CCE Speed?

[robinsj]

I just started trying this method and this has probably been posted before, but just wondering what kind of speeds people get with CCE? I am using CCE 2.50 and am getting about 3.008, is that good? Would I get faster if i used CCE 2.67? any help is much appreciated.. thanks.

[Eyes`Only]

holy crap thats fast.

[caprioles9000]

whats your setup ??/ What type of a rig you got there ...dual xeons ..or somethin else ..that is pretty fast ??

[influenza]

I must be doing something wrong here. more than 1,5 I've never seen.

[-JS-]

:eek: 3.008! :eek: I normally hit around 1.4!

[robinsj]

I have dual xeon 3.06ghz .... even that actually was when it just started, after i let it run a bit it got up to 3.150. I am just curious if version 2.67 would go faster because doesn't it support multiple processors? did any of you notice that 2.67.00.20 just came out october 20th?

[shoarthing]

robinsj - Hi - at last a first-hand CCE speed report from a dual Xeon 3.x GHz box :) . . . as regards speed-difference between CCE versions, I find CCE 2.5x if anything very slightly faster. AVISynth 2.5x is definitely faster than earlier versions.

All these CCE versions support SMP & the recent ones are [supposedly] PIV-optimised. Tell us what you find.

AVISynth is the bottleneck for fairly fast SMP boxes [it's not a multithreading app] - what CPU%'s do you see when encoding?

As a basis for comparison; a dual Barton box at 2.25GHz averages in the 2.8-2.9x range over a typical three-pass VBR encode usng CCE 2.67.00.13/AVISynth 2.52/current mpeg2dec3 - this is at 'normal' bitrates/complexity using the Big3 for what they were designed for . . . I don't see more than 80-ish% on both CPU's at this encoding-rate.

You may well get a speed-boost by using the switch idct=5 [ie SSE2] in your *.avs scripts - I got a [small] boost using idct=6

[edit: need hardly add these reported rates are with non-interlaced material]

[Vejita-sama]

Running CCE 2.5 SP

I have a AMD Barton 2500+ overclocked to 11x200 (3200+ speeds). I normally get CCE speeds of ~2.6-3 for most encodes. If I run a IVTC process it drops to 1.5~2.0.

On my old machine (AMD Palamino 2600+) I got speeds of ~1.7 for most encodes and like ~0.8 for IVTC processes.

I've notice that CCE seems much faster when the source and output directories are different (which doesn't really help with the 'Big3'). I'm a newbie to using the big3, but like it so far (only 2 discs completed so far).

Full system:
AMD Barton 2500+ [11x200] @ 1.7V
Abit NF7-S rev 2 MB
2x 512MB Buffalo PC3700 DDR (BH5/Winbound) @ 2.8V

[robinsj]

Well i am still trying to get my hands on a special version of 2.67.00.20 if you know what i mean... But I will post the results of what speeds i get with it. I am really new to this so you will need to explain to me what IDCT is? and where do i find it to change it?

[robinsj]

Okay i am trying it out with 2.67.00.09 version... I have getting a speed of 3.73 and my cpu usage is only at 52%... why is that, how do you get it to use the cpus to there potential?

[arlsair]

CCE 2.66+ support SSE2, so for all owner of such CPUs CCE 2.66+ is faster than CCE 2.5.
Original CCE FAQ
CCE-SP 2.50 is still the fastest choice for AMD processors while v2.66/2.67 will run fastest (and a little faster than 2.50) on Intel P4 systems.

And for the 52% load:
Original DVD2SVCD Q&A
Q49: Does DVD2SVCD support dual processors?
A: DVD2SVCD is a frontend to multiple steps. Most of these steps don't support multiple processors, in fact only CCE and TMPG do. The good thing about this is that the most time consuming process, video encoding, is performed by one of these two programs. On the downside, these programs rely on the frame serving by avisynth. Avisynth, sadly, does NOT support dual processors.
Practically, this means that:
1) when encoding with CCE, speed is mostly limited by avisynth. Avisynth is using 50% of your cpu resources, CCE only needs about 10-15%. Speed will be faster compared to a single processor pc, but not that much.
2) when encoding with the not so fast program TMPG, speed is limited by both avisynth and TMPG. Most of the time 100% of your cpu resources will be used. Speed will be a lot faster compared to a single processor pc, but it will still be slower than encoding with CCE.
3) This does aid in your ability to use your computer for other tasks while DVD2SVCD is running.
4) Of course, your OS must support dual processors, NT, 2000, XP only.
- thanks to da franksta

[robinsj]

I did the change to the idct that you mentioned... and just wanted to post my speed results with cce 2.67.00.11 is 4.09 .... that rocks.... with only 53% cpu usage... I wish someone would figure out how to make avisynth utalize both processors.... that woudl be awesome

[Grover]

Originally posted by robinsj
....I did the change to the idct that you mentioned... and just wanted to post my speed results with cce 2.67.00.11 is 4.09 .... that rocks.... with only 53% cpu usage...
Now your just "showing off" to make us mortals feel jealous and puny :o

FWIW, I am a PAL user and do a variety of de-interlaced and also progressive encoding with CCE 2.50.
The CCE encoding speed for my de-interlaced material is about from 45-55% the speed I get when I can leave stuff to encode as straight progressive which runs at 1.7 to 1.85 usually but sometimes as "high" as 2.04. Not sure why it varies so much.

Does HT on P4 make any difference (to CCE) "out-of-the-box" or do I have to adjust CCE to get any benefit ? Or am I kidding myself becuase at the end of the day there IS only ONE actual processor in my PC ?

P4-2400 800FSB HT(OC to 2.93GHz)
PC3200 x 512MB dual channel (2x 256MB)
Asus P4P800

Cheers...

[shoarthing]

Robinsj - Hi - hmm . . . nearly 10% gain for the idct6 switch [SSE2] sounds a heck of a lot to me . . . what average encoding bitrates are these figures based around?

How many CPU graphs are shown in your system's task manager? - I assume the thing is a dual HT Xeon, so you should see, or be able to see, 4: interested to see if one of these [presumably devoted to AVISynth] is very much higher than t'others.

Perhaps you might mail me at: shoarthing_NOSPAM@burningissues.net [remove caps/underline] with a *.jpg/*.png screencapture of the CCE & taskmanager windows while encoding - I'm writing a piece on tweaking CCE & would appreciate input from dual-Xeon users.

Grover - Hi - presume that v2.5x of CCE-SP is not getting the best from your hardware.

[arlsair]

Originally posted by Vejita-sama
I've notice that CCE seems much faster when the source and output directories are different (which doesn't really help with the 'Big3').
You can select in DoCCE4U/BatchCCEWS under "settings" where to save the output. You just have to copy later the output to the input directory.
But for me it is not faster and not slower. And it makes sense for me in consideration of the low traffic.

[Eyes`Only]

can't imagine that being faster at all. CCE even at speeds such as 4.x can't even come close to the speed of the slowest hdds found on today's systems. Even using the same hdd, the hdd is basically hurry-up-and-waiting for CCE's output.