Hacker News Reader

Comment by axilmar | Hacker News Reader

I will forever wonder why 8-bit computers of the 80s didn't have more powerful video hardware. It's not that the technology wasn't available, since most arcade machines at the time had a lot more powerful graphics hardware than 8-bit home machines.

(I will also forever wonder why Sega's 16-bit console didn't have hardware sprite scaling and rotation, since that was the main attraction of Sega games at the time...)

I don't think it was only economic reasons, because I really doubt it would be that much more expensive to have a few tiled background layers, a few more sprites on the screen and a few more colors in a decent (i.e. not 16o pixels wide) resolution...

View on HN

Well it almost certainly was cost. Memory was expensive. Address space was limited to 64k. And most importantly, user's displays were color TVs with terrible resolution. Arcade games shipped w/ high quality CRTs. Back in the 80s very few of us had that luxury. My display on my VIC-20 was a 9" B&W TV that my parents had lying around.

Also if you look at something like the Atari 800, it had at least a subset of what you're talking about as far back as 1979. As a result it was more expensive than the competitor machines, and also many software developers didn't make good use of the features that were there.

Our TVs where fine for the resolution of, for example, ZX Spectrum. C64 and Atari and Amstrad CPC graphics looked extremely blockly on our televisions/dedicated monitors (well the Amstrad had such a beast).

I had connected all the above (spectrum, c64, atari) first to a 40 inch b/w tv, then to a 32 inch color tv. Guess what? the spectrum image was visibly blockly, and the c64/atari image was incredibly blocky.

Furthermore, you didn't need another 64k for video ram; there where a lot of palette tricks that could be done. Even if 64k ram were added, the price difference would be small in the end.

So I don't buy it it was the cost; most probably it was the shortsightedness of the designers.

It would have been quite a lot more expensive, I assume? Roughly speaking, with 2MHz RAM, and assuming a setup where the CPU and the display system alternate (so 1,000,000 bytes/second bandwidth for each), you have a budget of 40 bytes for the visible part of the line, and large buffers are expensive, so it's just a question of how you spend it. Take the C64 as an example: you've got 320 px mono (40 bytes/line), 160 px 2bpp (40 bytes/line), or 2 40-byte character modes (40 bytes/line). I assume the fetches for the 8 sprites fit in the 24 bytes that are left, as each is 3 bytes wide.

Later machines had more bandwidth and more complicated memory systems, but the limitations are still the same. You have a fixed number of bytes you can fetch from RAM per second, the TV scans out at a certain rate, and that's going to limit what you can do.

Arcade games could also more readily use ROM, so tile and sprite data could just flow in parallel to the rendering subsystem -- without worrying about sharing with the CPU or refresh cycles.

That's assuming you're reading all the layers from the same ram. You could increase memory bandwidth by using separate chips for each layer.

Memory usually accounted for a majority of the cost in those days, so there's a lot to be said for a unified memory system in home computers. The standard memory architecture for home computers from 8-bits like C64 to 16-bits like Amiga was unified memory with half the memory cycles each going to the CPU and graphics, and PCM playback on later systems could conveniently steal a few graphics memory cycles during hblank. (The C64's VIC-II also stole CPU memory cycles every 8 scanlines and could be forced into stealing a lot more bandwidth with the 'badlines' technique.) ROM-based arcade machines and consoles generally had memory parallelism for their graphics systems. It puts heavy pressure on the chip pin-out and it requires you to physically allocate the RAM/ROM chips onto different buses. I guess theoretically you could put all the memory chips on multiple buses with a configurable arbiter in front of each so you could change the allocation per application, but the PCB for that would be massive and expensive. Instead there was usually a slow, awkward way to access memory from a different bus, e.g. the graphics bus could be accessed by the CPU only with DMA or a pair of memory-mapped addr/data registers with multi-cycle wait states. So, this kind of architecture isn't really suitable for a general-purpose home computer unless you can afford a large allowance of memory for each bus, but it did exist on high-end home computers like the X68000 that were less cost constrained. Also, Amiga did have the option for a kind of RAM expansion called Fast RAM which was only accessible to the CPU, unlike normal Chip RAM which was accessible on the standard chipset bus to the audio and graphics chips in addition to the CPU. Fast RAM didn't speed up the graphics system but it prevented the CPU from stalling when the other chips wanted to steal CPU memory cycles on the chipset bus (they always take precedence since they have hard real-time constraints).

Yes, and the C64's colour RAM is a good example, because accesses to it come for free on top of the 64 bytes/line from the main RAM. But more chips is exactly the sort of thing that would make it more expensive. And I assume it would also have been more complex to build, or something - because if it were cheap and easy, surely people would have done it?

64 bytes per scan line X 200 scan lines = 12,800 bytes.

1,000,000 / 12,800 = 78.125

i.e. frame rate = 78 frames per second could be updated with that speed.

I am sure 78 frames per second are not really desirable, so the resolution could actually be doubled per scan line:

128 bytes per scan line X 200 scan lines = 25,600

1,000,000 / 25,600 = 39 frames per second.

It's my understanding, that the reason why arcade boards were so much more powerful was because they were a lot more complex thus much more expensive to design, source components and produce in large scale.

More colors, background layers and sprites with slow CPUs and graphics chips in the 80s usually meant having several RAM chips to have multiple memory buses and parallel accesses to video memory and more video memory in KB. That's why arcade boards were usually very large and with a lot more components when compared to home computers and home video game consoles.

Sprite scaling was added to Sega's 16-bit console with the Mega CD, it probably wasn't included with the original console so Sega could beat the competition and launch it at a competitive price in 1988 before Nintendo could launch its 16-bit console.

