A real CCE SpeedTest

[LB]

Originally posted by Piaf
cce 2.50 TEST
Hi, heres my test for the 2.50 script,looks good it seems?!:)

http://people.zeelandnet.nl/rpjongepier/cce250_speedtest.jpg
system:
--------
Intel Pentium 4 3.0 Ghz met 800 MHz FSB
512 MB DDR400 dual channel geheugen
ATI Radeon 9800 XL 128 MB SDRAM
info:/http://www.hardware.info/reviews.php?id=435

greets piaf

Hmm, you did something wrong. Nothing goes that fast using this script. Even those with 3.2Ghz P4s are getting around 1.58... so I think you need to do your test again.

[Piaf]

lol :D my bad,updated my post,getting 1.265 now,damn that 4.xx fps :D

[GooglyBear]

I might as well contribute to this thread considering i just finished upgrading my pc

http://allwegot-is.us/imgz/funstuff/HomeTheatre/cce-speedtest1.jpg

configuration:
Athlon 64 3000+, 1GB DDR, Seagate SATA

oddly enough I swear it hit above 3 last night when I ran 5pass on 28days..

real life applications (using dvd-rb, cce5pass, i only have 267):
my old duron 800: 25+ hours
my old athlon 1ghz: 17+ hours
last night's run: 7+ hours to completion :)

[auenf]

Originally posted by GooglyBear
I might as well contribute to this thread considering i just finished upgrading my pc

[IMG]

configuration:
Athlon 64 3000+, 1GB DDR, Seagate SATA

oddly enough I swear it hit above 3 last night when I ran 5pass on 28days..

real life applications (using dvd-rb, cce5pass, i only have 267):
my old duron 800: 25+ hours
my old athlon 1ghz: 17+ hours
last night's run: 7+ hours to completion :)

this script is basically a torture test, so you probably were getting 3x on a real source.

Enf...

[MyNameIsNeo]

Originally posted by GooglyBear
I might as well contribute to this thread considering i just finished upgrading my pc

http://allwegot-is.us/imgz/funstuff/HomeTheatre/cce-speedtest1.jpg

configuration:
Athlon 64 3000+, 1GB DDR, Seagate SATA

oddly enough I swear it hit above 3 last night when I ran 5pass on 28days..

real life applications (using dvd-rb, cce5pass, i only have 267):
my old duron 800: 25+ hours
my old athlon 1ghz: 17+ hours
last night's run: 7+ hours to completion :)

Oh my f*cking god.
How the hell do you get 1.58? I'm confused. I believe mine is suppsed to be a bit faster. Is it the RAID Configuration that is making the difference.
I have an Athlon 64 3200+ and my CCE 2.67.00.23 is 1.32@3.00 minutes
and CCE 2.50 is 1.400@3.00 minutes

Athlon 64 3200+
Gigabyte K8npro (nforce3 150)
WD 80GB and WD 160GB HDD with 8mb buffer (no raid)
2x512MB GeiL DDR 400(Value series 2.5 6-3-3 with heatspreader)

Neo

[onesoul]

CCE 2.50.01.00: 1.568
CCE 2.67.00.27: 1.47

Motherboard: Asus A7N8X-Deluxe(rev1.04), Cpu : AMD Athlon xp2100+(AIHUB) @2070MHZ, FSB:180MHZ Synched with memory, HDD: 2*80GB WD IDE RAID0 with Rocket Raid100 PCI Card, Graphic Card : GeForce3 TI200 64MB, Monitor : Samtron 19" 96BDF, Memory : 2*256 DDR400 Corsair dual channel @180FSB timings:2,2,2,5, Mouse : Microsoft Intellimouse optical, Mousepad : Icemat Black Clanbase

Why do I see a tenuous vertical white line next to the yellow bar with cce 2.67 moving left-right? I don't see this happen with cce 2.5! Could be the reason of cce 2.67 being optimized for encoding motion? :confused:

Cheers

[shoarthing]

CCE-SP 2.67.00.23

1.58

Dual Barton Mobiles @ 15x 160MHz
1GB ECC unbuffered, set 'correct & scrub'

[br408408]

try 2.5, it runs faster on AMD's

[cypher_soundz]

CCE 2.5 : 0.831 (with loads of thing open)
CCE 2.5 : 0.848 (after reboot)
system spec below

[shoarthing]

cypher_soundz - Hi - is this at 13x 100 [presumably so]? . . . in my experience w/ the MPX chipset, CCE performance is overwhelmingly linked to FSB [then, to a lesser extent, DDR timings] . . . so perhaps interesting to try your hacked Durons at 9x 140 [say] & see how this 'speed test' measure compares: I'd be amazed if it didn't improve quite markedly

FWIW I've tried Palomino's, TBreds, Bartons, & Mobile Bartons in Iwill MPX2's at multipliers from 9-17x, FSB's from 133-167, & gross GHz from 1.2-2.5 . . . but never [alas] Durons . . . . the extra cache of Bartons appears to make no difference whatsoever to CCE encoding speeds.

