Reacting to http://www.hardwarezone.com/articles/view.php?cid=29&id=2146 today, I wrote the following (from my private e-mail account). I do remember fighting for this verification pass to be mandatory in the hardware, or the defect management scheme couldn’t work, despite the hardware folks’ pressure to do performance for review over reliability. Messaging it to reviewers is still pretty challenging. Luckily the dreaded competition from defect management-less HD DVD-RW isn’t taking place as Toshiba failed to deliver any retail recorder at this stage.
I read your review of some BD recorders and saw your confusion regarding the performance of BD-RE recording, in particular the delta between the performance announced and the performance measured. I actually have participated to the design of the BD specification and worked for Nero AG for several years. BD-RE is a hardware defect managed media, like DVD-RAM and the unreleased HD DVD-RAM. This means that in case of a defective block, the data will be relocated by the drive to a spare area. The computer software won’t see a thing – the disc will appear defect-free. However, to perform this relocation the drive needs to know what is the original data at this block. But if you are reading back and stumble on this defective block, you don’t know what was the original data. So instead the device will embed a step of verification after physically writing the block. If during this verification the block is found defective, the device still knows the data it attempted to write at this block and can safely relocate it to the sparing area. However the verification is basically a read*, meaning that the logical write operation (the operation visible to you and me, composed of a physical write and of a verify) will actual take twice as long as the disc rotation speed (the famous 2X) suggests**.
Sony is correct to state that this verification pass may be disabled, and actually the entire defect management, for video data. This requires however specialized software that will treat the video data differently and issue a so-called ‘stream’ write to the device (further specification in the device command set MMC-5 at www.t10.org under the WRITE12 command streaming bit). AFAIK, data mastering software do not implement this. In theory video recording software that stream video to the disc and need to ensure the stream is in-interrupted should use this.
The hardware defect management has the advantage of actually managing the defect away from the user, so that your burn works despite defective blocks (if any), while software verify pass would only be able to tell that the burn failed but won’t fix the defect (i.e. you have to toss the disc).
*: Some optimization of this step are possible, I don’t believe however that hardware implementing them has been released so far.
**: If the drive properly optimizes the seeking between write and verify. If not, then the write operation may be even slower.