CD/DVD Drive Accuracy List 2008

[thope]

Re: CD/DVD Drive Accuracy List 2008

Although the statistics are very interesting, I don't think this is a very good test. There are a lot of variables unaccounted for in this test a better test would be to take a relatively damaged cd and use a trusted control group to test drives on the same disc, then we'll have a good representation although there are factors that depend on the drive's condition and age.

[HiPerFreak]

Re: CD/DVD Drive Accuracy List 2008

Isn't the Pioneer 115D and 215 exactly the same drive except the interface and RAM burning support? How is it possible that the 215 belongs to the best and the 115 to the worst drives? I have a 215D and it would be interested, if it is like the 215 oder more like the 115D?

[BruceH]

Re: CD/DVD Drive Accuracy List 2008

Hi All,

Is there a way to know which of the top drives in this list is external (USB-attached)?

I read that LiteON was generally good; so I bought a DX-20A4PU. It seems to read data CDs ok and to burn ok. However, it errors out with "unreadable frames" on at least one track of practically every other CD that I try to rip with dBpoweramp. This does not strike me as normal.

Any ideas for a good external drive to dedicate to ripping? Internal is out of the question because my office Core2 Duo laptop is my only machine.

Thanks,
Bruce

[Spoon]

Re: CD/DVD Drive Accuracy List 2008

If your system has any sort of virtual CD drive which is used to rip a CD your results will not appear in the database.

[Zn0rt]

Re: CD/DVD Drive Accuracy List 2008

Only in the context of drive accuracies, results from multi-rive systems go into the database, they are just not used when creating the accuracy lists.

Does this mean that if I have one physical CDRW drive and one and or more virtual/emulated CD drives (ergo Daemon Tools to mount ISO files virtually) that my AR results won't be used? Oh, see, thats sad.

-Z

[Spoon]

Re: CD/DVD Drive Accuracy List 2008

Only in the context of drive accuracies, results from multi-rive systems go into the database, they are just not used when creating the accuracy lists.

[BugsBunny]

Re: CD/DVD Drive Accuracy List 2008

RE 5. Anyone who has 2x different drives on the same computer (most ripping companies), their results cannot be used as AR has no way of tracking results from a drive only a computer - integrity is maintained by purging a whole computers results from the db if a miss configured offset is highlighted. This drops about 30% to 50% of submissions from being used.

And I was wondering why a CD I test-ripped did not appear later on in accuraterip. Does this mean that even if all drives on a computer are correctly configured the submission is dropped by accuraterip?
Is there an update planned that will allow PCs with more than one CD drive to submit the results to accurate rip?
To simply drop 30%-50% of the submissions is really a big deal! If there can be impelmented something to accept these submissions, that definitely should be done.

[Spoon]

Re: CD/DVD Drive Accuracy List 2008

>If this is the case, there's really no way to tell on a per track basis if a
> match/mismatch is more likely an disc read error vs. a different pressing...correct?

We go off: if all tracks missmatch it is a different pressing (or a horribly damaged CD).

>pressings are the same audio material pressed at a different offset...correct?

Correct

>2. Are CD-Rs removed from the records and/or blocked from submission?

No

>3. Are low track count CDs removed from the results (e.g. CD singles, CD3)? These CDs

Not a problem with AR (ID collisions) as the new discs go in as different pressings, even though it is a totally different disc.

[bhoar]

Re: CD/DVD Drive Accuracy List 2008

Now on the suggestion of trimming before averaging -- consider how a single user like myself could impact this, and why you don't want it to happen...

Right. This is why it might be worth taking my first recommendation and remove some number of outlier results for each drive.

-brendan

[Porcus]

Re: CD/DVD Drive Accuracy List 2008

Nice statistics.


Now on the suggestion of trimming before averaging -- consider how a single user like myself could impact this, and why you don't want it to happen.

My PX-230A's had about all my troublesome CDs. That is, I have ripped about 100 CDs of which some 60 had issues with other drives, and 30 still had issues with the 230A (some of these were aborted, some were not in AR). For my submissions to produce a representative picture of the 230A, I should also have run the 5000 other CDs through it. Which of course will not happen, as they were run through my Sony changer without errors or warnings.

[bhoar]

Re: CD/DVD Drive Accuracy List 2008

Spoon -

1. I take it that the API to AccurateRip, standardized to be used from more than one program, does not facilitate a method for passing rip-process information to the database (e.g. burst vs. secure, secure settings, C2 error flags encountered, mismatched reread counts, etc.).

If this is the case, there's really no way to tell on a per track basis if a match/mismatch is more likely an disc read error vs. a different pressing...correct?

In addition, I believe most of the time different pressings of the same CD end up with all tracks not matching existing pressings. This is because, most of the time, different pressings are the same audio material pressed at a different offset...correct?

If so, there may be some easy heuristics to drop "very likely pressing related" mismatches from the results. e.g. if *all* tracks failed to match, but same user+drive also ripped accurate results the same day, then this set of inaccurate matches should not be held against the drive.

Of course, the pressing issue would also be solved by your work on AR2 in the long run.

This heuristic would also reduce skewed results in the drive ratings due to rippers who have large collections of rare/esoteric CDs (and are therefore often the initial and only submitter).