[cypher_soundz]

Cool, thanks for the tip, i'm not really in tune with overclocking (or under clocking ) could you run through the process? i tried my bios settings but to no avail.
Regards
cyph

[shoarthing]

Hi - I don't know whether your Duron has 4x or 5x L1 bridges: if the former, this link

http://www.ocinside.de/go_e.html?/html/workshop/socketa/socketa_resistors.html - should be relevant;

if the latter [more probable],

http://www.ocinside.de/index_e.html?/html/workshop/socketa/xp_painting.html

- the idea is to set your default multiplier via the L3 bridges to some number lower than 12.5, & to have all the L1 bridges open so you can then adjust the multiplier via the K7D BIOS up to 12.5 [the K7D cannot soft-adjust higher multipliers than this].

Advise you read around on this procedure - but ocinside is a good place to start & info is solid.

BTW: have just run a few encodes [real ones: CBR 6000 PAL ZZ/P] including 2 at 2325MHz: one at 17x137; 'tother at 15x155: latter was faster by just under 8% despite slightly slower DDR timings

[McDo0m]

CCE 2.5: 1.481

Asus A7N8x Delux-Athlon 2800+XP-Dual 80GB Seagate SATA striped-1GB RAM Kingston-PC3200-Nvidia GeForceFX5200 128MB PersonalCinema AGP 8X

EDIT:
CCE 2.67 (Manual AVS load with prescribed values of DDogg's second post in this forum) : 1.21

Also my 2*512MB kingston PC3200 rams will not run in DUAL Channel mode[Machine hangs up]
EDIT/

[DD51]

This message has been edited with proper values!

I found this thread fascinating and decided to try this myself.


After 3 minutes I get:

With 41 processes running in the background
CCE 2.5 - 1.381
CCE 2.67 - 1.42

Reboot and with 28 processes running in the background
CCE 2.5 - 1.392
CCE 2.67 - 1.45


My specs:
Motherboard: P4C800-E-DX GigaLan 800mhz Intel-i875 rev 2.
CPU: Pent IV 3.0C 512K 800mhz FSB PPGA-478 Northwood.
Overclock: 5% to 3.15GHZ
Memory: 1GB: 2x Kingston #KVR400X64C3AK2/1G PC3200 512MB DIMM.
Video: GeForce Ti4600 4X W/62.11 drivers installed.
DVD-ROM: Lite-on JLMS XJ-HD166S.
DVD Writer: Pioneer DVD-RW DVR-105.
Floppy: Basic Panasonic.
HDD 1: Maxtor SATA 120.0 gig 7200 rpm 8MB Cache 6Y120M0.
HDD 2/3/4: None.
Sound: Audigy 1 with Audigy 2 drivers installed
Klipsch ProMedia 5.1
Monitor: Viewsonic P95f+ Black 19"[COLOR=royalblue][COLOR=royalblue]

[PleXuS]

CCE 2.67.00.23
Speed: 1.36

System Config:
Asus K8V Deluxe
Athlon64 3200 @PR3400 (2,2ghz)(FSB448)
1mb L2 Cache
1 x 512MB Corsair DDR466 (Cas2.5)
1 x 200Gb Sata Maxtor
1 x 120Gb Uata Maxtor
Asus Radeon 9200 128Mb

Temps:
Cpu stressed: 48°C
cpu idle: 36°C

Cpu Config:
Cool&Quiet switched On!

u want to know more? let me know..

-GreetZ-
PleXuS

[onesoul]

Originally posted by onesoul
Why do I see a tenuous vertical white line next to the yellow bar with cce 2.67 moving left-right? I don't see this happen with cce 2.5! Could be the reason of cce 2.67 being optimized for encoding motion? :confused: About this problem, I tried different settings and I noticed that quality precision isn't equivalent between cce 2.67 and 2.5. For the ecl of 2.5 is at 17 and 2.67 is at 25. The range for cce2.5 is 0-100 and cce2.67 is 0-64. So equivalent setting for cce 2.5 at 17 should be cce2.67 at 11. Or for cce2.67 at 16 should be cce2.5 at 25. Just do the math.

And I found that although cce2.67 does have some shimmering on the transition of white bar (at the given bitrate) it does a better job on the transition of other colors.

