Jump to content

Talk:AX architecture

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
[edit]

Hello fellow Wikipedians,

I have just modified 3 external links on AX architecture. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at {{Sourcecheck}}).

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 12:29, 1 October 2016 (UTC)[reply]

resolution

[edit]

Just wondering, are we sure it's 640x480 not 640x400? The latter was somewhat of an extant standard for Japanese and certain western machines, typically with 25kHz-class monitors... key example being the PC98 itself. Super EGAs only seemed to add 640x480 to that when VGA and MCGA came along (which is suggested as after the intro of JEGA) and had a bunch of other strange interim resolutions besides. The Hercules graphics adaptor even had an option for that and was only developed so the inventor could use their IBM machine with eastern scripts.

I'd expect given the CHIPS branded ICs on the board that a good deal of the actual parts that make the thing work are shared with western S-EGAs as well, down to the BIOS modes though maybe not the character set (dependent on how much of the kanji was from ROM, and how much drawn as bitmaps by software), and the single layer of output (vs the multiple layers with different resolutions and colour depths that the X68000, FMT and I think the PC98 could do, allowing mixing of hi-rez monochrome text with lower rez colour graphics). And the later S-EGAs were software compatible with the majority of VGA software in the hi-rez modes (as they tended to have a standard, or at least fairly high contrast 16-colour palette, so would be compatible with 64-colour digital output as per EGA and compatible multisyncs), so unless there was a problem of really really needing to use the exact same drivers as IBM VGA (in which case why not just use a generic clone card?), the only reason to make a whole new thing if you already had 480-line was to provide the wider colour palette, especially in low rez.

The other clue is in the onboard RAM in the pictured card, which is marked as four banks of 64Kx4 total (each with four, presumably one-bit SIPPs), for a total of 128KB. That's enough for a single 640x350 16-colour page with EGA (but typically you'd get 256KB for smooth page flipping), and also a single 640x400 page with the same colour depth (the practical minimum for a lot of purposes unless you have something odd like an early PC98 8-colour card which has three RAM banks giving 96K...), but not 640x480. Unless you drop to a somewhat wasteful monochrome mode, or a weird 4-colour one, you're going to need 256K on your card (or at least 160, like the PGC...), as per all VGAs. Notionally you could jam a 640x480 8-colour mode into 128K, but the arrangement of the chips is key - the architecture of such things tended to rely on separate bitplanes being fed into the output circuitry in parallel (rather more efficient when you have limited colour and low clock rates), and in this case we have four planes of 32KB each (or 256Kbits/256Kpixels). Natively they simply can't store more than 640x409 pixels or the equivalent (e.g. 512x512), and if you want to extend the size, they'd have to be chained into two pairs of 64KB (giving 512Kpixels at a 2-bit depth / 4 colours)... and then you could take the resolution a good bit higher for an even clearer image (e.g. a MacII-like 832x624 may be doable, or at least 800x600, just either with some flicker or needing a higher clock frequency such as the mid-30s MHz used on higher end S-EGAs), though maybe not to the same lofty heights as the near-1Mpixel PC98 HiReso board.

It does have a true 25.175MHz clock on board, which is curious, given that it's exactly what VGA's 640-pixel mode runs at (maybe it's a fairly late JEGA and the originals used 24MHz or similar, unless IBM copied their clock rate straight up or it's a very strange coincidence - there's no real reason to use it outside of the way IBM chose to structure their modes to be NTSC linebuffer compatible) ... but that clock is used for all of 480, 400 and 350 line modes at that width. It just means you don't get 720-pixel width without some trickery (as true VGA has a separate clock for that... but very, very few applications ever used that as a graphics mode rather than single-byte block text).

Besides which 480 is a strange linecount to go for other than wanting square-pixel 4:3 graphics without letterboxing. For text, one presumes the aim was 40x25 kanji to match the industry standard 80x25 ACSII modes. 640x400 will give you a nice 16x16 pixel block to work in with that layout (vs 8x16 for ASCII on VGA, or 8x14 for EGA). Presumably 16x14 wasn't quite enough for clarity, or the reduced number of lines with the same font size was considered unacceptable? But 480 doesn't divide into 25... some other machines used 475-line with a 19-pixel high character block, which is a bit strange but presumably it helps... if you're going to make a custom mode for that purpose, as neither 350 nor 400 are satisfactory, why not 450 or 500 lines? Or even the aforementioned 800x600 that would give very clear 20x30 pixel characters? Or was a DEC style compromise reached with a 40x24 screen of 16x20 pixel characters? (in which case, if it's all coming from ROM, the memory requirements are much less of a problem even whilst maintaining 16 colour attributes, the actual physical resolution can be pretty much arbitrary so long as your hardware can handle the frequency demands, and you can fall back to a lower resolution for actual bitmap graphics).

I dunno. Maybe that is genuinely what they did, for some unstated reason. I have no proper inside line. But from what we see on that card, and what I know of modes supported by both Super EGAs and JDM computers of the time (640x400 was a standard PC98/etc mode for a good way into the 256-colour era of the 90s, even, and only added 480-line for Windows support pretty much, so presumably NEC could make kanji work just fine at that rez), plus the likely demands being met, it seems off. And even if it's stated in the sources, I've seen enough of those be hugely unreliable and pretty much written off the top of the web designer's head based on their rather more general existing computer knowledge and guesswork when seeing the timeline. (Which is no more or less than what I'm doing, but I have a closer interest in this stuff and have only even ended up on this page because I was trying to figure out something about MCGA in specific... which was a hopeless adaptor partly for the reason that it could have had more useful modes, including 640x400 at 4c, if IBM weren't deadset on pushing the 480-line mode as a benefit even if it meant wasting 3/8ths of the VRAM and having to run in monochrome... maybe to one-up the S-EGAs and JEGA even though it was at the expense of losing all the more regularly used in-between colour modes). 89.241.24.144 (talk) 13:16, 10 August 2021 (UTC)[reply]

I'm not a computer engineer, but I have some documents and magazines written about the AX. All of them describe the JEGA has 640x480 screen. The card of the picture has 512 KB of RAM; each with 64K words by 4-bit. You can see the photograph of another view. The RAMs are marked Mitsubishi M5M4464AL-12. Why did the AX choose 640x480? I don't have that answer, but I guess that 640x400 was not good for Japanese word processing because the line spacing was not enough. (And, the Japanese prefer to draw a grid between the columns and rows). eg: the IBM 5550 has a 18x21 pixel character box and the DOS/V has 16x19. No wonder that the developers chose 640x480 (or 640x475) to design the AX as a better word processing machine than the PC-98. Darklanlan (talk) 16:12, 11 August 2021 (UTC)[reply]
The PC-98 originally chose the 640x400 resolution which was the monochrome graphic mode of the PC-88, made it easy to port PC-88's softwares to the PC-98. It also enabled to use the PC-88's CRT without an adapter. The 640x400 resolution didn't have enough line spacing to display 25-line kanjis, so the PC-98 had 20-line text mode. The EGA chip with 640x480 resolution was developed by Chips and Technologies in 1985, according to this article. The resolution was used by IBM's PGC card in 1984. JEGA was developed by cooperation with ASCII Corporation and Chips. Its text mode supported not only Shift JIS kanjis, but also stretching and bolding text. I thought it was designed for word processor softwares, but the AX was developed after PS/2, so the resolution may have been influenced by VGA. Unfortunately, the PC industry was heading towards VGA when JEGA cards with 480-line planar mode were put into AX computers. Darklanlan (talk) 16:24, 15 August 2021 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:AX_architecture&oldid=1197164756"