All of this, plus an arcade machine was specialised. BomberMan needs 39 sprites?[1] Build the hardware to do that! Super Racer needs 3D polygonal graphics? Add a hardware scanline rasterizer. To put all these tricks into a home computer, though, would have made it cost many times as much as an arcade cabinet.

[1] Not real numbers.

The hardware budget for an arcade machine could be higher because the economics are to sell a relatively small number of units at relatively high unit cost. On the other hand the hardware development budget was more limited than for consumer home computers or consoles, as that is spread over only the small number of units.

The effect is arcade machines almost never had custom chips, but often had duck-taped-together designs that threw in extra chips to accomplish tasks.

Contrast to a machine like the Commodore 64 or Nintendo which had custom sound and graphic chips but where the overall chip count and unit cost was ruthlessly controlled.

The sprites and backgrounds use the same memory as the main CPU so more sprites and more complicated backgrounds means the CPU gets stalled more.

On the C64 for instance the first line of a character (so one every 8) is when the graphics chip fetches the list of characters on that line and that line is a ‘bad line’ where the CPU has much less cycles. And then the CPU is stalled for 50% of the rest of the time so the graphics chip can fetch the character bitmaps indexed by the data fetched during the bad line.

The first time I've got my hands on a Z80 clone it was like "my precious". For example AY-3-8500 was not unheard of but was unobtainium in every respect.

If you're considering clones in East Europe, I think the real reason was economic. It was useless to have a complex circuit when the devices were not available. For this reason, most early clones of Sinclair Spectrum used only 7400 series circuits around Z80.

Some 8-bit arcade systems also had multiple CPUs; IIRC Galaga had 3 Z80 CPU's working together and I think Qix had 2. Would have loved an dual or triple CPU 8-bit computer in the 80's.

Arcade cabinets are the wrong comparison. Better are home consoles, in particular the NES/Famicom. A fine-scrollable tilemapped background + hardware sprite multiplexing made it trivial to develop games for the Famicom that would be technological marvels on contemporary home computers, yet no western computer manufacturer came up with that design, despite being small additions on top of text-mode graphics and hardware sprites.

Some western micro computers did have some of that though. The C64 did have hardware scrolling, which is why you see a higher calibre of scrolling games on that machine than you would on the Amstrad CPC 464 (which did not have hardware scrolling).

It all comes down to expense though. You could throw the whole kitchen sink into the computer but then how much would it cost to manufacture and how much would you need to sell the thing for?

Costs need to be cut somewhere and micro computers had to additionally ship an interpreter ROM (typically some dialect of BASIC), more expansions ports (for printers, serial modems, memory expansions, additional storage devices, etc) a physical keyboard, and so on.

Plus lets also not forget that while many micros were sold as games machines - they were originally built as hybrid devices for doing work, finances, etc as well as recreation.

That sprite hardware also gave the C64 a reputation as a "game machine" in an era where the big money was in business sales.

There was a lot of "serious business only" attitude in many of the computers those days.

The hardware scrolling in the C64 is pretty trivial, it can only shift the screen up to 7 pixels horizontally and vertically and mask the borders a bit. If you want to do scrolling you have to move the memory contents around which costs a lot of cpu time.

On a console you typically have a circular buffer and the screen is a movable window on it, so you only have to paint one row every time.

Atari 8-bit could fine scroll horizontal and vertical. In 1979.

https://www.atarimagazines.com/compute/issue67/338_1_Atari_F...

Which is the same circular buffer. Great if you can spare the memory, it uses twice as much if you want to scroll in one direction or four times if you want both directions.

On the C64 that’s 8k which is a lot if 16k is all you get (the video chip only has enough address lines to address 16k).

Still, it wasn't like you had to push all the pixels. You only moved 8x8 (or whatever it was) characters. So the C64 was pretty decent, actually. Also, of course you had to shift characters once every 8 pixels.

I was envious as hell on my MSX, which had no pixel shifting at all.

It’s very helpful and the implementation is ingenious but not very complicated.

Vertical scrolling changes the timing of the bad lines and horizontal scrolling the timing of the bitmap data fetched and this causes the image to come out at a different position.

There are a lot of creative tricks you can do with this though, especially with vertical scrolling.

> (I will also forever wonder why Sega's 16-bit console didn't have hardware sprite scaling and rotation, since that was the main attraction of Sega games at the time...)

An interview with one of the hardware designers on the Mega Drive/Genesis VDP suggests they wanted this, but that didn't have enough die space. It's also not clear how comparable to their arcade hardware this would have actually been even if it was present. Scaling and rotation requires non-sequential access to the source data which the VRAM used in the MD/Gen is not very good at. My guess is that the limitations would have been similar to what you saw with the scaling hardware on the Sega/Mega CD (i.e. relatively low frame rate due to the limited bandwidth available for updating VRAM).

I don't buy that at all as well. They could have made the console a bit bigger, and a bit more flexible inside.

I wasn't asking for the full powerdrift arcade experience, but something that can emulate it to a certain degree.

> They could have made the console a bit bigger, and a bit more flexible inside.

They could have gone to a 2 chip solution for the VDP and implemented scaling, but it would have substantially increased costs. Now, Nintendo did notably go with a 2 chip solution for the PPU, but they were using a lower cost CPU, launched 2 years later and sold for a higher price.

There are other places they skimped that hold up much worse than dropping scaling hardware. For instance, having seen photos of the VDP die it seems pretty clear they could have made a nice improvement to the palette hardware if they took more time to optimize the layout (though I imagine that could have taken quite a while with 80's IC design tools).

Don't forget that arcade machines capabilities could be specialized. Consoles, and even more so, home computers, had to be a lot more general-purpose.

Reply on news.ycombinator.com