But 1600x1024 isn't
W---I---D---E screen it's a letterboxed
s-h-o-r-t s-c-r-e-e-n format that looks like an excuse for using lopped off LCDs that had too many bad pixels at top or bottom edge. [}
]
Uhm, no. We get that you think anything less than 1600x1200 is a periscope, but really, in the 1990's that was a pretty high-end resolution even on "Workstation" hardware. SGI's contemporary workstations typically ran at 1280x1024, and last I checked 1600 is wider than 1280. Also, when this monitor came out an LCD panel with a resolution higher than 1024x768 was a (rare, practically unobtanium, and very expensive) novelty. So, yes, in context it certainly was
*widescreen*.
That Sonata card's pretty interesting, but dollars to doughnuts its RAMDAC won't support 1920x1200
Why do you think the RAMDAC would be the problem here? It looks to me as if that card would have *plenty* of performance overhead to handle 1920x1200. As a refresher: What the RAMDAC does is read the VRAM at a rate accordant with the current set of video timings and convert the data into analog signals which get pushed out to the monitor. Or, in other words, what the RAMDAC limits is your *pixel clock*. That datasheet says the card has a 200mhz RAMDAC. So, how much performance do we need to do 1920x1200@60hz?
1920x1200x60=138,240,000
So, there you go, its 200mhz RAMDAC should *easily* do the needful. (Note that this is a really crude oversimplification.) The datasheet says it supports 1920x1080@69hz, which would have a pixel clock *very* close to 1920x1200. It also claims to support ridiculously much higher resolutions like 2432x1712, which would have way higher pixel clocks. What I do find something of a headscratcher is that it claims to support that resolution in "thousands" of colors, IE, 16 bit, which I call foul on because the card only has 4MB of RAM. 2432x1712=4,163,584. So doing that resolution in *
8 bit color* requires just barely under the 4,194,304 bytes you get in 4MB. Someone made a boo boo...
And, actually, doing that math points out the *real* problem with 1920x1200 verses 1920x1080. The latter is just barely under two megapixels, while the former is closer to 2.3. So... even though I would assert that the Sonnet card would be perfectly capable of running at 1920x1200 assuming the driver could be convinced to program the hardware to do it, it *would* be limited to doing so at 8 bit color depth instead of the 16 bit it supports with 1080i.
One day I'll get my 1988 SuperMac card with its amorphous RAMDAC setup to run some widescreen resolutions, but it's 2MB VRAM limited.
Yeah, so that same calculation applies again. With 2MB you theoretically might be able to convince it to do 1080i at 8-bit color, but you don't even have enough RAM for that with 1920x1200. I guess old versions of MacOS do support 4 bit modes, you've got enough for that assuming the RAMDAC is actually fast enough.
Is this the card where you've talked about swapping crystals to make it do weird resolutions? Again, remember, on newer video cards (almost anything made since, I dunno, 1993 or so) that's not a thing. Almost everything uses PLL clock generators to generate sync and pixel clocks. The primary limitations are the range of the PLL, the speed of the RAMDAC hardware, and the bandwidth of the VRAM. (Remember, even running at 8 bit color a 1080i display requires the hardware to push over 120MB a second from RAM out the monitor port. I believe that's about four times faster than the RAM->CPU bus in a Macintosh SE/30.)