poobah

In the 'classic era': OS/2: Easily the best OS of its time, what Windows should have been (literally) MacOS 9 Windows 2000: Stable and solid, a little weak in the driver department

No invoice here

I believe the 3dfx Banshee card actually had the fastest 2D engine of the era. All 256 windows raster ops were implemented in hardware, and if I recall correctly, it was one of (if not the first) to hit the maximum theoretical Windows 2d performance. It had some ungodly wide internal 2D engine (128 bit?). It was significantly faster than the awesome Tseng Labs ET6000w32, which had been my 2D card of choice before then. 2D was really the banshee's only saving grace, since it was missing the 2nd texture unit from the Voodoo2. Perhaps you are thinking of the Riva TnT, which had similar 2d performance to the banshee, voodoo3, and an assortment of Matrox cards, as by then pretty much everyone had maxxed out 2D performance.

I have the 2.0 version of that printer, it is a good value, and lots of people making bits and pieces to improve it. Adding diagonal braces, better bearings and a borosilicate glass bed really helped it make great prints

Time to fire up the printer and do a test print...

http://www.uclinux.org/

I'd take one

We used to call those crashes "random NEW"

Pretty sure i have some, any brand/model preference?

I have both. The NX has a built in USB hub, the VX does not. Otherwise, they appear identical to me.

Somewhere (and sadly I couldn't find it quickly), exxos details the great difficulties he had finding genuine PALs or GALs. Most of them had been remarked to faster speed.

Multi-track sampling is a whole different animal, I doubt there was a PC-based solution for it during that time period. The "high end distortion" comment seems odd, and the measurement suspect since the RAP-10 doesn't have a PA on it.

Lightning.... ugh Well start with the basics... do you have clock signal, does it come out of reset, do you see the activity on the bus, go from there

Define dead =D

So definitely check the voltages with it loaded down before you hook it back up to the ST. Often times when you lose a component like that, it can take other stuff with it.

Awesome link, it appears my memory is a little fuzzy. Looks like the TB card (barely) edges out the RAP-10 on most of the measurements, though they both have impressive specs. At any rate, both cards are evidence that you could indeed do professional music work on PCs of the era.

Rev. H TT030, others should be similar

The RAP-10 was essentially a Sound Canvas on a PC card. It wasn't meant (nor did it provide much support) for PC gaming. Serious musicians used them. I'm pretty sure the RAP part stood for "Roland Audio Producer." The Turtle Beach Multi-sound was an excellent card, but the RAP-10 had a better SNR. (For context, "back in the day", a good friend of mine used the RAP-10 & Cakewalk for professional music gigs.) So yeah, expensive as all heck, but a good bit past "a pipe dream".

Ahh yes, my old friend serial-bus data frame checksum error, and his buddy, Error 138, Device timeout.... How I loathed you both. Somewhat amusing anecdote.... I actually had an occasion to indicate a malformed serial transmission with some development hardware, can you guess what I wrote to the log?

The Roland RAP-10 was out in '92 or '93.

I guess what I was driving at, is that there are a lot of potential noise sources for the vanilla ST sampled output, and you sort of nibbled around the edges, but there's really a lot more to it than what you've researched. The distortions you have identified are definitely there, but not for the reasons you might think. You can generate reasonably good output with the vanilla ST hardware (or worse). Certainly, the limitations you correctly identify don't make it easy, but it really is more about the source material and how it is manipulated than the exact specs of the hardware. Would better sound hardware been a good move on Atari's part? Absolutely (hence the STe DMA sound).

Yup, nowhere near enough current.

DisplayPort is a bear in an FPGA, you need way more resources than with HDMI. I'm only aware of 1 publicly available implementation.

OK, I'll bite You really are conflating a bunch of different things and painting a worse picture than reality (not that the vanilla ST sample playback is stellar....) 4 bits of amplitude is certainly more coarse than 8 bits, but, given good samples and a decent sample rate, it could still sound fairly good. For instance, although CD's are indeed composed of 16 bit samples (well almost always, but that's another day), those samples are often played through a 1 bit DAC. Yes, much higher sample rate than our lowly ST sound chip, but by the same argument you present, 4 bits is WAY more than 1, so the ST is obviously better than CD quality? That said, the ST isn't massively oversampling on playback, so the limited quantization levels certainly can come into play, again depending on the original samples. Much like mixing colors or multiplexing sprites, toggling amplitude in the samples 'fast enough' can create the illusion of additional quantization levels. Also, keep in mind that we really discern amplitude as power, so there's a time factor involved as well. Regarding filtering, that's really going to come down to how the original samples were crafted and what rate they are played back. So long as our original samples don't have any frequency content greater than 1/2 our playback rate, you won't have any aliasing, and you won't have to rely on filters to address it. None of that is to say that vanilla ST is some kind of sample driven powerhouse, it certainly is not, but given the hardware limitations, the end result depends greatly on the source material and techniques, and only somewhat on the actual sound chip (and in truth, this is because of the limitations of the sound chip)

I'm in Ohio, I can mod yours if you like.