Divx 5 Performance based on CPU Speed Comparison

[billharrison]

Has anyone completed a comparison of speed for encoding using different speed cpu's and brands, IE intel and AMD?

I suppose one that used identical movies, as well as Identical settings in GKNOT for example.

I am just curious as my speeds seem a little slow, and I kind of wanted to know what others perhaps with similair systems were seeing.

I know that an assload of ram helps, I currently have 528mb Pc2100, does anyone that has 1GB or more feel that it improved their machines encoding speeds vrs having less?

I would not mind grabbing some extra ram, especially at the moment as prices have really come down. But if this does not make much difference, I dont really see the point.


Thanks!

Bill

[billharrison]

Ok, I just checked, and my average 2 pass encoding times for GKnot are in the 1hour 14 minute range.

Anyone else?

[hakko504]

1h14m doesn't say anything about your encoding speed. I can get that too for a twopass encode, no problem. If I encode a single episode of my Goofy collection @640x480 with Convolution3D (before lanczosresize that is) I get that kind of times too. The Gladiator will take me 8h though without any filtering, just a bicubic resize. Please use fps instead, first pass numbers is usually better. And if you want speed, try XviD in null mode. ;)

[billharrison]

Ok, so a good comparison would be the First pass FPS?

I will check that out ...

Thanks

[billharrison]

Ok, on the movie "Cruel Intentions" I got 42.371 FPS on the First pass at 512x272.

How does that compare?

[hakko504]

Originally posted by billharrison
Ok, so a good comparison would be the First pass FPS?
Yes, as long as you post your complete method at the same time so others can relate to it. (usually AviSynth script)

[billharrison]

Ok, method:

Divx 5 pro pro features enabled, using GKNOT

First pass FPS 57.076 at 512X272 using precise bicubic filtering, slowest quality.

[toliman]

i will contribute what i can, considering my only benchmarking tools would be an Intel P3 933MHz vs an AthlonXP 2100+, which is about 2x the raw MHz, and about 4x the raw grunt.

this question has problems, namely because it's the proverbial...
"how long is my piece of string",
and "how many knots can i tie in it"

i'm surprised it hasnt been flagged by a forum admin like it would have been if it were posted in avisynth or xvid.

if you want to create a valid, useful benchmark, use the same source for everyone's encodes, i.e. a few chapters of a dvd that everyone has, i.e. Matrix / SW TPM / LOTR / Fight Club, etc. or a movie trailer in mpeg2 source... make sure people are using the same source, then, get measurements from people.

first is the null-speed encode for measuring video subsystem/HDD speeds, and the general levels of speed when comparing YV12 to YUV to RGB on the same source. more of a test of avisynth's speed than the video codec. this is useful for judging the amount of video actually created by the encoder, vs work done pre-processing the image.

then a fast-recompress 1st and 2nd pass, no extra features, plain divx5... as plain as divx 3.11a, to measure the fastest time needed to encode a video.

then once more with all the pro features, including nth pass, mv hints, gmc, qpel, b-frames, psy enhance light/normal.

those figures would be a relevant starting point.

[Acaila]

I fail to understand the reason to WHY this thread even exists. A little bit of searching will provide a dozen threads where speed comparisons are given. Every once in a while someone asks "Is my speed like it should be?" and a thread exactly such as this is created.

Use the work people have done in the past to answer your question, instead of asking people to that same work again. Just because there's a new DivX version out doesn't make those old threads obsolete.

[billharrison]

Perhaps its because I did a search, and could not come up with much of anything.


Give it a go, maybe you can do better than I did....

[EagleDM]

I wrote a COMPLETE REVIEW of the new Pentium 4 3.06 Ghz with Hyperthreading and for base system (not for comparisson) I used a 1.4 Ghz Athlon CPU with KT266A memory working at 148 Mhz.

So if you like to see it, there is a COMPLETE info and graphics, it is in spanish, but you can check the graphs (they are in english)



http://www.maximopc.org/showreview.php?id=42

the main page is at: www.maximopc.org


I hope you like it, just read the numbers of the Athlon 1.4 machine and that will be MY machine, the other is just the BENCH machine that I will have to give back in less than a month :mad:

ah oh, btw: the test is done with Divx 5.03 (npasses) and XviD.

[Nethermancer]

And for further information you can find my Athlon XP3000+ review here:- http://www.hardwarereview.net/Reviews/AMD%20Athlon%20XP3000/AMD%20Athlon%20XP3000.htm

There are several Divx benchmarks as well as charts comparing it to a P4 3.06GHz and a Dual Athlon rig.