View Full Version : 1st pass high fps 2nd pass low fps
ny555soul
2nd February 2007, 11:37
I have a C2D E6600 OC'd to 3.0 GHZ and I was averaging 160 fps on 1st pass and 160 fps on 2nd pass using autogk 2.40.For some reason im now averaging 160 fps on 1st pass and 80-100 fps on 2nd pass, and my cpu usage is 95-100%.What would cause such a huge drop in fps?I havent changed any settings and the 30 min test video is the same.
Btw Im using xvid 1.2 with 2 threads.Both cores are being used in task manager with 95-100% usage.
Like i said i havent changed any of the settings but now the fps are dropping alot during 2nd pass.
check
3rd February 2007, 03:50
are you using a more complex avisynth input script?
ny555soul
4th February 2007, 00:04
are you using a more complex avisynth input script?
No i havent touched avisynth.All the settings in autogk are the same as before.Im wondering if it has anything to do with overclocking, but im using cold air to keep temps below 40 degrees C even at full load, so the cpu never got really hot.But even still why would only the 2nd pass slow down?
ny555soul
7th February 2007, 15:48
I wish i had saved my logs, so i could trouble shoot this better, but i do remember one one of the logs saying:
"trying to adjust settings"
"no adjustment possible"
Not sure what it meant, but it was right after the 1st pass.
Thinking that has something to do with it.
BigDid
7th February 2007, 16:53
...Im wondering if it has anything to do with overclocking, but im using cold air to keep temps below 40 degrees C even at full load, so the cpu never got really hot.But even still why would only the 2nd pass slow down?
Hi,
You can monitor the cpu clock in the process manager (Ctrl+Alt+Del) to see if the cpu throttles ...
Edit: my bad monitor with cpuz, not the process manager.
...but i do remember one one of the logs saying:
"trying to adjust settings"
"no adjustment possible"
Not sure what it meant, but it was right after the 1st pass.
Correct, that has to do with the 1st pass which can be redone IF the settings can be adjusted.
If it happens again, post the logfile
OT: 160fps is impressive, I'll have to overclock much more if I want to approach that speed :D /OT
Did
SeeMoreDigital
7th February 2007, 17:03
are you using a more complex avisynth input script?Indeed... Most probably generating encodes at a higher resolution...
The more pixels you encode, the lower the encode speed.
BigDid
7th February 2007, 19:02
Indeed... Most probably generating encodes at a higher resolution...
The more pixels you encode, the lower the encode speed.
Hi,
SMD, he already stated no and it is true in Autogk unless geeks like me use Agkpal :)
Typical script is
autocrop(movie,mode=0,wmultof=4,hmultof=4,samples=10,aspect=new_aspect,threshold=34,samplestartframe=0,leftadd=0,rightadd=0,topadd=0,bottomadd=0)
LanczosResize(out_width,out_height)
RemoveGrain(mode=2)+ colormatrix or leakkerneldeint if used
I just made a quick 2 pass encode of 1 short Vob and had the "Trying to adjust settings" indication and a slower 2nd pass; here is the partial log:
[07/02/2007 05:57:43] Predicted comptest value is: 92,92%
[07/02/2007 05:57:43] Running first pass.
[07/02/2007 05:57:43] Writing the following script to J:\AmelieP\agk_tmp\VTS_01_7 2_movie.avs
...
[07/02/2007 05:59:20] Duration was: 1 minute, 37 seconds
[07/02/2007 05:59:20] Speed was: 91,20 fps.
[07/02/2007 05:59:20] Expected quality of first pass size: 76,08%
[07/02/2007 05:59:21] Trying to adjust settings.
[07/02/2007 05:59:21] No adjustment is possible
[07/02/2007 05:59:21] Running second pass.
[07/02/2007 06:01:36] Duration was: 2 minutes 15 seconds
[07/02/2007 06:01:36] Speed was: 65,58 fps.
[07/02/2007 06:01:36] Job finished. Total time: 4 minutes 50 seconds
Did
ny555soul
8th February 2007, 06:11
I have my C2D e6600 clocked to 3 ghz with a 4:5 ratio for this test
Actually the "trying to adjust settings" log was when my 1st and 2nd pass fps were around 160-170 fps(depending on how high i overclocked)
Now i dont get the "trying to adjust settings" and my 2nd pass is slower.
Here is a more recent log with 2nd pass dropping fps:
[2/3/2007 6:16:28 PM] Source resolution: 544x480
[2/3/2007 6:16:28 PM] Found NTSC source.
[2/3/2007 6:16:28 PM] Source aspect ratio: 4:3
[2/3/2007 6:16:28 PM] Manual aspect ratio: 4:3
[2/3/2007 6:16:28 PM] Color correction enabled.
[2/3/2007 6:16:28 PM] Analyzing source.