[buzzqw]

@DD51

wrong script or wrong avisynth version

open the avs in virtual dub and see the message

BHH

[DD51]

buzzqw,

Thank you.:) I had copied the "code:" into the AVS script. When I removed that line it worked well.
Thanks for helping out.
I have edited my message above (4 up) to reflect the true numbers.

Regards,

[br408408]

Computer:

Abit NF7-S
AMD XP2600 Mobile @ 204 x 12.5 = 2544mhz with L12 mod
Thermalright SLK900A with Panaflo 92x25mm High Speed Fan
2 X 256 Twinmos Winbond CH5 dual channel

CCE 2.5 speed after 3 min = 1.831......WOW:D

This motherboard and cpu may replace my P4 setup in my main rig...and this was at 1.75 vcore and prime95 stable! 1.8 vcore is safe so I think it has more in it. Will be doing more testing soon.

[laura_glow]

Can I run this tests wit CCE sp trial version? I try to open the ecl files but i cant, with the same message as when i try to save a configuration (option disabled in trial versions).

I really would like to run the test, can someone tell me how to configure exactly CCE sp trial from scratch? also, what version ov avisynth is needed would be great...

i will edit this post with the results as soon i have them.

Thanks!!

[cypher_soundz]

UPDATE:
So i just got 2GB of DDR ECC REGISTERED 2700 (running at 2100) RAM and im running a single 2400XP+ (untill i mod my other one) and my speed is below:
CCE 2.5: 1.054 (avs 2.05)
1.163 (avs 2.54)

I feel my speed is a little low compared to others with a 2400?

Regards
cyph

[sh03z]

wow, this is a loooong thread

[Dox]

0.79 on Cce 2.67

I have over year old computer with p4 2.0 ghz.

[cypher_soundz]

So who is going to make these results in to a readable format?:rolleyes: anyone?;)
Regards
cyph
I might have a go at it:cool:

[DDogg]

New machine -

Speed 1.740 - CCE 2.67.00.27
Speed 1.836 - CCE 2.50

ABit NF7-S Rev 2 | Mobile XP 2500 @2.43 Ghz (11.5*211) VCore 1.75 v - SiSoft PR 3528 (Est.)| GFrce 2 MX400 |(2) 256 meg PC3500 HYP_X (7-2-2-2 VDim 2.8 v Dual Channel)

[DDogg]

Oh yeah!

New machine - Water Cooled

Speed 1.800 - CCE 2.67.00.27
Speed 1.940 - CCE 2.50

ABit NF7-S Rev 2 | Mobile XP 2500 @2.59 Ghz (12*215) VCore 1.85 v - SiSoft PR 3750 (Est.)/Model XP-3612 (Est.)| GFrce 2 MX400 |(2) 256 meg PC3500 HYP_X (11-2-2-2 VDim 2.8 v Dual Channel @ 431 MHz)

[stidyup]

CCE 266 I can get upto 2.50 :D that's frameserving with Avisynth.

Before moving to SATA raid, I always found encoding from one drive to another was the quickest.

Enermax 465, Asus A7V8X-E Bios 1008
Coolermaster Aero7
Raid 0 chunk size 128 2x120gb SATA Samsumg
1xSATA 120 gb, Highpoint 1520 SATA, 1xIDE 180 gb, 4xAkasa hdd coolers
2800+ Athlon XP-M, 2x512mb PC3200 DDR Crucial
Sapphire Radeon 9600XT

[DDogg]

stidyup, try it with the test method outlined in the first posts of this thread. The test is not based upon a normal encode where I can get slightly over 3.0 RT on a fullscreen source.

[stidyup]

Ooops

I'll give it a go when I've got time.

[LB]

Tomorrow, I'll possibly get my new chips in... 3.7ghz @ 1000fsb
I'll be posting in here if all goes well :D

[YaoMing11]

CCE 2.50 - 1.801



Abit NF7-s 2.0

Athlon 3200+

1GB 400mhz Dual channel RAM

Seagate 160gb SATA

[Diable]

Damn, those old ass XP's are spanking everybody. Shouldn't the A64's be just as fast when using CCE 2.5?

I was gonna get a A64 3200+/MSI K8N Neo Platinum combo but after reading this thread I think I'll order a Abit NF7-S and moble 2500+ for my encoding box.

[br408408]

My XP beats my P4 in the speed test by quite a bit....but, when backinig up a movie using DVD2AVI (DGIndex), AVISynth 2.5 and CCE 2.5 my P4 still wins. Might be because of the Hyper Threading on the P4C. I have tested using the same movie and the same versions of all the software.

