PS/2 CD "Burnfest" 2005/2006
A little later on I changed what I had plans to do. I first learned that for whatever reason, the Yamaha drive and a variety of IBM MCA SCSI adapters weren't getting along. For whatever reason, attaching the Yamaha drive to the external port of an IBM MCA SCSI controller resulted in the controller failing to work at all. I tried several other MCA SCSI controllers with a PS/2 Server 95a running Windows NT 4.0 SP6a. For whatever reason, only the lowest performing adapter (a Future Domain MCS-700) would work to burn CDs. Others would cause the process to fail almost immediately. I did a lot of diddling around with swapping cards, reloading drivers, reapplying the NT service pack and installing Force ASPI...all to no avail. I stopped the whole process shortly after finding that the Future Domain card would allow up to a 4X burn, but that it absolutely floored a Pentium 90 CPU in order to do so.
On the clone front, I found that the fastest I could drive the Yamaha burner appeared to be 12X from an Adaptec 1460 SlimSCSI PCMCIA card installed in a Dell Latitude D800. I figured this probably represented the best performance that the burner could offer, as a 2.0GHz Pentium M (Centrino) ought to have no problem running a PCMCIA SCSI board to its limits.
Finally, I came back to the whole idea of burning a CD on MCA hardware, long after the frustration had subsided. This time around, I set up a PS/2 Model 9585-0XF with Windows NT 4.0, a 486DX4-100 upgrade CPU from Intel, 64MB RAM, onboard IBM cached SCSI, a Seagate Barracuda ST12550N hard disk, a Madge Smart 16/4 MC Ringnode network adapter and an NCR SIOP SCSI adapter. The onboard IBM SCSI would handle the hard disk and booting the OS. The NCR 53C710 SIOP board (card-ID 01BA) would handle the Yamaha CD burner. I got the OS running and updated before installing Force ASPI and Nero Burning ROM 5.5.
To test the burning part, I prepared about 650MB of files. Some files were big, and others were smaller, just to help provide a balanced test. Big files generally don't fit into a program's buffer, although they cut down on the overhead of being found, opened and transferred to CD. With smaller files, the files do generally fit into a buffer, but the computer and CD burning software must find, open and transfer each one, which adds some delay in the burning process.
With the data prepared and a blank CD-R loaded, I fired up the burning software and told it to simulate the production of an ISO-9660/Joliet formatted disc. I picked the 8X speed option, and started the process. A short while later, the simulated burn process had completed without incident. While it ran, the software-based buffer of 32MB never dropped below 75% and the CD burner buffer stayed at 98%.
The 8X burn didn't make for a usable system. I didn't find that to be a big deal, as I could go use another computer, and it really is a good idea to just leave the CD burner alone while it runs. While I did not try it, a 12X burn may well have been possible even without turning the buffer underrun protection on. All things considered, the results shown here are pretty respectable.
A Word About Adapters
If you are familiar with the MCA bus and adapters that are available, you'll notice that I tried to select adapters capable of busmastering or streaming data transfer. Adapters on the MCA bus can (but do not have to) be intelligent devices with their own CPUs and controlling software. While the OS in use doesn't always support these features, an adapter that can work by itself without calling upon the main system CPU for help can ease the load on a heavily-worked system. While I don't know if the NCR 01BA SIOP (SCSI I/O Processor) adapter supports busmastering, it does support data streaming across the MCA bus if you elect to turn it on in system programs.
For this system's network connection, I picked a Madge Smart 16/4 MC Ringnode adapter. This is a token ring adapter, which isn't a big deal since I have two running TR networks, one for 16 megabit stuff and the other for 4 megabit stuff. (Both are linked to my Ethernet stuff, so every computer on the network can see every other without issue.) This adapter is running on the 4 megabit side. However, unlike the very common IBM "shared RAM" token ring adapter, this can busmaster and supports data streaming features if you enable them. If you have data residing elsewhere on your computer network that you wish to burn to CD, it's probably going to be a large amount of a data. An adapter capable of busmastering and data streaming can ease the load on the system and let you use it for other tasks while you transfer data in preparation for a CD burning session.
The point of this discussion is to help you decide what kind of MCA adapters you might wish to populate a CD burning PS/2 with. While at first you may have to use what you have, there is no reason why you can't look out for better performing options and hardware in the future. It will pay to do so for the most part.
If Things Don't Work
Unfortunately, as the experiences I haven't gotten into would show, burning CDs on a PS/2 can be quite an adventure. (It can be an adventure on any other type of older computer too!) Sometimes things don't work, and the failures are not easily explained.
If you run into this, my suggestion is to change your plans. If you have another PS/2 around, and it has a reasonably powerful CPU, give it a try. It may work where the other system would not.
Force ASPI on Windows definitely can help. You can try to burn CDs without it, and it might work. If it does not, try this before giving up.
Very modern versions of CD burning
software may really not work, or demand more CPU power than you can
easily get if they do work. I'd heartily recommend use of Ahead
Software's Nero Burning ROM...for me it has been a high quality CD
burning software package. Versions 4.0 through 6.0 all work well on