[2/3/2007 6:16:59 PM] Source is considered to be interlaced.
[2/3/2007 6:16:59 PM] Output will contain 53928 frames
[2/3/2007 6:16:59 PM] Encoding audio.
[2/3/2007 6:18:56 PM] Audio1 size: 24,408,264 bytes (23.28 Mb)
[2/3/2007 6:18:56 PM] Overhead: 2,157,056 bytes (2.06 Mb)
[2/3/2007 6:18:56 PM] Video size: 217,752,888 bytes (207.67 Mb)
[2/3/2007 6:18:56 PM] Running compressibility test.
[2/3/2007 6:19:36 PM] Duration was: 39 seconds
[2/3/2007 6:19:36 PM] Speed was: 68.40 fps.
[2/3/2007 6:19:36 PM] Compressibility percentage is: 19.13
[2/3/2007 6:19:36 PM] Using softer resizer.
[2/3/2007 6:19:36 PM] Chosen resolution is: 512x384 ( AR: 1.33 )
[2/3/2007 6:19:36 PM] Predicted comptest value is: 19.84%
[2/3/2007 6:19:36 PM] Running first pass.
[2/3/2007 6:25:01 PM] Duration was: 5 minutes 24 seconds
[2/3/2007 6:25:01 PM] Speed was: 165.95 fps.
[2/3/2007 6:25:01 PM] Expected quality of first pass size: 21.02%
[2/3/2007 6:25:01 PM] Running second pass.
[2/3/2007 6:33:56 PM] Duration was: 8 minutes 55 seconds
[2/3/2007 6:33:56 PM] Speed was: 100.71 fps.
[2/3/2007 6:33:56 PM] Job finished. Total time: 18 minutes 11 seconds
BigDid
8th February 2007, 07:04
Hi,
@SMD, My apologies. You were right but in reverse: the less pixels to encode the higher the encode speed :)
@ny555soul
Your encode is out of usual boundaries: you have 21% expected quality with a resolution of 512x384 :scared:
You may want to re-read the tutorial http://www.autogk.me.uk/modules.php?name=TutorialEN , specially the part concerning size, quality and resolution.
I suppose your source is an avi and you want to have your movie or clip in a smaller size. If so let AGK with auto width (not 512 minimum width)
I suppose the resolution will then drop below 400 or even 352 width with the expected quality rising to 50% or more.
Let it finish and compare the 2 encodes then choose what fits you best and give us the encoding speed for the other encode -> the 2nd pass should be lower but near the 1st pass speed ...
Did
check
8th February 2007, 10:07
Temperature could be causing a problem. With core2duo chips, if the temperature hits 70C (or was it 75 or 80?) the clocks will automatically scale down in speed in a manner that won't be picked up by CPUZ or similar. the only program I know of that picks it up is intels Thermal Analysis Advisor (definately a useful tool to have btw).
ny555soul
8th February 2007, 11:53
Temperature could be causing a problem. With core2duo chips, if the temperature hits 70C (or was it 75 or 80?) the clocks will automatically scale down in speed in a manner that won't be picked up by CPUZ or similar. the only program I know of that picks it up is intels Thermal Analysis Advisor (definately a useful tool to have btw).
I use coretemp to read the actual temp of both cores.
My temps never go above 46C on either core even at 100% cpu usage even overclocked to 3.3 ghz with a 1:1 ratio.
@ BigDid
Source is TS (mpeg2) 544x480 i have autogk set to fixed 512 to keep 4:3 ratio and to keep bitrate above 1000 kbs to fit 90 minutes on a cd.But that was a 15 minute length test clip.Custom size of 117 MB
I did notice on the last test i did, during 2nd pass cpu usage went down to 88-93% where 1st pass was a consistent 97-100%, but coretemp never went above 46C and actually dropped to 43C during 2nd pass.
I dont think my cpu throttled at those low temps.
Heres another test @ 3.3 ghz 1:1 ratio :
[2/8/2007 5:26:13 AM] Source resolution: 544x480
[2/8/2007 5:26:13 AM] Found NTSC source.
[2/8/2007 5:26:13 AM] Source aspect ratio: 4:3
[2/8/2007 5:26:13 AM] Color correction enabled.
[2/8/2007 5:26:13 AM] Analyzing source.
[2/8/2007 5:26:27 AM] Source is considered to be interlaced.
[2/8/2007 5:26:27 AM] Output will contain 26954 frames
[2/8/2007 5:26:27 AM] Encoding audio.
[2/8/2007 5:27:20 AM] Audio1 size: 12,139,536 bytes (11.58 Mb)
[2/8/2007 5:27:20 AM] Overhead: 1,078,144 bytes (1.03 Mb)
[2/8/2007 5:27:20 AM] Video size: 109,465,712 bytes (104.39 Mb)
[2/8/2007 5:27:20 AM] Running compressibility test.
