1041 replies to this topic

[Retro Rogue]

The_Laird, on Tue May 11, 2010 4:35 PM, said:

Its interesting how quite a few of those titles were never released such as 720 Degrees, Road Riot 4WD, Vindicators and Escape From The Planet Of The Robot Mosters on the Lynx. All of the 7800 titles and none for the ST but Steel Talons and Road Riot 4WD were released for the Falcon so I wonder if that is a mistake.

SEC filings are not mistakes, those are reports that all companies in the US are required to file and represent the most accurate info.  The listing is for all the licenses as mentioned, which had to be paid for regardless of whether or not they were published in the end.  Notice as well it states license (executed) and marketing dates on these, not release or publishing dates.  A license is a license, and has to be reported on an SEC filing, as do all financial dealings.  Companies here get in trouble if they don't, especially in the case of issuing stock for payment as this was in regards to.

Edited by wgungfu, Tue May 11, 2010 4:44 PM.

[The_Laird]

Fair enough, I just wondered as Atari hadn't self published any games for the ST since the early days and the Falcon is in the same line of computers being backward compatible and everything.

I am well aware of the legal implications etc thankyou. I was merely stating how its interesting that a company pays out for licenses they didn't even execute. But thats Tramiel Atari all over I suppose.

Edited by The_Laird, Tue May 11, 2010 4:54 PM.

[tyranthraxus]

The first thing I would do is go up to Sam Tramiel and tell him to cancel the Jag and reinstate the ST/Falcon line.

If Sam won't come to his senses then I'd tell him that I want to release the Jag but not use the Atari name on it because teenagers in the early 90s only associate Atari with that crappy old 2600 and with geeky computers. Then I'd blow the bank getting Mortal Kombat or whatever the next hottest arcade game was as pack-in exclusive.

[kool kitty89]

tyranthraxus, on Tue May 11, 2010 7:04 PM, said:

The first thing I would do is go up to Sam Tramiel and tell him to cancel the Jag and reinstate the ST/Falcon line.

But the computers had already stopped selling, they'd lost the market, even in Europe, it seems. (some have speculated that Jack would have fought longer/harder for the computers than Sam did though) You'd need to make changes well before '93 (like in the late 80s) if you want to try and change the outcome of Atari's computers. (both Commodore and Atari let their computers stagnate a fair bit and fall behind progression in PC hardware -and any advancements were largely limited to high-end units until too late: STe then 1200/Falcon; not sure why Atari didn't implement a cut-down version of the TT)

The jaguar ended up providing revenue at a critical time for Atari, that's why they were forced to launch early. (not wanting to risk private funds to support the company in the interim)

Atari Corp. had been trying to get a new home game console out since the late 80s. (and ST derivative followed by the Falcon, followed by Flare 2's Jaguar design created in 1990)

Quote

If Sam won't come to his senses then I'd tell him that I want to release the Jag but not use the Atari name on it because teenagers in the early 90s only associate Atari with that crappy old 2600 and with geeky computers. Then I'd blow the bank getting Mortal Kombat or whatever the next hottest arcade game was as pack-in exclusive.

That's most definitely untrue for the European market... (or anyone who still paid attention to newer Arcade games for that matter)

[tyranthraxus]

Can't argue with you about the lack of ST development, the ST and Amiga were both coasting on their initial designs but I can wish can't I!

However I will debate that the Atari name was mud in 1993. Whether Atari was big in Europe is irrelevant because they launched the Jag in the U.S. The promotion of the 2600 for so many years left the impression of them being yesterdays man.

And the much praised Atari Games wasn't anything special by 1993, take a look at the games they produced after 1990. Apart from some racing games they don't have much compared to their mid 80s prime. Who remembers Primal Rage compared to the success of Mortal Kombat or Street Fighter?

Atari just had no answer for the proprietary franchise games of their competition, perhaps if they secured a key pack-in exclusive and didn't carry the baggage of the Atari name then maybe it'd have had a chance.

[Retro Rogue]

tyranthraxus, on Tue May 11, 2010 10:35 PM, said:

However I will debate that the Atari name was mud in 1993.

