How do I properly setup dedicated multi-disk ripping machine

[computer-girl]

I am building a dedicated ripping system that is configured as follows:

1.1 GHz P3
Windows XP
IDE drives
Several PX-230a & PX-708a drives (as many as reasonable for this system)

I wish to rip using secure ripping.


1. What PCI to IDE card manufacturer and model # is best to retain DAE (C2, etc.) detection in dbpoweramp?

2. Am I to only use drive per port and/or PCI to IDE card because the optical devices need to be set to MASTER only in order to retain C2 detection in dbpoweramp?

3. Can I use batch ripper with this setup or should I use multiple instances of R13?

4. What is the reasonable maximum number of drives for this system?

5. Should I expect the PIO problem with this setup? If one optical drive encounters a problem, will it slow all of the other drives down?

6. Is there anything else I need to know that I haven't listed here?

Thanks,

computer-girl

[bhoar]

Re: How do I properly setup dedicated multi-disk ripping machine

1. I'm not familiar enough with the IDE card choices to help, sorry.

2. The rule (one IDE drive per cable, with drive set to Master) is mostly about preventing problems with a scratched disc in one drive from severly impacting the rip in the other drive.

3. It's six of this, one half dozen of the other. The batch ripper itself uses multiple instances of the cd ripper behind the scenes. The primary differences are going to be easy of use (thats personal and experiential) and the metadata licensing rules which differ when you use the batch ripper.

My general suggestion would be to separate out any CDs that look scratched and handle them manually with the cd ripper task, since you may have to rip several times to get the results you want.

As for the rest of the CDs...if you're using a robot or changer, that's easy, use the batch ripper. If doing it manually, well, use the batch ripper until your "free" AMG lookups run out. If you're happy with how that works, then buy more AMG lookups. If not, revert to the CD ripper.

Some bugs were reported months ago when using multiple CD rippers in parallel with the drives occasionally switching around, but from what I understand, Spoon has fixed those bugs so you should be safe.

4. Summary: that's an older machine. I wouldn't go above two parallel rips on a 1.1GHz P3. But you'll need to experiment to find out what's right.

Longer version: if you are concerned about throughput, my general guide is one drive per CPU. If you have very fast disk IO (e.g. a RAID 0 or RAID 10 array) and you aren't using a CPU intensive codec/setting, you might be able to get away with up to 1.5 drives per CPU and retain throughput.

Now, you can add more drives than that, but you'll see the rips slow down. If your primary concern is not about IO throughput but instead is about how many times you have to unload all drives and load them again, this might not be a problem. In this case, you don't care if the machine is bogged down, you're just trying to reduce the "latency" incidents (since that involves you getting up, going to the machine, emptying drives and refilling them - you want to cut that down to as few instances as possible).

But...also if you have *way too many* drive (say four...maybe less), this may impact the ability of the cd ripper tasks to keep up with drive IO and then you'll see more serious problems than just slow rips, though I haven't pushed things to this point.

5. Yes, this can happen. Yes, one drive in PIO mode causes the rest of the drive rips to suffer substantially, even if they are on different controllers. The reason is that PIO mode mean drive IO becomes interrupt-driven and the CPU has to stop handling all other tasks to do the interface IO one byte or word at a time. Ick.

6. Of course...but hopefully it will be a more fun adventure than frustrating one.

-brendan

[computer-girl]

Re: How do I properly setup dedicated multi-disk ripping machine

1. I'm not familiar enough with the IDE card choices to help, sorry.

2. The rule (one IDE drive per cable, with drive set to Master) is mostly about preventing problems with a scratched disc in one drive from severly impacting the rip in the other drive.

3. It's six of this, one half dozen of the other. The batch ripper itself uses multiple instances of the cd ripper behind the scenes. The primary differences are going to be easy of use (thats personal and experiential) and the metadata licensing rules which differ when you use the batch ripper.

My general suggestion would be to separate out any CDs that look scratched and handle them manually with the cd ripper task, since you may have to rip several times to get the results you want.

As for the rest of the CDs...if you're using a robot or changer, that's easy, use the batch ripper. If doing it manually, well, use the batch ripper until your "free" AMG lookups run out. If you're happy with how that works, then buy more AMG lookups. If not, revert to the CD ripper.

Some bugs were reported months ago when using multiple CD rippers in parallel with the drives occasionally switching around, but from what I understand, Spoon has fixed those bugs so you should be safe.

4. Summary: that's an older machine. I wouldn't go above two parallel rips on a 1.1GHz P3. But you'll need to experiment to find out what's right.

Longer version: if you are concerned about throughput, my general guide is one drive per CPU. If you have very fast disk IO (e.g. a RAID 0 or RAID 10 array) and you aren't using a CPU intensive codec/setting, you might be able to get away with up to 1.5 drives per CPU and retain throughput.

Now, you can add more drives than that, but you'll see the rips slow down. If your primary concern is not about IO throughput but instead is about how many times you have to unload all drives and load them again, this might not be a problem. In this case, you don't care if the machine is bogged down, you're just trying to reduce the "latency" incidents (since that involves you getting up, going to the machine, emptying drives and refilling them - you want to cut that down to as few instances as possible).

But...also if you have *way too many* drive (say four...maybe less), this may impact the ability of the cd ripper tasks to keep up with drive IO and then you'll see more serious problems than just slow rips, though I haven't pushed things to this point.

5. Yes, this can happen. Yes, one drive in PIO mode causes the rest of the drive rips to suffer substantially, even if they are on different controllers. The reason is that PIO mode mean drive IO becomes interrupt-driven and the CPU has to stop handling all other tasks to do the interface IO one byte or word at a time. Ick.

6. Of course...but hopefully it will be a more fun adventure than frustrating one.

-brendan

Brendan,

Thank you for your thoughtful and timely response! You have helped tremendously.

Regards,

computer-girl