[2/8/2007 5:27:49 AM] Duration was: 28 seconds
[2/8/2007 5:27:49 AM] Speed was: 74.39 fps.
[2/8/2007 5:27:49 AM] Compressibility percentage is: 20.69
[2/8/2007 5:27:49 AM] Using softer resizer.
[2/8/2007 5:27:49 AM] Chosen resolution is: 512x384 ( AR: 1.33 )
[2/8/2007 5:27:49 AM] Predicted comptest value is: 21.46%
[2/8/2007 5:27:49 AM] Running first pass.
[2/8/2007 5:30:18 AM] Duration was: 2 minutes 29 seconds
[2/8/2007 5:30:18 AM] Speed was: 180.31 fps.
[2/8/2007 5:30:18 AM] Expected quality of first pass size: 22.82%
[2/8/2007 5:30:18 AM] Running second pass.
[2/8/2007 5:34:26 AM] Duration was: 4 minutes 7 seconds
[2/8/2007 5:34:26 AM] Speed was: 108.80 fps.
[2/8/2007 5:34:26 AM] Job finished. Total time: 8 minutes 27 seconds
==============================================
MadMatty
14th February 2007, 01:33
I have same issue with my amd X2. 1st pass uses 100% cpu on both cores and 2nd pass is quite eratic but generaly used about 90% across both cores. C&Q is dissabled. I have a fresh xp install and a fresh autogk 2.40 install and pretty sure it didnt happen a few versions ago but cant be sure.
My source is 720x576 freeview footage. usualy 45 min tv episodes. minimum width set to 640 and vbr 192 mp3 and a target 350mb file size. Accurate colour fixing, sharp mpeg, and ess standalone support.
[08/02/2007 12:09:16] AutoGK 2.40
[08/02/2007 12:09:16] OS: WinXP (5.1.2600).2
[08/02/2007 12:09:16] Job started.
[08/02/2007 12:09:16] Input file: K:\Vids\PrisonBreakS2\Episode4.mpg
[08/02/2007 12:09:16] Output file: K:\Vids\PrisonBreakS2\Episode4.avi
[08/02/2007 12:09:16] Output codec: XviD
[08/02/2007 12:09:16] Audio 1: Audio Stream 0 MPEG
[08/02/2007 12:09:16] Subtitles: none
[08/02/2007 12:09:16] Format: AVI
[08/02/2007 12:09:16] Target size: 350Mb
[08/02/2007 12:09:16] Custom resolution settings: minimum width of 640 pixels
[08/02/2007 12:09:16] Audio 1 settings: VBR MP3 with average bitrate: 192Kbps
[08/02/2007 12:09:16] Standalone support enabled: MTK/Sigma
[08/02/2007 12:09:16] Started encoding.
[08/02/2007 12:09:16] Demuxing and indexing.
[08/02/2007 12:09:26] Processing file: K:\Vids\PrisonBreakS2\Episode4.mpg
[08/02/2007 12:09:26] Source resolution: 720x576
[08/02/2007 12:09:26] Found PAL source.
[08/02/2007 12:09:26] Source aspect ratio: 16:9
[08/02/2007 12:09:26] Color correction enabled.
[08/02/2007 12:09:26] Analyzing source.
[08/02/2007 12:11:55] Source is considered to be progressive.
[08/02/2007 12:11:55] Output will contain 61713 frames
[08/02/2007 12:11:55] Encoding audio.
[08/02/2007 12:15:33] Forcing encoding with sharp matrix.
[08/02/2007 12:15:33] Audio1 size: 54,915,024 bytes (52.37 Mb)
[08/02/2007 12:15:33] Overhead: 2,468,480 bytes (2.35 Mb)
[08/02/2007 12:15:33] Video size: 309,618,096 bytes (295.27 Mb)
[08/02/2007 12:15:33] Running compressibility test.
[08/02/2007 12:16:34] Duration was: 1 minute, 1 second.
[08/02/2007 12:16:34] Speed was: 50.16 fps.
[08/02/2007 12:16:35] Compressibility percentage is: 55.18
[08/02/2007 12:16:35] Chosen resolution is: 640x368 ( AR: 1.74 )
[08/02/2007 12:16:35] Predicted comptest value is: 55.18%
[08/02/2007 12:16:35] Running first pass.
[08/02/2007 12:30:23] Duration was: 13 minutes 48 seconds
[08/02/2007 12:30:23] Speed was: 74.48 fps.
[08/02/2007 12:30:23] Expected quality of first pass size: 56.62%
[08/02/2007 12:30:23] Running second pass.
[08/02/2007 12:49:37] Duration was: 19 minutes 14 seconds
[08/02/2007 12:49:37] Speed was: 53.47 fps.
[08/02/2007 12:49:37] Job finished. Total time: 40 minutes 21 seconds
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.