks for the reply.:thanks:

One follow up question, is it DVD-RB that decides whether or not to run two instances of HC? If so how does the NOSMP option in HC have any effect over whether or not two instances of HC are being run (specifically on a dual or multi core system)?

I'm trying an experiment on a video capture, where I have broken the total move into two halves. I'm currently running two instances of HC started from the GUI, whether or not the SMP option is selected only affects the instance where it is deselected. The fps shown is each slightly lower when two processes are running, but the net gain is almost 2x due to both instances running. (i.e. 1 instance fps avg 23, 2 instances fps avg 22 - on each). Also the slight slow down may be due to the increase use of the video card resources by the fft3dgpu plug in in the avs script.

Thanks in advance for taking the time to respond.:)

rcubed
jdobbs
23rd February 2008, 20:52
If you run multiple instances with v1.27.1 -- SMP is always disabled (in both instances). It is only enabled if you have deselected "Multiple Encoder Processes". In the next release you will also be able to override that with a hidden option.
rcubed
23rd February 2008, 22:57
If you run multiple instances with v1.27.1 -- SMP is always disabled (in both instances). It is only enabled if you have deselected "Multiple Encoder Processes". In the next release you will also be able to override that with a hidden option.

jdobbs,
I guess where I was headed was that in 1.27.1 initially DVD-RB was running two instances of HC, when the encode process got to where extraction started (I'm assuming since I wasn't watching the whole process) but any way at some point DVD-RB only ran one instance of HC022 from that point on until the total encode was finished which slowed down the overall encode. Ver 1.266 did not fall back to one instance of HC022. Is that something that will be optioned, or corrected in the next release?

I guess the basic question is why did 1.271 behave that way? A previous thread had mentioned the SMP/NOSMP option being changed, but that alone should not cause only one instance to be running. In my case the option for multiprocessors was selected. Apparently something else changed from version 1.266 to 1271 to cause the fall back to one instance of HC.

Thanks again for answering:thanks:

rcubed