Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > MPEG-2 Encoding

Reply
 
Thread Tools Search this Thread Display Modes
Old 3rd December 2003, 20:16   #1  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
A real CCE SpeedTest

The point of this exercise is to create meaningful comparisons of CCE encoding speeds. Many of us are tired of the nonsense posts and waste of bandwidth. Although this test is primitive and will reflect a speed much slower than you will actually get using proper source, it should allow an exact comparison of speed on various platforms. In other words, two machines of the exact same configuration should return identical results.

This assumes I have thought it out correctly and did not screw up If everybody uses the same source and the same ecl to ensure the same settings, it seems to me this ought to work. Feedback welcome if I missed something.

Instructions (follow exactly):

1> Create directory on your HD as "CCESpeedTest01"
2> Copy and paste AVS file below into notepad. Save into the above directory as "CCESpeedTest01.avs"
3> Copy and paste the two ECL files into notepad one at a time. Save each as "CCESpeedTest01_CCE2.5.ecl" and "CCESpeedTest01_CCE267.ecl"
4> Edit the ecl files using the replace function of notepad. Search and replace "G:\" to the HD drive letter where you created the dir in step 1 above.
5> Run CCE and load the appropriate ECL file. Make NO changes. Press encode and note the speed after exactly 3 minutes of encoding.
6> Report that speed along with the information on your machine.
Code:

#start of CCESpeedTest01.avs
ColorBars (720,480)
ShowSMPTE(24)
converttoyuy2()
#end 
CCE2.5x ECL
Code:

; Cinema Craft Encoder SP -- Encoder Control List

[item]
title=CCESpeedTest01.avs
vid_out=1
vaf_out=0
aud_out=0
vid_file0=G:\CCESpeedTest01\CCESpeedTest01_CCE2.5.mpv
vid_file1=
vaf_file=
aud_file=
file_focused=0
encode_mode=0
packet_size=4096
timecode=0x1000000
width=720
height=480
vmode=0
frame_rate_idx=4
cbr_brate=6000
opv_q_factor=32
opv_brate_min=300
opv_brate_max=8000
vbr_brate_avg=2257
vbr_brate_min=1194
vbr_brate_max=2565
vbr_bias=30
vbr_pass=3
quality_prec=17
use_filter=0
filter_val=6
seq_endcode=1
dvd=0
half_width=0
half_height=0
fast_mode=0
progressive=1
alternate_scan=0
non_linear=1
top_first=0
lum_level=1
intra_dc_prec=-1
aspect_ratio=2
gop_m=3
gop_nm=5
gop_hdr=15
seq_hdr=1
all_closed_gop=0
fix_gop_length=0
mpeg1=0
mpeg1_cps=0
samples_per_sec=48000
stereo=2
brate_idx=7
crc=1

[file]
name=G:\CCESpeedTest01\CCESpeedTest01.avs
type=0
frame_first=0
frame_last=107892
encode_first=0
encode_last=107892
CCE 2.67 ECL
Code:

; CCE SP Trial Version -- Encoder Control List

[item]
title=CCESpeedTest01.avs
vid_out=1
vaf_out=0
aud_out=0
aud_mode=0
vid_file0=G:\CCESpeedTest01\CCESpeedTest01_CCE267.mpv
vid_file1=
vaf_file=
aud_file=
file_focused=0
video_type=2
width=720
height=480
timecode=0x1000000
tc_ref_frm=0
drop_frame=0
cbr_brate=6000
opv_q_factor=32
opv_brate_min=300
opv_brate_max=8000
vbr_brate_avg=2230
vbr_brate_min=300
vbr_brate_max=2530
vbr_bias=30
vbr_pass=3
create_new_vaf=0
pulldown_detect=0
letter_box=0
panscan=0
aspect_ratio=2
credits_tweak=0
credits_start=0x00000
credits_brate=1000
h_filter=0
h_filter_idx=8
use_filter=0
filter_val=4
dither=0
dither_max=8
quality_prec=25
intra_dc_prec=2
alternate_scan=0
non_linear=1
progressive=1
gop_m=3
gop_nm=5
gop_hdr=15
seq_hdr=1
seq_endcode=1
dvd=0
all_closed_gop=0
fix_gop_length=0
fix_vbv_delay=0
offset_line=0
half_width=0
half_height=0
uscc=0
lum_level=1
packet_size=2048
frame_rate_idx=4
qmat_idx=0
mpeg1=0
mpeg1_cps=1
brate_idx=7
stereo=2
samples_per_sec=48000
crc=1