[LB]

2 x XEON 3.6ghz @ 848mhz (17x218) 2.5,3,3,6

CCE 2.67 1.80

[shoarthing]

2.16

Dual Opteron 250s [2.4GHz] on an Iwill DK8N

CCE-SP 2.67.00.27

IGB 2-3-2-6 Reg/ECC: HT link CPU<>CPU @ 1000

[Diable]

Motherboard: Gigabyte K8N Pro| CPU: Athlon64 3200+ @2.20 Ghz(10*220)| Memory: 1GB of PNY PC3500| Hard Drive: Seagate 160GB SATA|

v2.67.00.23 Speed: 2.18

I couldn't believe the speed so I ran it three different times and got the same speed.

[cypher_soundz]

Wow! Are you sure the script is working correctly? I'm very impressed!
Regards
Cyph

[DDogg]

Diable, if you have the chance, would you get the free version of SiSoft Sandra and do the memory benchmark test? I suspect the score will be very high and may be a big part of your excellent speeds. [Hmm, I read in reviews this may not be so]

To give you a comparison, my Nforce2 dual channel @420 shows 3200 MB/s. Your Nforce3 chipset should be much faster.

[shoarthing]

"I suspect the score will be very high and may be a big part of your excellent speeds" - well, this dual Opteron thing 'scoring' 2.16 has a SSSandra memory bandwidth of around 5400MB/s & is running at 2.4GHz, so mebbe there's another reason - like, say, a later/better AVISynth?

[DDogg]

shoarthing, I am sure glad you mentioned that. I checked speeds using the latest Avisynth Alpha build and there is a significant difference between 2.54 and alpha 255 AviSynth_200804.

The speed of encoding the simple test script is slower using 2.55 alpha (AviSynth_200804) than with 2.54 release. Using 2.55 gives 1.763 RT and using 2.54 shows 1.994 RT using the machine settings below. It may not apply to a real encoding source, but it certainly does mess up these tests. We should stick to 2.54 for them.

/Add: Note - The AviSynth_260804 RC3 build seems to bring speeds back to close to an equivalent to 2.54. Still, I think we should stay with 2.54 for consistency, IMO.


ABit NF7-S Rev 2 | Mobile XP 2600 @2.688 Ghz (12.5*215) VCore 1.85 v |(2) 256 meg PC3500 HYP_X (11-2-2-2 VDim 2.8 v Dual Channel)

[LB]

@DDOG

I'm having a problem. The script no longer works for me, even on a clean install. I'm trying it with 2.67.00.27

Regarding CCE 2.67.00.27. The people that have used that version had to have "modified" the script somehow. Why? It loads fine, but it tries to do a 1+3 pass; meaning it tries to encode the vaf first. I get a speed of 660 (yes, 660) and then when it tries to save the vaf I get an error obviously because vaf_out=0 is turned off and vaf_file= is blank. The error says:

vinfo_open() failed.
*** Could not open the file <> ***
The system cannot find the path specified.
[Windows error code: 0x3 3]


I was finally able to get it working on 2.67.00.23 though with a little help from eclcce and score a big 1.70 with dual 3.6ghz @ 920fsb using 2.55 and with 2.54 I get 1.73.

[DDogg]

LB, I have not heard of anybody having any problems. 2.67.00.27 works fine for me. No 'mods' should be required except maybe the path if you don't have an HD "D" drive. That was the whole point of the exercise. Just load and press encode. Just in case, here is my exact ECL and exact avisynth script again:

Save everything between the lines as "CCESpeedTest01_CCE267.ecl"
; 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=D:\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=D:\CCESpeedTest01\CCESpeedTest01.avs
type=0
frame_first=0
frame_last=107892
encode_first=0
encode_last=107892
Save everything between the lines as "CCESpeedTest01.avs"
#start of CCESpeedTest01.avs
ColorBars (720,480)
ShowSMPTE(24)
converttoyuy2()
#end

So long as you use the exact scripts above [mod the paths or use the exact ones that I did] you should have no problem. If you do, I would guess you have something wrong with your installation of avisynth or CCE? Let us know what you find. Btw, what happens when you 'play' the avs script above in WMP or VDub? You may be throwing an error.

[DDogg]

I got a little motivated, bumped VDIM, Vcore, and chipset voltage, and jacked up my speed settings up to 2.74 gigaherz. It seems pretty solid so far. Interesting though, I only just barely cracked 2.01 realtime using CCE 2.50. The speed I mentioned in a previous post of 1.994 RT was achieved at 2.688 Ghz (12.5*215), presently I am at 2.74 Ghz (13*210 Vcore @ 1.925, Chipset Vdd @ 1.70). The slower memory speed of 420 versus 430 for my older settings seemed to nearly counter the increased CPU speed.

