Factors to make DVD Shrink run faster

vbimport

#1

What can be done to make DVD shrink run faster? I have it set on deep analysis mode, and it takes on around two hours to analyze/encode a movie after ripping it with Ripit4me.

I am guessing that since the CPU runs at 100%, having a more powerful processor would help. Would something like a dual-core processor be a big improvement, and what other factors help to make it faster? Would adding more RAM also help DVD Shrink run faster?

System: AMD Athlon 64 3200 socket 939, 1GB RAM (512x2 in dual channel), Epox M1697 motherboard


#2

I use shrink v3.2 with the default settings, turn off video preview and i’m done in approximately 6 minutes.


#3

Yes, a new CPU is the only thing will help, although transcoding times like those usually mean that you are over-compressing the movie. You really should stay above 75%. If you’re watching on a smaller screen, you could turn off the quality settings and just use deep-analysis. That would speed it up.

2 things will make it go faster, CPU-wise. More L-2 cache will hugely speed up the analysis phase (1 MB), and more speed. You can get a 939 X2 CPU for next to nothing these days, any one of them will be a big help over what you have now. With 1 MB L-2 cache x 2 and 2 cores, you’ll be impressed.Shrink is multi-threaded, so dual core really helps.

More RAM will not help, but [I]faster[/I] RAM will help. Either higher clock speed or tighter timings. The benefit there will be small.


#4

Another way to improve DVD Shrink processing time, is to remove all unnecessary material from the DVD before starting the shrinking process, because the less compression you have to perform, the faster it will be.

First step would be to only choose the Main Movie (unless you really want the extras or the menus).

The you can remove audio tracks that you don’t need, e.g. commentary audio tracks, audio in other languages, redundant audio tracks in different formats (e.g. AC3 2ch, AC3 5.1ch, DTS).

Removing subtitles will not matter much because they take up very little space.


#5

Would either of these processors be a significant improvement over what I currently have (Athlon 64 3200 socket 939) when using DVD Shrink? Or would something more powerful be needed to notice a difference? Thanks

AMD Athlon 64 4000+ San Diego 2.4GHz 1MB L2 Cache Socket 939: $70

AMD Athlon 64 X2 3800+ Toledo 2.0GHz 2 x 512KB L2 Cache Socket 939: $95


#6

Dual core is the only way to go, regardless of the clock speed it will kick the pants off your current CPU. I would really try to get 1MBx2 L-2 cache too, cause analysis speed will almost double compared to 512x2.

But if you have to choose between those 2, the Toledo X2 is the clear choice.


#7

While having dual cores is nice, it won’t necessarily help transcoding speeds unless the transcoder is multi-threaded (or if you run two transcoding jobs at once). If the transcoder is single-threaded (which most, if not all, are), then getting a cpu with a higher-clocked single core will be faster than getting lower-clocked dual cores. However, I would still get a dual core processor because it’s more future-proof and it would help with multi-tasking.


#8

Um, yes it does. All applications will run considerably faster, whether they are single-threaded or not. As much as 20% faster. In my own experience, more like 25% faster in single-threaded encoders. May be in part due to having 2x the cache, who knows. But it’s clearly faster based on encoding times and indicated speeds.
Here, Procoder went from 0.7x to 0.9x on identical encoding jobs when I changed from single core with 1MB L-2 to dual core with 2x 1MB L-2. (same clock speeds, same RAM). Experience with other single-threaded encoders is similar. That, plus the fact that you can actually use the machine while encoding in normal priority.

There are plenty of compelling arguments for having dual core. Although it’s common to see people use the “single-threaded” argument against dual core, nobody who actually uses one will make that argument. Even at lower clock speeds, dual core always wins.


#9

Which won’t bring much in terms of speed since Shrink does not touch the audio streams anyway.


#10

Um, by its very definition, a single-threaded application can only utilize one core at a time. I’d suggest you do some reading about dual core processing, specifically the part that says:

A single-threaded application running on a dual-core CPU simply will not benefit from that second core.

The only times a single-threaded application will benefit from having a dual core as opposed to a single core (assuming the cores are otherwise exactly the same) is if you’re multi-tasking, which in itself is multi-threading. I have a Core 2 Duo myself, and I’m perfectly fine with making the argument that dual cores are not always faster.


#11

Yes it will bring some speed, because when you remove some of the audio streams the video streams don’t have to be compressed as much, and shrinking will run faster when you don’t need to compress as much.

So removing anything that you don’t need from the DVD before shrinking will indirectly make the compression go faster, and will also improve the quality of the video that you do keep, because less compression is necessary.


#12

Oh please. ALL systems are multitasking, all the time. Single-threaded applications always benefit from being run on a dual-core system, for this very reason. I have around 40 processes running all the time. So, yes, they are in fact [I]benefiting[/I] from having that second core. But since this thread is about Shrink, it’s a moot point. Still, I’d be interested in hearing about your [I]personal experiences[/I] with ANY single-threaded video encoder that does not run faster on a dual core system. I know I have yet to find one, and I use most of them.


#13

Ok, let me restate my point. You said:

Even at lower clock speeds, dual core always wins.

A lower-clocked dual core will only be faster than a higher-clocked single core in a single application if that application has to compete with other processes that consume enough CPU time that the higher clock speed is negated.

For example, let’s say you have one single core CPU that is 3ghz and one dual core CPU that is 2 x 2ghz. The cores are exactly the same (same architecture, same cache, etc). Let’s assume that these CPUs are installed in a normal Windows OS that idles at 5% CPU usage (which would be fairly high for idle). If you run a single-threaded application on both of these systems while the system is idling, then the single core CPU will complete the task first because it has 2.85ghz available (95% of 3ghz). Even if the dual core system could devote 100% of its CPU time to the application, that would still only be 2ghz from one core while the other core sits idle.

So a dual core CPU is NOT always faster simply because it has dual cores.


#14

AMD Athlon 64 4000+ San Diego 2.4GHz 1MB L2 Cache Socket 939: $70

AMD Athlon 64 X2 3800+ Toledo 2.0GHz 2 x 512KB L2 Cache Socket 939: $95

Would either of these processors be a noticeable improvement over what I have now (Athlon 64 3200+) when running DVD shrink?


#15

Since the main function of DVD Shrink is ripping your DVD and that is depend to lots of thing like S/L or D/L, the format of that DVD been produced either is B/W or color the speed of your drive, is the drive in DMA and your amount of RAM and so on.


#16

It depends on your behavior with your computer. If you heavily multi-task while transcoding (as in simultaneously run another CPU-intensive task), then the X2 will be better. If you don’t do anything intensive while transcoding, then the 4000+ will be better.


#17

I :iagree:


#18

Thought I’d through my two cents in. I recently upgraded to an AMD 5200 dual core socket 940 with 1gb DDR2-800 and I noticed something odd. DVD Shrink is apparently multi threaded.

When transcoding with DVD Shrink windows reports 100% CPU utilisation. Switching to DVD2one, windows reports ~ 54% CPU use.

I did another quick test on two different machines, my 5200 and an AMD 3700 single core based machine. Naturally the DS setting were the same, no analysis, maximum sharpness, Both machines had Pioneer 11x dvd drives.

The 5200 was roughly 1.85 times faster, not the 1.20 predicted by clock speed alone.


#19

Almost any ripping program utilizes %100 CPU function and this is mostly true with Shrink.


#20

You can’t have read the thread, or my message. A single threaded transcoder will only use one half of a dual core cpu. I specifically said:

So “almost any” is certainly not true, and in line with point of this message thread, a dual core cpu is a very effective way to speed up DS if you only have a single core cpu.