[file]
name=G:\CCESpeedTest01\CCESpeedTest01.avs
type=0
frame_first=0
frame_last=107892
encode_first=0
encode_last=107892

Last edited by DDogg; 4th December 2003 at 14:31.
DDogg is offline   Reply With Quote
Old 3rd December 2003, 21:23   #2  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
My tests:

CCE 2.5x 1.50
CCE 2.67 1.41

[Edited - Better speed after reboot]

Biostar M7NCDP | XP1700 (AXDA1700DUT3C - JIUHB 0312VPMW) @2.1 Ghz (11*192) VCore 1.70 v - SiSoft PR 3064 (Est.)| GFrce 2 MX400 | 4 Maxtor 40g| Zlmn Flower w/ 120 fan @ 2000 rpm| Enermax 330 watt r82 tweak | (2) 256 meg PC3500 HYP_X (7-2-2-2 VDim 2.8 v| Enermax CS 307 Case

Last edited by DDogg; 4th December 2003 at 18:15.
DDogg is offline   Reply With Quote
Old 4th December 2003, 03:01   #3  |  Link
KeyserSoze
Registered User
 
Join Date: Nov 2003
Location: Greece
Posts: 27
I'm getting the following error:

D:\CCESpeedTest01\CCESpeedTest01_CCE2.5.ecl is not a ECL for Cinema Craft Encoder SP."

I'm getting the same with CCESpeedTest01_CCE267.ecl file!!
__________________
AMD Athlon 64 3200+,Asus K8N,2x256Mb Kingston CL2.5(Winbond) PC3200 DDR,ASUS ÁX800 XT Platinum Edition ,17" LCD Sony SDM-HS74P/B ,Plextor DVD+R PX-716A ,Pioneer DVD-121,80Gb WD Caviar WB800JB , 120Gb WD Caviar WB1200JD SATA ,Creative Inspire 5500 Digital 5.1,Thermaltake Tsunami Case,Antec TruePower 430W,Zalman 7000B-Cu,Arctic Cooling ATI Silencer 4

Last edited by KeyserSoze; 4th December 2003 at 03:03.
KeyserSoze is offline   Reply With Quote
Old 4th December 2003, 04:28   #4  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
Sure thought they should load OK, crap. Well just set either of them up for One Pass VBR Q32 Min 300 Max 8000 and no VAF creation. Set Zigzag and Progressive on. No internal filters engaged of any kind. Make sure DVD compliant is not on. That should work, but without loading the ECL's we can't consider it apples to apples.

What exact version/s of CCE are you trying to use?

Has anybody else got them to work ok?

Last edited by DDogg; 4th December 2003 at 04:32.
DDogg is offline   Reply With Quote
Old 4th December 2003, 08:38   #5  |  Link
coona
Registered User
 
Join Date: Mar 2002
Location: Prague
Posts: 358
Hi DDogg,

I really appreciate your work . Solution (as usual ) is very simple - ECL file has to start with this line:

; Cinema Craft Encoder SP -- Encoder Control List

My result is:

CCE 2.5x 0.391

I´ll post my configuration later because I´m at work and I have to check what hardware I really use here, hehe.

***EDITED

Asus CUSL2-M, PIII 866, 512 MB SDRAM, Ati Rage XL AGP,1x IBM 60 GB, 1x IBM 75 GB, Toshiba SD-M1502

***EDITED

Last edited by coona; 4th December 2003 at 09:11.
coona is offline   Reply With Quote
Old 4th December 2003, 14:40   #6  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
jarvis1781, did you use the ECL above? That speed surprised me.
DDogg is offline   Reply With Quote
Old 4th December 2003, 14:53   #7  |  Link
KeyserSoze
Registered User
 
Join Date: Nov 2003
Location: Greece
Posts: 27
2.67.00.10---->1.42

2.50.01.00---->1.514


XP2400+ at 2000Mhz(12x166,harmless o/c ,Vcore 1.65) Dual Channel Winbond CH-5 (5-2-2-2,VDim 2.6v)

CCE 2.67 is ofcourse optimised for P4 CPUs, and CCE 2.5 for Athlons .

