DVD-RB speed database....

[fjhdavid]

I think it could be useful to have a "kind of" database of DVD-RB speed, as it does exist for other soft.

Anybody can put their speed results, in this thread,for example, this way:

processor type: AMD, Intel core2 duo...
processor frequency: 2Ghz, 2,5Ghz...
clip size: 704x480 NTSC, 720x576 PAL...
clip origin: video, film, anime, DVD, TV
encoder: CCE, HC-encoder....
encoder quality settings: normal, best...
Multiple encoder processes: 1,2,3 or 4....
others settings: if you use specific avisynth script, give the script....(even if for initial purpose, no avisynth script should be used)

Thank You and it is up to you in this thread....

[Wombler]

I can see where you're coming from but unfortunately there are too many variables to make the results directly comparable.

Every video clip is unique and stats for different clips can't be correlated on the same hardware let alone different specs.

So regrettably it wouldn't work.


Wombler

[SpazzHH]

So regrettably it wouldn't work.

Have to agree with Wombler on this one. Even doing just plain Jane DVD's with no filters at all, there is just way too much difference from disc to disc to give a true reading of an encoding speed.

[fjhdavid]

another possibilty, would be to test a very wellknown movie like starwars 1 to 5?

and even if we stay with others movies and if there are 15% or 20% of difference from one disc to the other one, it should be seen as a starting point, no?

for me for example, It goes from 50fps to 60pfs with a lot of film with one of my computer, and between 15fps to 17fps on another older computer....

[Wombler]

Even having a standard film doesn't provide a valid comparison as some encoders may perform better than others with certain types of hardware.

Simplifying things greatly for the purposes of explanation, the basic problem is you're trying to chart two different variables at the same time, the hardware and the software encoders.

In reality very few people will have the same hardware combination (different processors, different speeds of hard disk) which complicates things further.

Even on the same hardware the set up affects performance.

For example in systems with multiple hard drives do you read and write the data on the same hard disk or use separate drives for each.

That's why benchmarks are done as far as possible on identical hardware with identical software configuration and varying only the element they wish to comparatively measure (e.g. processor etc).

In real life that never happens as people just don't operate that way and effectively every system has a unique configuration of hardware and software.

Have a go by all means but whilst it's possible to collect the data I think you'll have very great difficulty putting it across in any meaningful sort of way.


Wombler

[turbojet]

It's the encoders you really want to benchmark right?
If so, I think the best way to do it is have a specific video and settings in a pack like TechARP (http://www.techarp.com/showarticle.aspx?artno=520&pgno=0) did for their x264 benchmark.

[fjhdavid]

yes it is exactly this kind of bundle we need.

I don't know how to do it, but we should have a 2mn video clip downloadable with fixed HC-encoder settings and a excel file where we can output our result along with the most important PC spec (frequency, memory amount and hard drive speed...).

It won't be perfect, but sufficient to have a good starting point.

What Jdobbs and others are thinking about that?