C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset)

[Spoon-]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

>Can I assume that these will work with the configuration above?

No, each PCI card might be different. I cannot remember the names of the cards I used.

[computer-girl]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

There are instances where c2 will not work in EAC but will in dBpoweramp (atleast on our systems), so it is half a dozen of one, half a dozen of another...

I recently built a system - quad core, 4 drives on PCI IDE cards (darn SATA MB's), on that system you could rip all 4 drives at x40 ripping speed (at the end of a disc) with less than 5% CPU usage (when using test conversion). Using Firewire, or USB you will quickly fill the systems bus allocation for these when using multiple drives.

I have my eye on the following drives


Plextor PX-230a
Plextor PX-708a
Plextor PX-760A

Can I assume that these will work with the configuration above?

computer-girl

[computer-girl]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

There are instances where c2 will not work in EAC but will in dBpoweramp (atleast on our systems), so it is half a dozen of one, half a dozen of another...

I recently built a system - quad core, 4 drives on PCI IDE cards (darn SATA MB's), on that system you could rip all 4 drives at x40 ripping speed (at the end of a disc) with less than 5% CPU usage (when using test conversion). Using Firewire, or USB you will quickly fill the systems bus allocation for these when using multiple drives.

Spoon,

How many PCI IDE cards, which manufacturer and model number, and which drives did you use for this setup?

[bhoar]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

One of the other things I've considered is playing around with alternate firewire and ATA drivers under windows, to work around the 800mbit issue and the PIO issue, respectively:

http://www.unibrain.com/Products/DriverAPI/ubcore.html

http://alter.org.ua/en/soft/win/uni_ata/

Once I get ULCLI up to the release v1.00, I'll return to experimenting with drive, IO rates and feature support.

-brendan

[pls1]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

Some on-MOBO SATA connections work and some do not. On board RAID MOBOs, at least versions i have, do not.

Not all PCI SATA cards work and not all drives work via SATA. It was a very frustrating and rather expensive month to get my set ups working . Beats me as to what is going on.

Perhaps we can put together a consolidated list of combos that someone has gotten to work as a FAQ

Phil

[bhoar]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

Yeah. Ok. But the fact that on-board SATA connections aren't working makes me believe this is a software issue of some sort.

-brendan

[pls1]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

brendan:
Most common (current) RAID oriented SATA MOBOs will not give dbpoweramp C2 detection off their on-board connectors. I don't know why but I spent enough weekend time convincing myself to just give up on certain machines.

Multiple PCI IDE cards (one card-to-one-drive) always work fine for me on any of my machines. Some eSATA PCI card combos work with the PX-755/760 and Samsungs but some don't. One USB plus multiple PCI IDE cards work fine on my quad core workstation with very highspeed ripping and low CPU utilization.

Every other attempted combo has either plain not worked or can't reliably deliver high speed ripping with consistent C2 detection over eight different other (MOBO) machines.

After Spoons' recent post I'm going to sell the Firewire enclosures, cards and cables on ebay.

Phil

[bhoar]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

There are instances where c2 will not work in EAC but will in dBpoweramp (atleast on our systems), so it is half a dozen of one, half a dozen of another...

Yes, but asking "why" here is an important thing to do to reduce the headaches that a lot of users go through trying to get this right. If you (spoon) finally know why this happens, perhaps you can tweak the code so that it always works for dbpoweramp in both the current positive circumstances as well as when it has only historically worked for EAC.

It might be as simple as the buffer sizes used at some point in the chain, which might be too small for certain drive data to be reported back correctly or too large for certain interfaces/drivers to handle correctly.

So much hardware and so many drivers out there are supposed to follow the T10 recommendations and just get it completely wrong.

I recently built a system - quad core, 4 drives on PCI IDE cards (darn SATA MB's), on that system you could rip all 4 drives at x40 ripping speed (at the end of a disc) with less than 5% CPU usage (when using test conversion). Using Firewire, or USB you will quickly fill the systems bus allocation for these when using multiple drives.

That still doesn't explain why I see the problem with any mix of the following: on-board SATA, on-board firewire, PCI firewire, on-board USB. If the problem had ever gone away in any circumstance, I could understand, but surely the situation with all on-board SATA (tried in both native and emulation modes) parallels your all PCI IDE cards well enough. I don't think that's the explanation for what I was seeing, but I do need to start from scratch with experiments on that hardware again.

-brendan

[Spoon-]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

There are instances where c2 will not work in EAC but will in dBpoweramp (atleast on our systems), so it is half a dozen of one, half a dozen of another...