(no background apps this time)
__________________
AMD Athlon 64 3200+,Asus K8N,2x256Mb Kingston CL2.5(Winbond) PC3200 DDR,ASUS ÁX800 XT Platinum Edition ,17" LCD Sony SDM-HS74P/B ,Plextor DVD+R PX-716A ,Pioneer DVD-121,80Gb WD Caviar WB800JB , 120Gb WD Caviar WB1200JD SATA ,Creative Inspire 5500 Digital 5.1,Thermaltake Tsunami Case,Antec TruePower 430W,Zalman 7000B-Cu,Arctic Cooling ATI Silencer 4

Last edited by KeyserSoze; 4th December 2003 at 15:33.
KeyserSoze is offline   Reply With Quote
Old 4th December 2003, 18:04   #8  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
KeyserSoze, were you able/how did you get the ecls to load? Note your speed and mine are nearly exact, so we can see the benchmark works and returns similar results on like machines.

Last edited by DDogg; 4th December 2003 at 18:20.
DDogg is offline   Reply With Quote
Old 4th December 2003, 18:55   #9  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
Quote:
forgot to note that my CPU is OC'd to 3.5
yeah, that would make a bit of difference
DDogg is offline   Reply With Quote
Old 5th December 2003, 00:42   #10  |  Link
KeyserSoze
Registered User
 
Join Date: Nov 2003
Location: Greece
Posts: 27
Quote:
Originally posted by DDogg
KeyserSoze, were you able/how did you get the ecls to load? Note your speed and mine are nearly exact, so we can see the benchmark works and returns similar results on like machines.
Your old version of scripts had that extra line on the top.....
Those didn't work in CCE2.5 but worked in 2.67(EclCCE).
Now everything is ok on both scripts.
I just followed coona's advice
__________________
AMD Athlon 64 3200+,Asus K8N,2x256Mb Kingston CL2.5(Winbond) PC3200 DDR,ASUS ÁX800 XT Platinum Edition ,17" LCD Sony SDM-HS74P/B ,Plextor DVD+R PX-716A ,Pioneer DVD-121,80Gb WD Caviar WB800JB , 120Gb WD Caviar WB1200JD SATA ,Creative Inspire 5500 Digital 5.1,Thermaltake Tsunami Case,Antec TruePower 430W,Zalman 7000B-Cu,Arctic Cooling ATI Silencer 4

Last edited by KeyserSoze; 5th December 2003 at 18:40.
KeyserSoze is offline   Reply With Quote
Old 5th December 2003, 15:33   #11  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
OK, thanks. As you noticed I edited out those extra lines in the ECLs above. Now everybody should be able to just copy and paste them into notepad and save them. The only edit needed is to change "g:\" to the drive they use.

The benchmark seems to work well now. Let's see some more testing results, people!
DDogg is offline   Reply With Quote
Old 5th December 2003, 17:31   #12  |  Link
chipvideo
Registered User
 
Join Date: Nov 2002
Posts: 150
Doesn't do smp very well with this test. I get 58% on my processors. While I do the dvd2dvd process with the highest bicubic resize method I get 72% on average usage.
chipvideo is offline   Reply With Quote
Old 5th December 2003, 18:19   #13  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
chipvideo, the data we are interested in was what speed you obtained. Once we get enough of that data we can then develop a factor to convert the result to one that more closely indicates real world encoding. Keep in mind that this method removes avisynth and mpeg2decX from the equation, as well as resolution/aspect ratio.