2. Are CD-Rs removed from the records and/or blocked from submission?

If so, is this by policy (all calling programs are required to not submit CD-Rs) or by code (AccurateRip DLL checks the source disc)?

If not (or if the calling program does this wrong), this too could be a significant source of noisy results for certain users and could skew drive results significantly.

3. Are low track count CDs removed from the results (e.g. CD singles, CD3)? These CDs are very likely to have incorrect cddb ID matches, are they more likely to have incorrect AR matches (hash collisions with unrelated CDs)?

---

My points are not meant as criticism, just some thoughts that might help in the future with AR2 or AR3...

AR, in terms of the track hashes alone, is an incredible asset for you. But the drive ratings could also become a useful asset as well.

-brendan

[bhoar]

Re: CD/DVD Drive Accuracy List 2008

RE 1.

Best
Drive: TSSTcorp - CDDVDW SH-S202N (51 users): 907 Accurate Submissions 13 InAccurate Submissions 98.5870 % Percentage Accuracy

Worst:
Drive: TSSTcorp - CDDVDW SH-S203P (53 users): 1754 Accurate Submissions 98 InAccurate Submissions 94.7084 % Percentage Accuracy

Similar user count, the latter did have a high number of bad submissions, really when the users is 53 it is a little low, one user could influence the results by 2%

Yeah. But the N is supposedly the B with Lightscribe added, while the P is supposedly the D with Lightscribe added. It just jumps out of me that the B/D/N are all clustered up in 98.x, while the P is at 94.x. It points to some large error bars in the survey (caused by source disc quality, perhaps), IMO...something to keep in mind.

RE 5. Anyone who has 2x different drives on the same computer (most ripping companies), their results cannot be used as AR has no way of tracking results from a drive only a computer - integrity is maintained by purging a whole computers results from the db if a miss configured offset is highlighted. This drops about 30% to 50% of submissions from being used.

Ah, ok. Storing and submtting the drive maker/model/firmware data in the next revision of AR might be useful. I doubt it would increase the database size much after processing, since you could use an index table for all unique values (perhaps a 16-bit value...or if there are more than 64k maker/model/firmware variations out there, 32-bit might be safer).

-brendan

[Spoon]

Re: CD/DVD Drive Accuracy List 2008

RE 1.

Best
Drive: TSSTcorp - CDDVDW SH-S202N (51 users): 907 Accurate Submissions 13 InAccurate Submissions 98.5870 % Percentage Accuracy

Worst:
Drive: TSSTcorp - CDDVDW SH-S203P (53 users): 1754 Accurate Submissions 98 InAccurate Submissions 94.7084 % Percentage Accuracy

Similar user count, the latter did have a high number of bad submissions, really when the users is 53 it is a little low, one user could influence the results by 2%


RE 5. Anyone who has 2x different drives on the same computer (most ripping companies), their results cannot be used as AR has no way of tracking results from a drive only a computer - integrity is maintained by purging a whole computers results from the db if a miss configured offset is highlighted. This drops about 30% to 50% of submissions from being used.

[bhoar]

Re: CD/DVD Drive Accuracy List 2008

Finally in 2006 the results were questioned as these drives:

Drive: PLEXTOR - DVDR PX-740A (131 users): 3505 Accurate Submissions 69 InAccurate Submissions 98.0694 % Percentage Accuracy
Drive: BENQ - DVD DD DW1640 (213 users): 2510 Accurate Submissions 38 InAccurate Submissions 98.5086 % Percentage Accuracy

Were effectively the same drive by had differing results, not any more.

Spoon,

1. Looking at the data imported into excel to play with a bit, I noticed the TSSTcorp (Samsung) drives are all over the map, even within the same model family. E.g. the SH-S20xy (x=2,3,4 for interface and y is a feature-set related letter) models, which I am concerned about in particular.

I wonder if it might make sense, before collating, to remove the top and bottom accuracy submitters (perhaps 2-3 at each end) for a particular drive - that is, the outliers...or perhaps weigh them differently.

I suspect some of these results might be skewed because a small number of users have exceptionally clean or exceptionally dirty collections or perhaps severely worn out drives, and, if they are large collections, they may be skewing the results.

Thoughts?

2. I was also hoping to see some comparison of SH-20xy vs. SH-22xy, but the latter must be too new to make the cut offs...

3. I'm familiar with the pathetic behavior of the GSA-S10N, having a few myself. I'm much more surprised that Pioneer allowed the atrocity that is the DVR-115D on the market. It seems that as time goes on, Pioneer drives are doing less and less well with audio CDs.

4. I'm seeing a lot more _NEC units than I would have expected. I wonder if these are built in to some of the most popular desktop/laptop models out there?

5. I wonder if the ripping companies using the batch ripper are feeding their accuraterip results back in. If so, the numbers for that market don't seem very good...

-brendan

[Spoon]

Re: CD/DVD Drive Accuracy List 2008

Finally in 2006 the results were questioned as these drives:

Drive: PLEXTOR - DVDR PX-740A (131 users): 3505 Accurate Submissions 69 InAccurate Submissions 98.0694 % Percentage Accuracy
Drive: BENQ - DVD DD DW1640 (213 users): 2510 Accurate Submissions 38 InAccurate Submissions 98.5086 % Percentage Accuracy

Were effectively the same drive by had differing results, not any more.