As an opinion only, I have to agree.  I still remember going to the Summer CES in Chicago in '94 and seeing this huge huge area for Nintendo for the upcoming release of Donkey Kong Country.  Big jungle theme with a mountain in the middle (reached almost to the ceiling of McCormick), people in jungle outfits, a virtual city of sub booths surrounding it (everything from support software for the design of DKC to general Nintendo related stuff).  Just at the edge of this were some girls in jungle outfits behind a counter.  I walk up to it to see what part of DKC it had to do with, and see an Atari sign revealed on the front of the booth as some people walk away.  As I get closer, one of the girls starts talking to me about this new console called the Jaguar.  I looked at her and said "Atari?  You guys are still around?".  Got a pretty icy look from her for that, but it was the truth.  At that point I hadn't heard of the system, and if I hadn't walked close enough I wouldn't have been able to even tell they weren't part of Nintendo's display.  Really sad looking back at it.

Edited by wgungfu, Wed May 12, 2010 12:03 AM.

[kool kitty89]

tyranthraxus, on Tue May 11, 2010 10:35 PM, said:

However I will debate that the Atari name was mud in 1993. Whether Atari was big in Europe is irrelevant because they launched the Jag in the U.S. The promotion of the 2600 for so many years left the impression of them being yesterdays man.

Which is a big reason why I think it was a mistake not pushing harder in Europe... If it wasn't a question of getting more interest from US investors, I'd say it would have been most definitely in their best interests to push for the Jag in Europe/UK first and US second. (Japan not at all unless they could license it -hence the NEC related post a while back)

Quote

Atari just had no answer for the proprietary franchise games of their competition, perhaps if they secured a key pack-in exclusive and didn't carry the baggage of the Atari name then maybe it'd have had a chance.

At best, the Jaguar could have established a niche market in the US, and people who still had fond memories of the brand name (perhaps some 7800 owners, or even Lynx -for 7800 there had been some 3.7+ million US owners in the US, and of course their computer users too).
Yes, they had the negative connotations as well tied to the name, but honestly, you have to temper that with the alternative: having an "old" name vs a completely new, unknown brand.

A big part of that was a lack of a new console on the market since the 80s. The 7800 was moderately popular but pretty much dead by 1990 and while there were plans for several other machines in the late 80s/early 90s, most fell though. An ST derivative might have been OK for '88, though having the 2600 and 7800, plus releasing the XEGS in '87 confused/divided things a bit. The Panther was planned for 1990, but that ended up really not being that practical, at least in the configuration intended for release (only 32 kB of unified RAM among other issues). Then there's the Lynx, but that pretty much speaks for itself: Nintnedo stole their thunder, it had some problems of its own (cost, battery life, bulk, load times -no direct access to ROM, treating carts more as mass storage, loading into RAM -left over from the original cassette design). It was mainly limited marketing combined with Nintendo's obvious strong competition that doomed the Lynx. (it did manage to beat out the Game Gear in a few regions though, the UK iirc -still well behind the GB) Then there was the unfortunate issues involving delays/price changes with the screens for the Lynx II. (which Atari sued over iirc)

For both the Lynx and Game Gear I wonder why cut-down models using simple (unlit) reflective screen weren't used to save cost, size, and lengthen battery life. (plus improve performance for bright lighting conditions), whether as standard, or alongside the "deluxe" models. (or why no one attempted simple built-in side/front lighting at the time for reflective screens -like many wristwatches did, either using incandescent lights usually -white LEDs being expensive, though red LEDs would be OK for a grayscale display like the GB; the GBA SP did that with LEDs years later, of course)

wgungfu, on Tue May 11, 2010 11:14 PM, said:

