This evening I tried a little esperiment.
With the original board, P III CPU etc, I removed the two graphics/display cards and replaced them wih the best PCI graphics card I could find, an old HD 4350, which even sports and HDMI output as well as DVI and VGA.
Before doing is, I took some waveform update rate measurements using Agilent's own test criteria documented in the scope's data sheet. This entails turning off all measurements, single channel, no deep memory, no sin x/x and no joining the dots, I achieved pretty close to book figures of about 3.2kwfm/s. Turning on joining the dots halved that.
After loading the drivers, the main scope app runs and thinks it's an MSO8104A, so a good start.
This time, the waveform update rate under the same criteria was up to about 4.1kwfm/s, but this time joining up the dots made little if any difference, so it was using the card's own interpolation.
I did a few other tests, to see how reliable it was, and to check out the intensity grading, which is quite an improvement. I couldn't find any problems, so I increased the resolution from the stock 640x480 to 1024x768 which is standard for an MSO8000 series scope. In the past this has proved unreliable when using equivalent time sampling, but this time I couldn't get it to fail at all, which leads me to believe that the MSO8000 might well use the same acquisition hardware as the 54830D series after all.
Next step is to see if I can fit a third party LCD. I have both a VGA and an XGA panel with the right screen size, and VGA/DVI/HDMI driver boards for them, the problem is whether or not they'll physically fit.