In your case, this shows CCEs processor usage alone. It may be that jarvis1781's single processor machine will actually encode faster than yours. If so, that is useful data to know (although not something you may particularly appreciate )

Last edited by DDogg; 5th December 2003 at 18:33.
DDogg is offline   Reply With Quote
Old 5th December 2003, 19:47   #14  |  Link
chipvideo
Registered User
 
Join Date: Nov 2002
Posts: 150
Well I doubt that he can encode 70% faster than my machine. He would have to encode a dvd with bicubic resize at around RT5.3 on Titanic.

I know that your machine isnt the same speed as mine for encoding. I have my procs OC'd to 2400mhz each. LIke I said this test shows CCE using only 58% processor while using the real method shows a CCE using 72%.

That is all I am saying. Based on this test one would assume that a dual processor system is a waste and doesnt perform any better than a single one when in fact a dual system enocdes much faster.
chipvideo is offline   Reply With Quote
Old 5th December 2003, 20:05   #15  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
That's interesting, but as before we don't have your speed on this test so as to put it into perspective.
Quote:
when in fact a dual system encodes much faster.
I don't have experience with dual processor systems. However, I do wonder if what you said above is always so. My understanding was that a single instance of CCE would not use the capacity of both processors. Is that correct? If so, could not a single very fast processor like jarvis1781's potentially out perform your slower duals?

Last edited by DDogg; 5th December 2003 at 20:16.
DDogg is offline   Reply With Quote
Old 5th December 2003, 21:30   #16  |  Link
chipvideo
Registered User
 
Join Date: Nov 2002
Posts: 150
My speed was 1.51 using your test. If cce didnt utilize the both processors then I would get a max of 50% on cce. I get 58% with cce using your test and 72% when using the dvd2dvd method.

His single processor is not beating my dual encodes by 70% encoding a dvd material. You must be crazy to think that your system is as fast as mine as well. But it isn't. I think a better way to figure this out would be for a few of us to use the dvd2dvd method and use the same settings. Bicubic resize and use ntsc source. Maybe we could make a list of some movies that people have and then see if we can get some results.

I have Titanic, Saving Private Ryan and that other hanks movie on the Island. Those are good movies as they are long ones and would show that true power of the system. I mean I get 3.3RT with Titanic. I highly doubt your processor running at much lower than mine is goign to do as well. I know this is a early stage test and I am glad your willing to take the time, but we need to get more realife results.
chipvideo is offline   Reply With Quote
Old 5th December 2003, 22:20   #17  |  Link
Kedirekin
Registered User
 
Join Date: Oct 2001
Location: Minnesota
Posts: 1,110
I feel compelled to comment, as I've spent a fair amount of time thinking about multiprocessor (MP) systems.

This test isn't just testing CCE encoding speed, it's also testing the speed of AviSynth.

On a single processor system, the script might be using something like 86% of processor cycles, leaving the other 14% for CCE.

On a dual processor system, all of one processor (100%) can be dedicated to the script. All of the left over processor could be dedicated to CCE, but the script is only capable of delivering data so fast, so CCE only ends up using 16%.

The apparent result is a 15% increase in CCE encoding speed, but the test is somewhat biased by a slow script. With a more efficient script, the ratios are completely different, and tend more and more in MP's favor.

Just a point of clarification, CCE is MP enabled. The reason you don't see 100% processor usage is that AviSynth is not MP enabled.

A more fair test of CCE sans frame server might be to use the script to create a Huffy (or raw) avi file, then encode from that. That would really show off the advantages of MP systems. Unfortunately that is misrepresentative of what you're really trying to test (frame server+encoder performance) in the other direction.

BTW: why is that script so slow? I wouldn't think ColorBars or ShowSMPTE would slow the script down so much.

Last edited by Kedirekin; 5th December 2003 at 22:26.
Kedirekin is offline   Reply With Quote
Old 5th December 2003, 23:09   #18  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
Quote:
You must be crazy to think that your system is as fast as mine as well.
My goodness, chipvideo I would never think that. What I am attempting here is a reliable method of comparing CPU platform to CPU platform that removes from the equation issues like NTSC/PAL, aspect ratio/total encoded pixels, variant of mpeg2dec and IDCT, color conversion, etc., etc., etc., ad nauseum.

