Would you believe CCE speed of 5.335?

[Roveer]

I just finished building my Dual Xeon 2.8ghz 533mhz fsb water cooled monster CCE encoder and was dying to see what cce speed I'd get. Well, here's the result:

http://users.rcn.com/roveer/ccespeed.jpg

I don't currently have enough HD space on the machine to do a whole DVD, and this was just a quick run of 600 megs or so (3 pass). I was expecting 2.x or possibly 3.x I'm completly blown away by this and can't wait to get my bigger hard drive running so I can do a whole DVD. Also, this was reading the VOB across a 100 meg network connection (because I don't have enought disk on the machine yet.

I sure hope these speeds stay where they are, if they do, I'll be very happy.

Edit: The results I posted were too good to be true. Something caused CCE to to process the file properly (vob across network etc), and the actual speeds were more like 2.3. Should have checked the results before posting. Shame on me :eek:

[Mancubas]

v Jealous, you obviously have a very nice bank manager.

I have a nice Dual AMD MP setup at work which I intend to use once I get my head around cce and maestro etc etc...

[slk001]

Well, okay, but how about some more information?
What version of CCE are you using?
Are you frameserving?
What does your frameserving script look like?
How are you frameserving?
What was the original source?
Etc, etc, etc.

Give us this info so someone could duplicate your results, or even see if your technique is applicable to the way someone else does it.

[ux-3]

I got about 4 myself on a single Athlon XP1700+. So given your investment, you deserve a 5...

[int 21h]

Originally posted by slk001
Well, okay, but how about some more information?
What version of CCE are you using?
Are you frameserving?
What does your frameserving script look like?
How are you frameserving?
What was the original source?
Etc, etc, etc.

Give us this info so someone could duplicate your results, or even see if your technique is applicable to the way someone else does it.

Looks like he is definitely frameserving, using a dvd2avi project as the source, vfapi as the frameserver.

[Arky]

What framesize input; what framesize output; and what standard - PAL or NTSC??


Arky ;o)

[Roveer]

Remember, this was a quick little test. I'm still working to get the rig in shape to do a whole flic. Here' the answers to a few of the questions:

What version of CCE are you using?

A: CCE 2.5SP (what I had lying around)

Are you frameserving?

A: Yes, VFAPI

What does your frameserving script look like?

A: I used the VFAPI GUI to prep the input

What was the original source?

A: first few minutes of a flic (don't even know what the title was), but not the studio logos, I DVD2AVI cropped past them to the actual beginning of the movie. I'm running it through on my 1ghz machine to make sure it was a valid test. So far the CCE looked normal. I got .7 on my 1ghz machine (for the VAP file) didn't hang around to see the actual encode.

Input 720x480 output the same. NTSC. I never crop on DVD's

I'll put together a more extensive test when I get my SCSI running on the machine and do a head-to-head comparison between my new rig and my 1.7ghz P4. It will be interesting to see how they stack up.

Roveer

[mb1]

Bitrate 0.2 Mbps !

If we all would encode just black frames we`ll get speeds over 2 with our single processor systems.
So it is nothing special for a double processor system.

Nice try :D

[Roveer]

Hang on a sec. I just encoded the same material with my 1ghz machine and the frames weren't blank. In fact the first fist 7 minutes of 3000 miles to graceland. I'm in the process of bringing the stuff I encoded on the new box back over to my DVDMaestro box so I can master and see if I got the same footage. Hold your horses. I could be wrong, but then again, so could you. ;)


Edit: Yup, you caught me. Something is wrong. I'm going to re-run the test this time with the VOB and everything on the same machine. Something is not working right. I'll report back on the results: Looks like I'm gonna eat some crow. Wonder what the problem is? This is my 'typical' test for CCE speed? Only thing different here (besides new machine), is that I'm sucking VOB across network. Let's see what happens when it's all on the same machine.

Edit: Well, don't ask me what doing this across the network did, but It sure did something. I moved the VOB over to the new machine and ran basically the same thing and now I'm getting 4.5mbps and 2.3 (on the vap, and 2.3 on the encode). I should have looked at the results before I ran here with my good news. Shame on me.

Now, what version of CCE is going to get my Dual 2.8ghz xeons to work beyond 65%? I've read all kinds of stuff about how certain versions are more SMP than others. Who wants to take a stab at that one?

[mrbass]

I got that CCE speed too. That's when I was overclocking my 2.53GB to 2.8GHz...my ddr 400 was at 412...needless to say it was doing something whack..Wait...maybe I was overclocking it more than that...Hmm I can't really remember but I did get a speed around that.
.If I let it go for more than 30 mins..it would crash. Needless to say I don't overclock anymore it just isn't worth it. Here's my CCE speeds. (http://www.mrbass.org/dvd2svcd/ccespeeds/)

ok I found that thread...I was getting 6.2 CCE speeds
http://forum.doom9.org/showthread.php?&threadid=34604&highlight=cce

[mb1]

what version of CCE is going to get my Dual 2.8ghz xeons to work beyond 65%
I would recommend version 2.64.01.10 or 2.66.01.07 for dual cpu

From CCE history list:
Version 2.64.01.02 Tuesday, November 20, 2001

Changes
- Tuned for Dual CPU environment.

And another hint for significantly more speed:
Don't frameserve with vfapi !
Use avisynth (2.0.7 or higher) in mpeg2dec-mode instead ...

[ux-3]

Originally posted by mb1
Bitrate 0.2 Mbps !

If we all would encode just black frames we`ll get speeds over 2 with our single processor systems.


hehehe... so you caught it too. That's the reason why I posted my 4 for a single xp1700+ cpu. Empty 8*16 frames are quick to encode...

cheers
ux-3

[Pauly]

Originally posted by mb1
I would recommend version 2.64.01.10 or 2.66.01.07 for dual cpu


2.66.01.07 also benefits from enabling hyperthreading making use of at less 3 of the 4 virtual CPUs giving a worthwhile speed increase which you don't see with 2.5.

[Roveer]

Well, here's some new results.

CCE 2.66 using AVISynth 720x480 NTSC. Settled in around 3.75 or so. Still waiting to look at finished results, but at least the mbps looked normal.

CCE Results (http://users.rcn.com/roveer/cce1.jpg)

Dual Xeon 2.8ghz with HT turned on.

Roveer