Confused, which Avisysnth/MT gives best performance?

[kostik]

I am trying to run the folowing script:
Dither_convert_8_to_16()
SmoothCurve16("0-0;32768-36768;65535-65535")
ditherpost()

and see no difference using MtMode or not using it.
What version(x86) of avisynth should I install and which MT should I use?

I've got c2q 2.7Ghz (Oc) 6 GB RAM.
Thanks!

[LaTo]

SmoothCurve16 (and all filters in SmoothAdjust) is natively multi-threaded... So you doesn't need MT.

edit: And SmoothCurve16("0-0;32768-36768;65535-65535") do nothing... So you can't expect any speedup :)

[7ekno]

Benchmark them :P

I know for some of the filter chains I use Avisynth v2.60 a3 Standard is faster, for some it's Avisynth v2.58 Standard and for one other filter chain that I use Avisynth v2.60 MT is faster ....

7ek

[TheFluff]

In general, Avisynth MT will give you a very low (or even zero) number of frames per second, because it's too busy crashing to get any real work done.

[Groucho2004]

for some of the filter chains I use Avisynth v2.60 a3 Standard is faster, for some it's Avisynth v2.58 Standard

Could you give me an example where 2.58 is faster than 2.60? So far every script I threw at 2.60 runs much faster than with 2.58.

[henryho_hk]

If u have enough memory, it may be even faster to keep them single-threaded and then run multiple encoding tasks concurrently.

[CruNcher]

Also on 2.6 threadrequest still works and does a great job for non multithreaded script parts