This is only the first stage of that process. Typically I normally have a bit of method in my madness. Over the years I've found that the only way to accomplish something on this board is to cause a challenge scenario. It is just the way people are. Its neither good nor bad, it just is. So thanks for the input.

Perhaps you and/or Kedirekin can make a major contribution so we can actually have an intelligent method of reportage instead of the silliness of "WoW, bubba, I got 9 x realtime!!!! er, 'course that was 160x120 resolution, but who cares cause 9x makes me feel like I have a big Johnson I suspect this irritates you, and especially Kedirekin, who knows more about this stuff than most will ever learn.

I think you are completely correct about this particular test as it not only concerns multi processors, but the general full process itself. I would wholeheartedly concur with a more proper method as you suggested that uses real source. When I have suggested something similar (repeatedly over the years) the response has been underwhelming. As said, this is the first stage, and it does supply a legitimate method of comparing [single ] CPU systems. In other words, a machine that returns 1.25 will be proportionately slower than a machine that returns 1.55. What that proportion is is still up in the air at the moment. For that reason alone I think this method is useful and serves a strong purpose.

Now we just need to work together and come up with the next iterations. However, that method must be transparent to NTSC and Pal, else it is a waste of time. Nobody wants a dual benchmark, IMO. This gets us into a easily downloadable common source and the required bandwidth which has always been one of the hangups.

So, you want to help or just bitch at me? (I really am grinning)

So now
Quote:
BTW: why is that script so slow?
That always puzzled me too. Maybe because it is so static. Once...it was either Dividee or Steady wrote this brilliant avs script for me that better mimicked a real video source. I don't know what came of it. I think that was on the old EZboard which kinda tells you how long I have wanted a decent benchmark.

Last edited by DDogg; 5th December 2003 at 23:25.
DDogg is offline   Reply With Quote
Old 5th December 2003, 23:40   #19  |  Link
Kedirekin
Registered User
 
Join Date: Oct 2001
Location: Minnesota
Posts: 1,110
Actually I think your proposed test is perfectly valid for SP systems. I wasn't trying to denigrate it. It may not represent real-world performance, but as an apples-to-apples test is should be accurate. As you said, the relative performance differences should track into real-world circumstances.

It's probably valid for MP systems as well, but only for comparing to other MP systems.

Unfortunately things break down when trying to compare SP and MP systems, and it isn't really the fault of the test; the breakdown is present in real-world situations as well. For example, your MP setup could encode one project 40% faster than a comparable SP system, and the next project might only be 10% faster.

Taking that into account, I don't think it is possible to create a test that will ever allow apples-to-apples comparisons between SP and MP systems. It's kind-of sad really - if everything were MP enabled, the MP system would always be 85-95% faster than the comparable SP system.

BTW: chipvideo, if you want to see how much added oomph your MP system gives you, try setting your CCE processor affinity to only use one processor.

Last edited by Kedirekin; 5th December 2003 at 23:45.
Kedirekin is offline   Reply With Quote
Old 6th December 2003, 01:08   #20  |  Link
DDogg
Retired, but still around
 
DDogg's Avatar
 
Join Date: Oct 2001
Location: Lone Star
Posts: 3,058
When we get more data, that is, when we can get a few more people to get motivated, run the test and report the results, we can then pick a few folks to also run an encode on a common DVD source. [Hey, it only takes five minutes!]

That data should establish the proportion of this artificial benchmark to a real world encoding job. Ultimately I think (hope) we can end up with a multiplier to apply against the benchmark number that will yield a close approximation to real world speed using the common encoding method of mpeg2dec3/avisynth. Hopefully that would also use, as part of the factor, the amount of pixels to be encoded so we can take into account aspect ratio/res. I'm still fuzzy on that. Most folks know I am a math dropout.

There is something else tickling the back of my brain but I can't get a handle on it yet I think it is something to do with pixels and some way of a PAL or NTSC person computing equivalence. May be clueless.

Last edited by DDogg; 6th December 2003 at 01:21.
DDogg is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 23:16.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.