As an opinion only, I have to agree.  I still remember going to the Summer CES in Chicago in '94 and seeing this huge huge area for Nintendo for the upcoming release of Donkey Kong Country.  Big jungle theme with a mountain in the middle (reached almost to the ceiling of McCormick), people in jungle outfits, a virtual city of sub booths surrounding it (everything from support software for the design of DKC to general Nintendo related stuff).  Just at the edge of this were some girls in jungle outfits behind a counter.  I walk up to it to see what part of DKC it had to do with, and see an Atari sign revealed on the front of the booth as some people walk away.  As I get closer, one of the girls starts talking to me about this new console called the Jaguar.  I looked at her and said "Atari?  You guys are still around?".  Got a pretty icy look from her for that, but it was the truth.  At that point I hadn't heard of the system, and if I hadn't walked close enough I wouldn't have been able to even tell they weren't part of Nintendo's display.  Really sad looking back at it.

Well yeah, but that really says nothing about their name... just their marketing budget. Imagine if, instead of Atari, it had been some totally new/unknown name, could be worse, might be better, but regardless the key was really lack of product awareness and there's not a whole lot more they could have done except use some resources a bit more efficiently and realize that viral marketing or late night infommecials weren't particularly viable. Prime ad space was expensive, obviously, but a sparese amount of ads that are likely to bee seen are probably better than more/longer advertisements that are less likely to get interest, in the same vein, a few ads of relatively high quality, packing a lot into a short amount of time -or space in print- would probably be making the best of things. (the Doom and AvP commercials are some of their better TV ads, though I also like the "get bit by jaguar" slogan better than "do the math")

[kskunk]

kool kitty89, on Wed May 12, 2010 1:38 AM, said:

For both the Lynx and Game Gear I wonder why cut-down models using simple (unlit) reflective screen weren't used to save cost, size, and lengthen battery life. (plus improve performance for bright lighting conditions), whether as standard, or alongside the "deluxe" models. (or why no one attempted simple built-in side/front lighting at the time for reflective screens -like many wristwatches did, either using incandescent lights usually -white LEDs being expensive, though red LEDs would be OK for a grayscale display like the GB; the GBA SP did that with LEDs years later, of course)

It's exactly because they were color displays.

There was no technology that allowed usable unlit color displays at the time.  Reflective color LCDs have less than 1/3rd the contrast ratio of equivalent black and white displays.  LCD contrast ratios were really low until the mid-90s.

Look at the Game Boy Color -- unlit, came out 9 years later than the Lynx, with LCD technology that was 4 generations improved.  Still, it's very difficult for a lot of people to play in ordinary lighting conditions.

To call white LEDs "expensive" in 1989 is an understatement.    They didn't exist.  White LEDs with sufficient brightness weren't on the market until 1995 and THEN they were expensive.

So side lighting was not an option for color.  To use the kind of compact bulbs that could fit, a light pipe was a requirement anyway.  Unless your screen is a quarter inch high (like a watch), the thickness of the bulb makes it much easier to direct that light behind the LCD rather than in front.  Side lighting is not a very good option with bulbs and large screens.

It's hard to tell if Nintendo succeeded despite the black and white screen or because of it.  It definitely gave them a smaller form factor, lower cost, and long battery life.  But I don't think those reasons were as important as Nintendo's great games and great marketing.

I feel like it's just one of those times where killer games and marketing allowed Nintendo to win despite it's technology handicap.

At least the Lynx was taken fairly seriously by reviewers and developers, if not the public.  I think this was exactly BECAUSE it had a big technical advantage over the Game Boy.  If Atari only had a crippled black and white Game Boy-like system, why bother targeting it (as a developer) or buying it (as a user)?

- KS

Edited by kskunk, Wed May 12, 2010 10:55 AM.

[tyranthraxus]

kskunk, on Wed May 12, 2010 10:38 AM, said:

It's hard to tell if Nintendo succeeded despite the black and white screen or because of it.  It definitely gave them a smaller form factor, lower cost, and long battery life.  But I don't think those reasons were as important as Nintendo's great games and great marketing.

I feel like it's just one of those times where killer games and marketing allowed Nintendo to win despite it's technology handicap.

