how fast is Gordian Knot our my PC?

[goauld]

I bougth a new PC. Pentium 4 2000Mz, 256DDR, MotherBoard: Asus p4s533, Winfast Geforce 4 MX A170 64DDR, HD: 80 GB....

I tried code Matrix and I've gotten around 18-22 FPS in VirtualDub.

I used Neutral Bicubic filter and codec 5.02 PRO



Is this FPS like yours?

[Gollum]

I have an Athlon XP 2100+ (1733MHz) I usually get 50 - 60 fps in virtual dub. I also use Neutral Bicubic filter and DivX 5.02 Pro codec. I think yours should be higher. I can get 30 - 40 fps, while watching the DVD at the same time. :D

Gollum

[Milo]

50-60? Wow...

I have a pentium 4 running at 2.4b (133mhz bus), with 400mhz DDR, fast hard-drive, all that stuff and I get tops 35 fps on the first pass. 45-50 on the second.

[Gollum]

Wow...I thought yours would have been higher...:confused:

I just encoded Braveheart tonight, while I was watching it. Playing the DVD takes up some processing power. I also had a few other programs running. Here are some GKnot stats in case anyone was interested:

11:37:35 PM: Finished DivX5-First Pass. Duration: 1 hour, 44 minutes, 51 seconds.
11:37:35 PM: Trying to open Log-file.
11:37:39 PM: Success: Log-file open.
11:37:39 PM: Encoded: 247852 Frames.
11:37:39 PM: Speed: 39.396 Frames per Second.

1:02:47 AM: Finished DivX5-Second Pass. Duration: 1 hour, 25 minutes, 7 seconds.
1:02:47 AM: Speed: 48.524 Frames per Second.

Total Encoding Time: 3 hours, 17 minutes, 32 seconds.
8/09/2002 1:07:10 AM: Job "Braveheart" finished.

The movie is 2hrs 57mins. So for 20 minutes the movie was not playing.

[goauld]

sorry, I posted only the compression Check values before. Here you are more for Matrix:


Finished DivX5-First Pass. Duration: 2 hours, 1 minute, 39 seconds.
Speed: 25.513 Frames per Second.

Finished DivX5-Second Pass. Duration: 1 hour, 30 minutes, 48 seconds.
Speed: 34.177 Frames per Second.

Total Encoding Time (video+sound): 4 hours, 20 minutes, 59 seconds

Uhmmm...doesn't look as good as yours
:confused:

I know that AMD are faster but I don't like them.

Milo, is your hardrive 5400rpm or 7200rmp?. Mine it's 5400.

[manono]

Hi-

Unless all the parameters are the same, the comparisons are pretty much meaningless. Are these all at the same resolution, resizing filter, Force Film and not IVTC, no other filters-that sort of thing.

Larger resolutions slow you down. SimpleResize is faster than BicubicResize. Force film is faster than using IVTC. Are you all using the same versions of AviSynth? Are you all using the same MpegDec.dll? Are you all using the DivX5 codec? Only B-Frames, or Q-Pel and GMC as well? Are these all NTSC or PAL DVDs (does that matter?)? Then there are the hardware issues such as do you have a second hard drive to encode to (which speeds things considerably). Are you using SDRAM, DDR, or Rambus, and at what Bus Speed? Are you overclocked? I think that even the movie you're working with makes a difference. I like to compare as much as the next guy (I have a reasonably fast machine), but it's real hard to make apples to apples comparisons.

[t0rrente_]

Bueno ya he hecho algunas pruebas con este nuevo filtro y efectivamente si lo elegimos lo normal será que en el lateral derecho nos aparezca la dichosa rayita verde ( rosa en algunas escenas ) y que podremos ver con tan solo pulsar el botón "preview" antes de crear el *.avs.

Pero aquí viene lo bueno xD: Efectivamente tal y como lei en el foro de doom9 la rayita famosa desaparece totalmente si las dimensiones del croping "Width x Height" son divisible por 4 ( ambos parámetros ) por lo que he decidido actualizar el manual y explicar como hacer esto para quien desee utilizar este filtro a la hora de crear el *.avs.

Me encanta este filtro, a partir de ahora lo utilizaré siempre en todas mis codificaciones ya que es rapidísimooooooooooooo.....xDD y la calidad obtenida es bastante buena.

Como ejemplo os pongo los tiempos de codificación para una película de prueba que he hecho y ya vereis como os quedais sorprendidos:

EQUIPO: AMD ATHLON 1 GHZ, 256 RAM, WIN XP
PELICULA: "LA LENGUA DE LAS MARIPOSAS"
DURACION PELICULA: 1 h 31 m 32 s

ATENCION!!!!!!.......La película quedo codificada totalmente en tamaño CD en sólo 3 h 36 m 19 s. Como la hize a cachos ( aunque evidentemente no sea necesario ) os pongo tambien el tiempo de cada uno de los procesos por separado:

1ª pasada: 1 h 48 m 33 s ( 21,1 fps )
2ª pasada: 1 h 29 m 23 s ( 25,6 fps )
codificación audio: 16 m
multiplexacíon video + audio: 2 m 23 s
total ( suma de todo ): 3 h 36 m 19 s

increible !!!!!!!...pero cierto!!!!!!.........mirar la tasa de fotogramas por segundo tanto de la 1ª como de la 2ª pasada....pufffffff...........¡ en la segunda pasada me hacia más de un segundo de peli por cada segundo de compresión................................¡ Peaso de filtro ! Me quedo con él para siempre !!!!!!!!!!!!!!!!!!!