/Add: 2.67.00.27 results were 1.85 RT [better after reboot]

Still, not too shabby at all to reach over 10,300 MIPS with a $95.00 CPU on homemade watercooling using a $15.00 '95 ford pickup heater core and an $19.00 fountain pump from Lowes.

http://img70.exs.cx/img70/9418/6801.jpg
http://img69.exs.cx/img69/4167/cpu.jpg
http://img69.exs.cx/img69/2262/cpu-mm.jpg


ABit NF7-S Rev 2 | Mobile XP 2600 @2.74 Ghz (13*210) VCore @1.925 volt, Chipset Vdd @ 1.70 volt - SiSoft PR 3974 (Est.)/Model XP-3848 (Est.)|(2) 256 meg PC3500 HYP_X (11-2-2-2 VDim 2.9 v Dual Channel @ 420 MHz)

[influenza]

Hi DDogg nice initiative!

It's not a very fast reply but here are my results for CCE 2.67:

Initial test: 1.62
Test with the mpg file : 1.77

Amd Xp-m (mobile processor, desktop pc) 2600+ @ 2,5 Ghz (12,5 X 200 IIRC). Just on air cooling
1 GB pc 3200 on an asus board
2 IDE 7200 maxtor drives (not in a raid configuration)

[coona]

Hi influenza,

Test with the mpg file : 1.77
what does it mean? Real encoding speed? It differs only a little from this DDogg´s torture test :confused:. I guess there should be much bigger difference between test´s and real encoding speed.

BTW - Can we find some volunteer for sorting these result into a table? This thread is becoming little bit useless...

[influenza]

Hmm sorry, Real time encoding yes.

Well honestly not a test with only CCE busy (mirc, IE and MSN), but indeed not too much difference.

(On a side note: Normally i get speeds around 2.2 on progressive PAL dvd material and 2,6 on progressive NTSC dvd material. Interlaced material around 0,3 lower than that)

[DDogg]

BTW - Can we find some volunteer for sorting these result into a table? This thread is becoming little bit useless... coona, I completely agree that the results need to be compiled so we can go to the next step. Thanks so much for volunteering! You are awesome! I know you will do a great job. :D

I really screwed up when I mentioned the true source test and then left on a 2 week vacation. That caused a lot of confusion. That should really have been the second step. Maybe after these results are compiled we can move to real source as a second iteration of these tests in a completely separate thread if folks are still interested.

IMO, the results using the test script have been useful, even if not indicative of real world encoding. They do seem to show clear and replicatable values that can be used to compare and identify how to optimize similar platforms. [think dual channel vs single channel ram, tweaked ram timings, fsb speeds, etc.]

However, as was mentioned, it may be that this test does not clearly show the clear speed winner between dissimilar platforms using real source, like Intel vs AMD or chipset vs chipset. That will take a separate test using real source. Maybe we can try to put out heads together on how best to do that.

I'm not so sure if using an AVI might not be an interesting way to go about it. That would then remove having to create a d2v. Making a d2v worries me as there are now so many versions of mpeg2dec floating around on peoples machines. Clearly that would render all the tests suspect.

I'm thinking maybe we could generate a huffy AVI of the test script used for these tests? Then that source could just be popped into CCE. The upside of that would be the removal of AviSynth and mpeg2dec from the equation so that the multi-CPU folks could be better involved and show their true results.

I'm also thinking to get this to work and make sure the source AVI was exactly the same would require a command line for VDub that would generate the AVI equally for all? Thoughts? I'm still not clear on the best way to do it so that the results are clearly fair and unbiased.

@Flu, I went back and edited in my 2.67.00.27 results above for comparison. Sorry I left that out. Oh, are you dual channel ram? What ram timings?

[influenza]

Yes I have dual channel ram, guess I have to reboot for the timings though

[Diable]

Originally posted by DDogg
Diable, if you have the chance, would you get the free version of SiSoft Sandra and do the memory benchmark test? I suspect the score will be very high and may be a big part of your excellent speeds. [Hmm, I read in reviews this may not be so]

To give you a comparison, my Nforce2 dual channel @420 shows 3200 MB/s. Your Nforce3 chipset should be much faster.

Sorry I didn't respond faster but the motherboard in box died four days after I ran the speed test. I'm back to a 2.2GHz P4 so I can't run Sandra :(

[vipera]

My results using the original script test:

CCE 2.5: 1.920
CCE 2.67: 1.88

A64 3000+ @ 2470mhz (247x10)