Well as someone who got a Gameboy in 1989, I'm sure my parents liked the price and I liked that it had Super Mario Land and came with Tetris. It being B&W never really factored as a drawback as I had played many of the primitive Game & Watches which made the GB appear revolutionary. The Lynx and Game Gear looked cool but Nintendo kept rolling out great games so I was never jealous of the other systems.

[kool kitty89]

kskunk, on Wed May 12, 2010 10:38 AM, said:

It's exactly because they were color displays.

There was no technology that allowed usable unlit color displays at the time.  Reflective color LCDs have less than 1/3rd the contrast ratio of equivalent black and white displays.  LCD contrast ratios were really low until the mid-90s.

Thank you! I've brought this up several times on both AtariAge and Sega-16, and no one provided that answer before. So backlit LCD displays had higher contrast ratios than reflective ones?

That still doesn't entirely kill the idea of a later redesign using a reflective screen. (though if they were still too expensive by '94/95 it didn't too much matter as Atari Corp didn't last past that and Sega decided to drop the Game Gear in late '95 -along with pretty much everything but the Saturn -in terms of first party software support)

Of course, there's also the extremely expensive active color LCD display NEC used for the Turbo Express (PC Engine GT), but that's another issue. (that was the ultimate deluxe portable console -a full fledged, full resolution 4th generation home console put into a handheld with a very high-end screen)

Quote

Look at the Game Boy Color -- unlit, came out 9 years later than the Lynx, with LCD technology that was 4 generations improved.  Still, it's very difficult for a lot of people to play in ordinary lighting conditions.

I actually never had trouble with that except in rather dim conditions (still playable even then), the GBA screen wasn't any better as far as I can see, and both were rather similar to the GBP (and better than the original GB in terms of clarity and motion blur -though not battery life)

However, backlit screens suffer the opposite problems they excel in dim conditions, but become very hard to see in bright conditions, especially outside (ie at school, after school -waiting for the ride home- or in a car on a bright day -etc). I've honestly had more problems with that than with reflective screens. (the main problem for reflective color screens is that the colors are off with the gray/silver reflective layer -if you could use a near white reflective layer, it might work better)
That's true for modern handheld game systems (or phones, etc) too, rather like older LED numeric displays (watches, 2-way radios, etc) vs reflective LCD numeric displays. (and now going back again with backlit LCD displays on cell phones -I have to cup my hand over the display on my cell phone to see it outside)

Quote

It's hard to tell if Nintendo succeeded despite the black and white screen or because of it.  It definitely gave them a smaller form factor, lower cost, and long battery life.  But I don't think those reasons were as important as Nintendo's great games and great marketing.

I feel like it's just one of those times where killer games and marketing allowed Nintendo to win despite it's technology handicap.

yeah, at least compared to Atari for the games/marketing, Sega had the marketing and (for the most part) the games, though they were 2 years late to the game and with similar disadvantages to the lynx. (more expensive than the Lynx by that point, I believe) Still, Sega was probably the biggest competitor (worldwide) to nintendo's handheld lines prior to the PSP. (of course, Sega dropped it in '95/96 -not formally discontinued, but halting software support and advertising, later handing off distribution to Majesco)

Sega had the added advantage of an existing library of Master system games that were simple to port to the GG due to the nearly identical architecture (just the screen viewing area and color palette changed -the former meaning that some ports felt a bit cramped compared to the SMS versions), and doing so wouldn't have really menat recycling games to the Japanese or North American markets for most people, as the Master System wasn't very popular ther. Not only that, but they could continue cross-platform SMS/GG games for the European and South American markets where the SMS was still popular. (apparently the GG didn't do very well in Europe, even in Sega friendly countries like the UK -where the Lynx seems to have been more successful)
So for that case, it's the cost, bulk, and battery life, plus the 2 year lead for the Game Boy.

Quote

At least the Lynx was taken fairly seriously by reviewers and developers, if not the public.  I think this was exactly BECAUSE it had a big technical advantage over the Game Boy.  If Atari only had a crippled black and white Game Boy-like system, why bother targeting it (as a developer) or buying it (as a user)?