I recently built a system - quad core, 4 drives on PCI IDE cards (darn SATA MB's), on that system you could rip all 4 drives at x40 ripping speed (at the end of a disc) with less than 5% CPU usage (when using test conversion). Using Firewire, or USB you will quickly fill the systems bus allocation for these when using multiple drives.

[bhoar]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

I can get four drives going with four separate R12 instances at X22+.
...
I'll bet you will run into all sorts off hardware/interface quirks with multi-drive batch rippers.

One thing I want to clarify for you about how the batch ripper works is that it basically does the same thing you do with R12. The batch ripper launches a separate instance of the dbpa cd ripper process for each drive. Normally they are hidden, but you can always unhide them and bring them to the front if you think there is an issue. Any problems seen with the batch ripper would therefore be seen with the standard R13 ripper (and vice versa) under this architecture.

If you look in the task manager, you'll see one process for the batch ripper, one process per drive for each instance of the cd ripper and then, of course, a bunch of encoding processes, one or more per track currently being encoded.

-brendan

[pls1]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

That's great news since this means it is a bug that is "fixable". Multiple firewire ports will be much more palatable to folks than multiple esata or expensive USB interfaces.

Thanks for the persistence.

Phil

[boneshaker335]

EAC is able to detect C2 errors with my PC setup!

I made another test:

Exact Audio Copy (current version) is able to detect C2 errors with the same exact PC/external case firewire setup that dbpoweramp fails to detect any C2 errors.

So the hardware seems to work.

[pls1]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

Try to not use USB/Firewire at the same time as SATA. Then use only two drives per sata PCI card (or MOBO SATA interface). Better yet one interface per drive sending each ripping stream to a fast HD that is different from your systems disc using a quadcore processor.

My MOBO uses the nvidia chipset since there is an elusive bug in the intel chip that slows the fastest discs in RAID zero like configs. I can get four drives going with four separate R12 instances at X22+. Also make sure that the SATA card can handle the full 3 gbs. Many cards really don't.

This is only partially OT since I have lost C2 detection due to overloading the interfaces on my P4 machine.

I'll bet you will run into all sorts off hardware/interface quirks with multi-drive batch rippers.

I hope Spoons is not as touchy as the TOS enforcers over at HA.

Phil

[bhoar]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

From my systems engineering experience "kill the problem with hardware" may not be elegant but is usually cheaper and easier from the "outside-in" unless the programmers have and show you secret dials somewhere. At $50-$75 per drive for adapters, interface and cable this seems like a no-brainer for a commercial operation.

I was hoping the "throw as much hardware at it" approach would work, which is why I was surprised.

FWIW, the experiments I was talking about with more than 2 or 3 drives causing all drives to slow down was with each drive on a separate channel (tried all Firewire, then all USB, then all SATA) using the "Test" codec that throws the data away and doesn't write to the hard drive, on a core 2 quad CPU with 8GB of RAM on Win 2003 Server EE. The problem continued to exist if I mixed and matched interfaces (one on USB, two on SATA, one on firewire, for example). My gut feeling is that it is a DMA or interrupt issue of some sort in Windows. I'm going to try a few different windows installations to see if I can get the problem to go away.

Of course, now I'm going off topic, sorry.

-brendan

[pls1]

Re: C2 erorrs not detected when using Plextor drives via firewire (Oxford 911 chipset

I don't use batch ripper (i use multiple instances of R12 on XP). From "black box" monitoring R12 on my overclocked quadcore workstation, I don't think R12 can make optimum use of an individual interface-to-PCI bus in terms of latency and throughput. Of course, just getting the low-level drive firmware to engage with dbpoweramp must be challenging enough so this is NOT a knock at Spoons.

Just casual observation but multiple cards, each with one drive, give the most consistent high speed (x18-24) on multiple simultaneous drives, while two drives on the same interface aren't consistent even though multiplication of expected through put is below the max of the spec.

For batch ripping I would really recommend the newer Intel CPUs over P4s. Casual observation also indicates the same leverage in performance that the typical video encoding benchmarks showed. The P4s are much less predicable in full CPU loading with multiple drives going. Also total L2 cache seems to make a difference although this seems to be a second order effect.

From my systems engineering experience "kill the problem with hardware" may not be elegant but is usually cheaper and easier from the "outside-in" unless the programmers have and show you secret dials somewhere. At $50-$75 per drive for adapters, interface and cable this seems like a no-brainer for a commercial operation.

Phil