DVD-RB free with HC - Extensive Conf Tests

[techreactor]

These test were conducted by me to determine best configuration in terms of performance of DVD-RB (in terms of time).

Standard Parameters
A 10 min 16:9 Video clip with 1 AC3 track and 1 English Subtitles
DVD-RB 0.96 Free
No Avisynth Filters
HC Encoder mode = BEST
Only encoding times are represented here(Time = Min:Sec)
MPEG2source(), IDCT mode was used in all to maintain some level of quality.
Different IDCT modes = Used three click mode and changed IDCT numbers in AVISynth script files

Machine Configuration
Pentium Centrino 1.6 Ghz(SSE2), 512 MB RAM on Windows XP SP2

Software's used:
Avisynth
HC encoder
DGdecode ver's
DVD-RB Free v0.96

Test configuration A

Avisynth 2.5.6a, HC Batch 0.16A beta, DGDecode Nan rel 1.1.1.0

IDCT7 = 31:56
IDCT6 = 32:24
IDCT2 = 33:29
IDCT1 = 32:18

From above results I shortlisted the least 3 timelines and used the corresponding IDCT modes for my next test.

Test configuration B

Avisynth 2.5.6a, HC Batch 0.16A beta, DGDecode ver bundled with DVD-RB 0.96 free (I think its Nan rel 1.1.0.0)

IDCT7 = 31:30
IDCT6 = 32:18
IDCT1 = 31:59

Since it became evident that IDCT7 mode was the fastest(moreso with the bundled DGDecode release), I used only this mode for my next tests.

Test Configuration C

Avisynth 2.5.5, HC Batch 0.15 beta, DGDecode ver bundled with DVD-RB 0.96 free (i.e. unaltered DVD-RB 0.96 free package)

IDCT7 = 29:15

okay so the 0.96 bundle gives the max speed performance, so I also wanted to check now if updating the HC will make any difference.

Test Configuration D

Avisynth 2.5.5, HC Batch 0.16A beta, DGDecode ver bundled with DVD-RB 0.96 free

IDCT7 = 31:45

Hmmmm....it increased the encoding time considerably. So what if I revert it back to HC ver 0.15 beta but change only Avisynth to latest.

Test Configuration E

Avisynth 2.5.6a, HC Batch 0.15 beta, DGDecode ver bundled with DVD-RB 0.96 free

IDCT7 = 29:33

which is the closest to timelines in "Test Configuration C" and means upgrading avisynth also adds to the time. (Jeo, I think this answers you)

So it means that dont upgrade anything in the original package of DVD-RB 0.96 free for BEST speed performance and use IDCT7 expert option.

Please also note these results are from my desktop and therefore results may vary when similar tests are conducted by you, since these tests extensively depend on the Desktop configuration and also on the source video used.

[UncasMS]

i wouldnt want to use hc015

stick to the latest 016 and let it take a little longer!

[techreactor]

If you were to ask me, i would always want to use the latest versions of all related software and thats why I did these tests, so ppl now know what combinations of softwares (latest vs Old versions) can boost or decrement their encoding speeds and then decide their options as they prefer.

[UncasMS]

thanx for testing, techreactor!

[jeo]

(Jeo, I think this answers you)

yes,......(and no).
i'm waiting your final conclusion

best configuration in terms of performance of DVD-RB (in terms of time)

because my pc(duron1.8) is very slow and i can't feel differences changing "versions".

regards.

[techreactor]

Like I said Jeo, I will not give a final conclusion to this, since I want ppl to be their own judge by lookings at these results and decide their choice of versions. Also cos there is a parameter of quality, which I am not addressing thru these tests. Like UncasMS said, "I would never use HC15 ver".

Also as you see the resulting time diff is very less in these test so if you say your PC is already slow, then you might not feel any speed gain.

But for you, my friend, i can suggest, dont change anything in the bundled installer and use only SSE* capable or optimized filters.

* = 1/2/3 as per your CPU.

[BR7]

Thanks for taking the time to do this tech and thanks for the heads up on the SSE filters.

[techreactor]

One more trick I just discovered is to put "*CPU SSE2" in the HC.ini file. Now my HC.INI file looks like this.

*BIAS 20
*CPU  sse2

This forces HC to use the SSE2* extentions and improves speed.

I did this test with 3 filters as below and conf as in Test B(IDCT7), the clocked time came out to 29:44.

Removegrain(mode=2)
STMedianfilter(3,3,1,1)
asharp(1.1,2.5)

which is much better than when I was not using the SSE parameter, considering I was not even using any filters in my earlier tests.

* use as per your CPU,i.e. SSE1/2/3