Yeah, well except for cases where brand loyalty applied, perhaps with some remaining Atari/Arcade fans in the US (indeed the Lynx did have an advantage for arcade games, generally speaking) plus brand loyalty in some Atari Friendly regions, again, the UK.
Some of those other issues didn't so much apply to Sega except very near to the GG's launch. (that and they continued to stay only modestly popular in Japan) Sega might have had a chance to compete on a more direct cost/size/battery life level later on, but again they'd cut it by '96. (but that's another topic anyway)



The only other thing was an unnecessarily bulky design apparently due to responses from consumer study groups waning "more" for their money. (like the 5200 all over again, except now they listened to the study groups but still had a similar outcome) Supposedly there's a ton of empty space inside the Lynx, but given that the Lynx II is on;y moderately smaller, and the GG is still a fair bit smaller than that, but perhaps the Lynx II still had a fair amount of empty space.
Then there were extra features, probably not really that necessary like the left/right hand modes (if that added noticeable cost, it probably wasn't worth it -left/right handed people tended to conform to the same gameplay layout on consoles, regardless, so that may have made it a bit of a moot point, though Gravis did that too).

There was the one technical issue left over from the originally intended casette media, with games having to run from RAM only, loading data from ROM. (I wonder if cutting ram in 1/2 and mapping the cartridge slot directly to 32 kB of main address space would have been a good idea:  that would limit work RAM to 16 kB with a double buffered display -but you could then work from ROM as well. (having an onboard bank selection scheme to expand address space would have been preferable, allowing the full 64 kB of DRAM to stay -though removing that would cut a little cost) Perhaps there were other reasons it wasn't redesigned before release. (iirc it had been planned to release earlier, but was delayed)

[kskunk]

kool kitty89, on Wed May 12, 2010 10:47 PM, said:

Thank you! I've brought this up several times on both AtariAge and Sega-16, and no one provided that answer before. So backlit LCD displays had higher contrast ratios than reflective ones?

That is also true (still is true).  But, I was pointing out that color displays have much poorer contrast ratios than black and white ones.  (Still true today.)  Low contrast ratios make reflective screens unviewable.  However, LCD technology improved dramatically through the 90s, so eventually contrast ratios were good enough to allow reflective color screens.

Quote

There was the one technical issue left over from the originally intended casette media, with games having to run from RAM only, loading data from ROM.

Where did you hear this was a problem?  It seems like a nice performance boost.

ROMs are slow.  The DRAMs in the Lynx are much faster than ROMs.  So copying to RAM allows improved performance.

To get back on topic, the Jaguar maps its ROMs into memory.  But games still load into RAM.  Running from ROM is only usable for very slow systems.

- KS

Edited by kskunk, Fri May 14, 2010 12:05 AM.

[kool kitty89]

kskunk, on Fri May 14, 2010 12:04 AM, said:

ROMs are slow.  The DRAMs in the Lynx are much faster than ROMs.  So copying to RAM allows improved performance.

What I meant was that it would have been nice to have the flexibility either way. (sans the 32 kB comment)

Quote

To get back on topic, the Jaguar maps its ROMs into memory.  But games still load into RAM.  Running from ROM is only usable for very slow systems.

Wouldn't the Lynx fit in that category?  (it's significantly slower than the older PC Engine/TG-16 using a similar CPU, plus the SNES if you take into account the 2x bus speeds necessary for the 65816 multiplexing -though I think the necessary access times for the MD's 68k would be lower)

Wouldn't the 68k's accessing in the Jaguar be slow enough to not be hindered working in ROM compared to RAM?

Edited by kool kitty89, Fri May 14, 2010 3:23 PM.

[kskunk]

kool kitty89, on Fri May 14, 2010 3:23 PM, said:

Wouldn't the Lynx fit in that category?  (it's significantly slower than the older PC Engine/TG-16 using a similar CPU, plus the SNES if you take into account the 2x bus speeds necessary for the 65816 multiplexing -though I think the necessary access times for the MD's 68k would be lower)

Wouldn't the 68k's accessing in the Jaguar be slow enough to not be hindered working in ROM compared to RAM?

