-->Ripping improving performance<--

[Jestak]

Greetings,

Maybe someone already was able to solve this issue or we can find a solution. It is why I am popping the question here.

I would like to improve the performance of ripping and encoding DVDs. Here is my configuration:

1 dvd player (Liteon) connected to IDE port1
1 dvd player/burner (NEC 2100A) connected to IDE port1
1 internal HD connected to IDE port2
1 internal HD connected to IDE port2

I tried to swap the ports or connect one dvd on the same port of a HD.

I know I can use RAID to improve HD I/Os. But when I rip and encode or rip 2 dvds at the same time, there is a jam with the I/Os even if I send to different HD (The cpu is still low level).

I use DVD Decrypter, DVD Shrink or Nero.

Any of you know of a setting with Windows to improve the performance?

Another question: If I use an external USB HD, will it be better performance wise than the current configuration?

Thank you

[bazooka]

Don't rip 2 dvd's at the same time.

Make the burner the master on the cable.

[Jestak]

Greetings,

The burner is the master indeed.

Also, I know ripping 2 DVDs at the same time is more taxing on the system than ripping only one.

However, the point is to be able to rip efficiently 2 dvds at the same time. I don't rip often 2 DVDs at the same time, but I process a lot of files through the DVDs readers, it is why the ripping thing is very good test wise. Like i said, I just want to try to improve I/O performances.

Thank you

[Gurm]

Without multiple CPU's, there is NO efficient way to rip/encode two DVD's at the same time. Even on the beefiest single-chip system, it takes more than twice as long to do 2.

If "d" is the amount of time it takes to do a single DVD operation, be it ripping or encoding, then the formula for the amount of time it takes to do two operations in a row is:

2d + k

Where "k" is the amount of time it takes you to swap discs and hit "go" again.

But the formula for doing it simultaneously on a PERFECTLY OPTIMIZED SYSTEM is:

(2 + q)d

Where "q" is the amount of CPU thread-swapping overhead. On a ripping operation, "q" is relatively low. On an encoding operation, "q" is another factor of d, making the formula more like:

4d

And on a system which is NOT perfectly optimized (and if you aren't using SCSI or SATA RAID, it's not optimized) then the equation is:

(2 + q)d + i

Where "i" is the IO nightmare time. Usually it ALSO adds up to a factor of d, meaning that it can easily take more than twice as long as it would just take if you swapped discs.

So that's the bad news. Now, if you get SCSI or SATA RAID, AND a dual-CPU board, then you could do two at once "efficiently".

[Jestak]

Greetings,

Thank you very much for the detailled explanation! I see now there is no way out of it.

Thank you

[bazooka]

Greetings,

The burner is the master indeed.

Also, I know ripping 2 DVDs at the same time is more taxing on the system than ripping only one.

However, the point is to be able to rip efficiently 2 dvds at the same time. I don't rip often 2 DVDs at the same time, but I process a lot of files through the DVDs readers, it is why the ripping thing is very good test wise. Like i said, I just want to try to improve I/O performances.

Thank you

It is not possible.

[Gurm]

It isn't immediately intuitive. For those of us who have been working with computers for years, we respond without thinking about it, because it IS intuitive if you have enough experience and prior knowledge under your belt.

But to the casual user, it DOES seem counterintuitive. I mean, assuming no IO bottlenecks, why not save the time and trouble of swapping disks? Because of CPU switching overhead, but most people are ignorant of that - no offense!

[powaking]

The only way I see this as an option is if say you are going to be stepping out for a few hours then yah sure do it up. Who cares if it takes 2 hours or so to rip 2 dvds at a time. If you're going to be out for 3+ hours then it doesn't really matter.

Catch my drift?

[blizno]

I rip two DVDs at the same time routinely. My CPU is hardly working at all so CPU bottleneck is not the issue. Here's my setup:
IDE RAID card in PCI slot with four hard drives in RAID 0 array (C:).
SATA RAID adapters on motherboard with two SATA drives in RAID 0 array (D:).
One optical drive on the motherboard's primary IDE adapter as master (E:).
The other optical drive on the motherboard's secondary IDE adapter as master (F:).

The bottleneck with most systems is not the CPU - it's one or both IDE channels. If an optical drive and a hard drive are sharing the one channel - or two opticals - the channel switches between them and the data flow is slowed way down. My setup has a separate channel for each data stream. No hard drive or optical drive has to share a channel with any other. When I have one drive ripping and start the other, there is zero slowdown. My two drives rip twice as fast as one drive.

[jimmalenko]

https://www.videohelp.com/forum/viewtopic.php?t=187112 might have some useful information.

[MOVIEGEEK]

I agree with blizno,when I rip 2 discs on different channels I can still maintain full speed and my CPU's utilization is ~50%.If I encode or transcode two files at one time my CPU will be at ~100% and the total time will be twice as long.
This is why you should make a on-the-fly disc copy with drives on seperate channels.

Here's how I have my PC setup:

Primary channel
Master-HDD
Slave-DVD-ROM/CD writer

Secondary channel
Master-BenQ DVD writer
Slave-Optorite DVD writer