NOTA: La prueba se ha hecho sin las mpeg tools del codec, es decir, sin marcar la casilla "Use Pro Features ( use only if Divx5 Pro is installed):" en la pestaña Divx 5 del "DivX Encoding Control Panel

otra prueba mas hecha con el mismo equipo pero ahora con una película un poco mas larga:

EQUIPO: AMD ATHLON 1 GHZ, 256 RAM, WIN XP
PELICULA: "MARUJAS ASESINAS"
DURACION PELICULA: 1 h 44 m 40 s

ATENCION!!!!: La película quedo codificada totalmente en tamaño CD en sólo 3 h 52 m 14 s. Como la hize a cachos ( aunque evidentemente no sea necesario ) os pongo tambien el tiempo de cada uno de los procesos por separado:

1ª pasada: 1 h 54 m 44 s ( 22,8 fps )
2ª pasada: 1 h 35 m 56 s ( 27,7 fps )
codificación del audio: 19 m 5 s
multiplexacion video + audio: 2 m 29 s
total ( suma de todo ): 3 h 52 m 14 s

guauuuuuuuuuuuuuuuuu!!!! todavia aun mejor ( es normal que se tarde un poco mas ya que es una peli mas larga que "La lengua de las mariposas" pero lo que realmente importa son los fps y aunque son muy similares son superiores". Lo importante es que hemos hecho ya una peli medianamente larga, a doble pasada y con tamaño CD en menos de 4 horas...............increible de nuevo!! y eso que mi AMD es "sólo a 1 GHZ y 256 RAM" quien tenga un PENTIUM/AMD a 2 GHZ con 512 ó mas de RAM pues seguro que con este filtro hace pelis como "churros"....

NOTA: La prueba se ha hecho sin las mpeg tools del codec, es decir, sin marcar la casilla "Use Pro Features ( use only if Divx5 Pro is installed):" en la pestaña Divx 5 del "DivX Encoding Control Pane

[theReal]

do you have a second hard drive to encode to (which speeds things considerably). Manono, do you think it makes a lot of difference? I always thought as long as 99% of the cpu power is used for the encoding task, the bottleneck must be the cpu, not the HD.

btw. I also get framerates in the mid twenties for the first pass using avisynth 2.03 (64MB frame-buffer), resizing with neutral bicubic to 704xnnn or 640xnnn (not much difference in speed). In Divx 5 I'm using GMC and b-frames, light psychovisual settings (mostly, sometimes not) and I'm using an .mv file.

My cpu is an Athlon XP 1900+ @ 2000+ (1680MHZ, FSB 140), Epox 8K7A+ AMD 761 board with 512MB Crucial PC 2100, CAS2.
In Sisoft Sandra, the system benchmarks almost exactly the same like an XP 2000+ with a VIA KT266A board and CAS2 RAM at FSB 133 (my AMD 761 chipset is a little slower, but it's the same through the higher FSB).

Really, I can't believe the 50-60fps - maybe with other settings in Divx5 (no pro settings), or with smaller resolutions. But using the same encoding settings, maybe a new Dual Intel Xeon with RAMBUS can make 50-60fps, but not a single XP 2100+

[manono]

Hi theReal-

Manono, do you think it makes a lot of difference?

Yes-it makes quite a bit of difference I think. You're not doing the reading from and writing to the same hard drive. Maybe I should run some tests to quantify it on my own system.

A few minutes later. I did a short 2 minute Quant 2 test encoding to the same drive as the movie files, and then encoding to a different drive. I got only 2% more data written to the different drive in the same amount of time. I figured it would have been greater than that. But the badly fragmented state of the drives may have something to do with it-too many movies coming and going these days.:) And 2 minutes isn't enough time either. I'll try it again later.

Really, I can't believe the 50-60fps -

Yeah-I paused to think about that one when I saw it also. I figured he must be using a lower resolution. You take the resolution down low enough, and the speeds really pick up. There's just so much information we don't have. That's why I said the apples to apples comparisons are difficult to do. Also, I'm kind of out of the FPS comparison game since I started to use Convolution3D. It's a good filter, but more than doubles the encoding time.

[Gollum]

Really, I can't believe the 50-60fps - maybe with other settings in Divx5 (no pro settings), or with smaller resolutions.

I use the pro settings, with a very small resolution. 416 X 176.

The quality of the rip is Excellent. I watch my movies on my TV using TV out. Honestly from where I sit to watch, you can barely tell the difference between the actual DVD and the rip. You can see minor quality loss from up close. But who watches movies right in front of the screen?

I tested using higher resolutions and the results were little if any improvement. I figured most people would be using the lower resolutions, I guess I was wrong. Try it out, you will like the results. It is a lot faster.

I try to bring the % of the commpressibility test above 70 but less then 80. Here is a quote from the GK guide.

60% or higher will pretty much guarantee that you'll get pretty good looking results. Going over 80% is a serious overkill and you should rather increase the resolution. Values in the range of 50-60% still look pretty good and 40-50 is still OK, especially when you activate B-frames.

So it say's that over 60% will pretty much give you good results, but over 80% is overkill. So between 60 and 80 is your best bet according to this guide. I usually go a bit over 70. Why don't all of you? Great results...short encoding time? What more could you want?

[theReal]

I like 640xnnn or 704xnnn with compress test values of 70%.
This is where it's getting hard to be able to see the difference to the DVD, even on a 21" monitor :-)

I never liked resolutions below 512x384 for DVD rips, and even that is too low for my current taste (I always watch Divx on my 21" monitor - I currently don't have a tv set).

The encoding isn't that slow, I think - also, I'm using convolution3d on more and more movies (only those that need some filtering). The framerate is down to 9 fps in the first pass, but I don't care if the result looks good :D