So I've had a frustrating start with my 812S. I thought I'd post my experience in case it's helpful to others.
My setup is an ASUS A7N8X Deluxe nForce2 motherboard, Athlon XP 2600+ (actually, overclocked 2100+), 1 GB RAM. 2 WD hard drives on the SATA controller, 1 HD hard drive, 2 CD burners and 1 DVD-ROM drive on the IDE controller.
On IDE primary channel I had a LiteOn LTR-48246S CD-RW drive (actually it's a 52246S underclocked as it gives me much better write quality on Fuji/TY at max speed) and Pioneer DVD-ROM DVD-105.
On secondary channel was a 24x LiteOn CD-RW burner and the 3rd hard drive.
I replaced the 24x burner with the 812S. However, the 812S is actually shorter than all the other drives, which was a problem as the IDE cable now couldn't reach it. So I swapped the positions of the 812S and the DVD-105 so I could plug everything in. This put the 2 burners on the primary IDE channel.
Everything seemed to go fine, except everything I burned with the drive was a coaster. Simulation and burns in Nero (latest version) worked fine but verification would always end at an unreadable part towards the 70-80% mark. I tried different firmware versions, different IDE drivers, settings, etc.
9 coasters later I finally figured out that the 2 burners just aren't happy being on the same channel. I don't really understand why this is, as only 1 drive was every being used at the same time. If someone has an answer to that I'd appreciate it if you could enlighten me.
Regardless, I bought a longer IDE cable and returned the drives to their former positions so the 2 burners are on separate channels, and the problem was solved. (I vaguely recall having similar problems when I added the 52x burner to the system, but had obviously forgotten the lesson when I installed this new drive.)
I've burned a number of CD-Rs now. They're RITEKG04's. I was a bit disappointed that my spindle won't burn useable discs at 8x or 6x (although I've read there's a lot of variablity between batches and I shouldn't be surprised). However, at 4x the burns and verifications are uneventful. Scanning with kprobe2 showed every burn was of very high quality. P0 generally below 20 and PI generally below 5, with 2 being the peak for most of the disc(s).
I also burned some video on to a Memorex 2x CD-RW disc (RITEKW01). I tried writing at 4x but the result was an unrecognizeable disc. I thought I might have destroyed the disc, but reburning at 2x produced a disc that scanned in kprobe just as well as the RITEKG04s. It was also succesfully recognized and played on my standalone, rather old Hitachi DVD player.
Less luck with a Verbatim Datalife Plus 4x CD+RW. Seems to burn fine at 4x or 2.4x and even verifies succesfully under nero. However, kprobe2 is unable to scan it, giving a "Test range error!" instead. I don't know what that means and this is the only disc that does that. kprobe 1.1.29 does manage to scan it and the quality is very good. However, the scan ends prematurely due to a read error. The disc is unrecognizeable in both the Hitachi standalone player as well as the Pioneer DVD-105 DVD-ROM player. It can only be read in the 812S and while it can be copied back to the HD performing some operations like binary compare between the disc and the original HD files causes the comparing program to hang, taking up all available CPU cycles.
I'm using the US0J firmware. I don't know if the Verbatim disc is just crap or if this is symptomatic of the CDRW quality problems observed in the full review recently posted, using the older firmware. The fact the Memorex CD-RW burns perfectly is encouraging, but I suspect the firmware may still need some work on CD+RW. I'd be interested in hearing other experiences with writing on these or similar discs.
I also played with using the Nvidia IDE driver or the standard Microsoft IDE driver. Generally I didn't notice much of a difference in performance (at least casually), although the Nvidia driver seemed to cause less of the short freezes I see when inserting a disc into one of the optical drives. What I did find, however, is that kprobe2 with the Nvidia driver wouldn't recognize the drives using the ASPI driver, but only the SPTI driver. So I ended up going back to the Microsoft driver for the best compatibility.
Finally, while playing with kprobe2 scanning discs on both burners simultaneously, I found the performance on both suffered when performing C1/C2 or P0/PI scans. This turned out to be due to having enabled realtime graph, which uses a lot of CPU time, for some reason. So with this enabled in both instances of kprobe2 there just wasn't enough cpu power. Disabling this feature allowed simultaneous scans at full speed on both drives with no problems.
However, performing performance scans simultaneously showed that during a DVD scan, if I start a CD scan the 812S would drop it's RPM several times while the CD scan was active. CPU usage was always below 70% and since the drives are on separate channels and the scan shouldn't require HD access, I don't really understand why this happens. Any ideas?
With the exception of the Verbatim disc, I'm very happy with the drive. Hopefully with a little more firmware tweaking and wider availability of 8x medium it'll be just about perfect. (Well, if it's flash upgradable to dual layer, even better...)
Oh, another question I had is why is the drive locked to 8x read speed for DVDR and DVDRW? Why can't it go to 12x? Any way to unlock that?