Low-cost ROMs run at between 350 and 500ns.  For example, the Lynx ROMs are rated at 440ns.

Lynx DRAM is between two and four times that speed:  250ns in random access mode and 125ns in page mode.  The Lynx uses page mode frequently, for handling graphics and even for sequential 6502 accesses such as opcode fetches.

On the Jaguar, the ROM speed is 375ns.  This is slower than the memory access time of the 68000 which is 225ns.

For comparison:  The Genesis ROM speed is 525ns.  The somewhat newer SNES's ROM access time is 375ns.

- KS

[kool kitty89]

kskunk, on Fri May 14, 2010 8:51 PM, said:

Low-cost ROMs run at between 350 and 500ns.  For example, the Lynx ROMs are rated at 440ns.

Lynx DRAM is between two and four times that speed:  250ns in random access mode and 125ns in page mode.  The Lynx uses page mode frequently, for handling graphics and even for sequential 6502 accesses such as opcode fetches.

Why would the 6502 ever need faster than 250 ns accesses since it runs at 4 MHz?

Quote

On the Jaguar, the ROM speed is 375ns.  This is slower than the memory access time of the 68000 which is 225ns.

225 ns would be 3 68000 cycles, right? The 68k needs memory to respond in 3 cycles? (I know it takes 4 cycles for a complete access, I also recall it being mentioned that it needed memory to respond in 2 cycles to avoid wait states, but that could be off)

Quote

For comparison:  The Genesis ROM speed is 525ns.  The somewhat newer SNES's ROM access time is 375ns.

Well, that matches almost perfectly for 4 cycles on the MD 68000 (particularly the 7.6 MHz of PAL MDs, a little longer than NTSC 7.67 MHz), but again, if memory needs to respond in fewer than 4 cycles, that would mean wait states. Would that apply to only some, or all MD cartridges? (I know the speed of SNES ROMs varies -even eliminating slowdown for some re-releases iirc)

For the SNES, due to the 65816 bus multiplexing, doesn't memory have to be clocked 2x as fast as the CPU? (every memory cycle being 1/2 a CPU cycle rather than 1 for the 650x) So to have 375 ns access times, it'd need memory that could respond at 2x that speed (187.5 ns), wouldn't it?  Plus, while early games used 2.68 Mhz CPU accessing like for ROM (which is also what DRAM uses), I believe there are notable examples of later games using 3.58 MHz access speeds for ROM.

Or is this: incorrect:

malducci, on Thu Apr 1, 2010 1:10 PM, said:

It goes for anything on the bus. The 65816 is missing the upper address lines, so externally you only have 64k of address lines. To overcome this, the multiplexed the upper address lines on the data bus. So what would normally be 1 complete memory cycle, is now 1/2 + 1/2. 1/2 the memory cycle you take the data bus and store it in a simple register latch (external or on the die/package), then on the second 1/2 cycle you use the 16 address lines of the normal 6502 pins and the 8 from the latch to form a 24bit address line, you then read in the contents from the data bus. But the problem there in lies, you need memory to respond at 1/2 memory cycle or twice the speed of a normal memory cycle.

The 65816 on the snes is custom. It's not stock. It has another memory layout ontop of the existing memory layout/system. It facilitates somethings, and hinders performance in other situations. You use long addressing more often. Anyway, if the cart itself had ram mapped to second half of the external address range - you could turn on the CPU to 3.58mhz mode and access that ram at full speed. But that still doesn't fix the whole DP (re-termed from ZP) address range. It's still going to get a penalty since it's in original ram. DP are the 65x address registers.

But anyway, a 3.58mhz 65816 requires a little faster than 139ns memory. A 65xx at 3.58mhz only requires a little faster than 279ns memory.

Also, I recall it being mentioned that ROMs used for Hucards on the PC Engine and TG-16 were exceptionally fast for the time, I'd gotten the impression of it being roughly as fast as the 8 kB of SRAM used by the CPU. (which I assume is accessed at around 139 ns)
That could have simply been relatively speaking though, compared to the slower ROM of contemporaries. (particularly with the PC Engine's 1987 release -though it was released in 1989 in the US, like the Genesis)



Oh, and one thing I just realized I've been assuming throughout this discussion: when the term "OPL" is used, that's referring to the Object Processor, right? (what does the "L" stand for?)

