Thank you very much for spending so much time to respond to my immature opinions, I feel being honored.
The loss with bigger than >1ECC sampling size are:
- loss of true minima and maxima (they become rounded-off estimates)
- The above also leads to a more smooth graph overall (removing spikes)
- Inability to check for reading of the disc to be within DVD speficiation (<=4 PIF/1ECC)
Granted, those are not neseccarily big losses in overall compatibility scanning.
I can only wish that I could explain things as clear and sound as you did.
I've been trying to find a suitable test method for cd- and dvd-/+r scanning as well (for the past three years).
I've also entertained the idea of finding a one 'picky' reader.
However, here's where the problem starts.
1) There is not just one way to make an 'incompatible' disc or a 'picky' reader. A disc may be readable in drive A, not in B. With another disc the roles are reversed. There is NO 'universal picky' reader that I've been able to find. This is due to way the dvd reading is left to the implementation of the drive more than that of cd reading. There are more than one way to read a dvd disc 'correctly', leading to different compatibilities and incompatibilities.
The information is invaluable to me. Being music oriented, I have limited experience with DVDRs, but I knew the possibility is low from the start. It is good to get the answer much less than three years, and to save money also. But do we still have a chance if the +R or -R drives are considered separately? If not, then I will turn my search instead to drive/disc combinations that will produce DVDRs playable by most or all drives. Do you know any such a writer? How about CD drives?
Remember, errors are not on the disc. One cannot expect results between drives to be similar, because the drives influence how many errors are produced in the reading process.
I take issues with the opinion that all of the errors are not on the disc. Jitter errors, eccentricity, asymmetry, etc. may be largely on the disc. The cases with C1/C2 errors are more complicated, as they are mostly the results of disc/reader combination. I did not say that the errors are on the disc in the present case, but for some types of copy protections, C2 errors are incorporated into the disc. And there are cases where C1/C2 errors are primarily due to defects on the disc. If you want some examples, I can show you.
Erik has not stated that the unit of measurement for error rates on DVD discs within CD Speed is bytes. It's just that one drive that is known to report each erroneous byte in a PO row.
We don't really know what CD speed reports.
This has been asked from Erik, but no straight answer has been gotten yet.
Does it report the basic sampling size error counts without calculation? Apparently not, because Erik has started 'calibrating' results across various drive models.
However, do they all report 8ECC floating window SUM for PIE and 1ECC PIF - or statistical averages of those?
Apparently not, because there are no units on the graphs and Erik hasn't replied on this issue either :)
I have re-read Erik's post
about this matter, and both of us are not entirely correct. I have extended Erik's statement of the unit of measurement too far. I am waiting for his words about the units for other chipsets. On the other hand, it is clear to me that CD-DVD Speed plots each 64 ECC sum returned by the chipset in this case; you seem to have overlooked this point. Perhaps previously I should have also mentioned another post
I came across. It is obvious then that PI errors reported by ASUS E616P2 at 5X are not significantly higher than those given by CATS when normalized, even if we assume that both use row/sample as the unit for error rate. It is unfortunate that the ASUS drive does not return PIF.
I think Erik has stated that E616P2 drive reports not PIF (or even PO), but erroneous bytes in PO columns (summed for 64 ECC blocks).
I do not understand what you mean. Erik said that it reports PO errors.
To put it shortly: I think the motives are pure and the goal is lofty, but overall I think it is folly.
Drives will see different amounts of errors in reading of a disc. Some drives are good at reading certain types of discs, other drives other types of discs.
There is NO WAY that both drive types can report the same error rates for the two different types of discs.
Yes, it is possible to somewhat correct for these differences, but it makes no sense, imho.
The point is to show how many errors did a drive see in the reading, while letting the characteristics of the drive show through in the error rates.
But showing somewhat 'universal' quality score is impossible, because there is no known universal one correct way to read a dvd disc (and as such, to measure it for errors in reading).
Also, basing this calculation on PIF is also false, imho. NIST has already shown with their testing that the crucial indicator of disc compatibility is it's reading error rates being below 280 PIE (in a CATS scanner).
So, there are two problems.
1) Disc quality score is not _disc_ quality. It is _disc compatibility in the drive doing the testing_. Errors are not on the disc.
2) PIF Is not the only or not even necessarily the best indicator of disc compatibility. PIE should be included as well. It could even be the determining factor, because it is a lower level error rate, not taking into account the PI row error correction strategy of the drive (which PIF/PO do).
That's all for now. I hope I didn't bore you.
As a final note, I have great respect for Erik. He's doing something I don't know how to do.
Still, that doesn't stop me from disagreeing with some of his assumptions, when I see them being clearly in opposition to the best scientific data I've been able to find on the issue.
I have not seen any discussion on Erik's attempt to normalize quality scores before, have you? I have doubts about its accuracy, too. But before his effort, with error test what being measured were reading abilities of readers rather than qualities of disc in most cases. So I welcome his endeavor. It is beyond me to comment on its feasibility. He has been responsive to requests and questions, including our concerns about units of measurements. I believe that he needs feedback to make improvements, if possible. Undoubtly he will cancel it if proven to be useless.
Of course, I'm fallible myself too, so I can be wrong as well. I don't want a flame war, just a bigger understanding of the issues involved.
Flame war? No. I am interested only in finding the truth. Besides, I am still in the process of learning and I have learned a great deal from you here and elsewhere.