Edited by kool kitty89, Sat May 15, 2010 1:43 AM.

[kskunk]

kool kitty89, on Sat May 15, 2010 1:23 AM, said:

Why would the 6502 ever need faster than 250 ns accesses since it runs at 4 MHz?

To share the bus more fairly with other functions.  It can't consume more than 4 million memory accesses per cycle, but you can space pairs of accesses more closely together to make room for the LCD controller.  This is what the Lynx tries to do by detecting page mode opportunities.

In any case, the main point is that, compared to low-cost ROMs of the late 80s, the Lynx 6502 is 2x as fast and the graphics system is 4x as fast -- definitely worth moving away from the ROM bottleneck!

Obviously you can require high-cost high-speed ROMs but that's a great way to kiss your 3rd parties goodbye...  Fast DRAM is a much better idea.

Quote

225 ns would be 3 68000 cycles, right? The 68k needs memory to respond in 3 cycles?

Yes, correct.  At top speed, the 68K generates one access per 4 cycles (300ns = cycle time).  But the memory must respond inside 3 68K cycles (225ns = access time).

I'll try to distinguish between access time and cycle time more clearly.  When you're talking about DRAM, both numbers matter.  With ROM, access time is all that counts.

Quote

For the SNES, due to the 65816 bus multiplexing, doesn't memory have to be clocked 2x as fast as the CPU?

I don't want to contradict malducci on this because I'm not much of a SNES expert.  I've also heard the SNES 65816 has differences from the stock chip.

It's true that bus multiplexing in the stock 65C816 affects memory access time, but not by 2x.  According to the datasheet, the worst case for a 4MHz rated chip is -75ns address-to-data read.  In other words, worst case ROM access time at 2.68MHz is 375ns-75ns or 300ns.  This assumes you wire up a transparent latch on the bank bits as described in the datasheet.

Write access is more complicated, so RAMs need to be faster still, but ROMs aren't so bad.

Maybe the SNES just required expensive/fast ROMs... that would have been stupid, but Nintendo had their 3rd parties over a barrel so maybe they could get away with it.

Quote

I believe there are notable examples of later games using 3.58 MHz access speeds for ROM.

Moore's law marches on... but we're probably pretty far from 1989 by now, right?  Another way to think about this is that larger, more expensive, ROMs are probably going to be faster too, since they have to be made on newer silicon processes.  And SNES ROMs got pretty big compared to Lynx or GBA ROMs.

Quote

Oh, and one thing I just realized I've been assuming throughout this discussion: when the term "OPL" is used, that's referring to the Object Processor, right? (what does the "L" stand for?)

It's also known as the Object List Processor, which makes sense.  Maybe OPL is a typo on OLP?

- KS

Edited by kskunk, Sat May 15, 2010 1:36 PM.

[kool kitty89]

kskunk, on Sat May 15, 2010 12:59 PM, said:

In any case, the main point is that, compared to low-cost ROMs of the late 80s, the Lynx 6502 is 2x as fast and the graphics system is 4x as fast -- definitely worth moving away from the ROM bottleneck!

That still wouldn't be a reason not to even allow ROM to be accessed directly, other than the cost of onboard bank switching circuitry.

Quote

It's true that bus multiplexing in the stock 65C816 affects memory access time, but not by 2x.  According to the datasheet, the worst case for a 4MHz rated chip is -75ns address-to-data read.  In other words, worst case ROM access time at 2.68MHz is 375ns-75ns or 300ns.  This assumes you wire up a transparent latch on the bank bits as described in the datasheet.

Hmm interesting. The discussion involved the 65816 in general, the SNES got mixed in too, but it was mainly on the 65816.


Maybe the SNES just required expensive/fast ROMs... that would have been stupid, but Nintendo had their 3rd parties over a barrel so maybe they could get away with it.

Quote

Moore's law marches on... but we're probably pretty far from 1989 by now, right?  Another way to think about this is that larger, more expensive, ROMs are probably going to be faster too, since they have to be made on newer silicon processes.  And SNES ROMs got pretty big compared to Lynx or GBA ROMs.

That would have made such a provision useful in the Lynx, at least if it lasted long wnough to pay off.

[kskunk]

kool kitty89, on Sat May 15, 2010 1:53 PM, said:

That still wouldn't be a reason not to even allow ROM to be accessed directly, other than the cost of onboard bank switching circuitry.

Are you basing this off a specific complaint by a developer?  That could help me understand why copying is considered a flaw of the architecture.  To me, it seems like smart engineering.

The Lynx already has its "bank switching" circuitry on-board.  The Lynx provides a full 1MB address bus on its cartridge port.  The only difference is that it uses copying instead of memory-mapped access.

If you want memory-mapped ROM access instead, that requires bank switching, since the 6502 can only access 64KB.

For a developer, memory-mapped bank switching can be more painful than fast copying.  Bank switching requires you to "bank pack" -- packing different sized assets into bank sizes you didn't choose, trying to keep related assets together, worrying about "wasted bytes" at the end of each bank, etc.

With the Lynx (or N64, or DS, etc) you just link all your assets end to end in whatever order you like.  No wasted space, no extra work.  Much nicer!

For that era of games, most of the ROM data was graphics, not code.  All consoles of that generation, including the Lynx and SNES, required graphics data to be first copied into RAM.  Direct access from ROM is not possible because the performance is too low.

You could argue for level data, but... level data needs to be copied into RAM anyway as it scrolls by.  So again, the Lynx is no worse off here...

At this point, the only major thing you AREN'T copying is code.  Games of that era were programmed in assembly and there just wasn't that much code.  The "game engine" for a given stage was rarely bigger than 10KB...  Since code uses only a small part of RAM, and ROM runs the code slower, why not copy it into RAM too?  Your performance will be better!

Obviously I feel that memory-mapped access is unnecessary.  But why not offer it as an option anyway?  Cost.

The Lynx custom chips have their DRAM interface baked directly into the chips.  This is a really nice cost reduction -- not only is there no need for external DRAM multiplexing or more-expensive PSRAM type memory, but you use less PCB space, fewer traces, fewer pins, etc.  The downside is that it's harder to memory map ROM devices.  I think it was a good design choice.

- KS

Edited by kskunk, Sat May 15, 2010 6:04 PM.

[kool kitty89]

kskunk, on Sat May 15, 2010 5:20 PM, said:

The Lynx already has its "bank switching" circuitry on-board.  The Lynx provides a full 1MB address bus on its cartridge port.  The only difference is that it uses copying instead of memory-mapped access.

If you want memory-mapped ROM access instead, that requires bank switching, since the 6502 can only access 64KB.

Wait, so does the Lynx support bank-switching for ROM (to direct CPU access)? (or only copying to RAM)

Quote

For a developer, memory-mapped bank switching can be more painful than fast copying.  Bank switching requires you to "bank pack" -- packing different sized assets into bank sizes you didn't choose, trying to keep related assets together, worrying about "wasted bytes" at the end of each bank, etc.

That would be mitigated by an MMU expanding the address range for the CPU wouldn't it? (like on the Z180, and HuC6280, and the Atari 8-bit MMUs)

Quote

With the Lynx (or N64, or DS, etc) you just link all your assets end to end in whatever order you like.  No wasted space, no extra work.  Much nicer!

But the N64 could use direct accessing too, couldn't it. (I recall reading about some developers taking advantage of that by streaming ROM data in real-time in certain games -in part to address RAM limitations without requiring the expansion PAK iirc)

Quote

For that era of games, most of the ROM data was graphics, not code.  All consoles of that generation, including the Lynx and SNES, required graphics data to be first copied into RAM.  Direct access from ROM is not possible because the performance is too low.

Much of the code is run from ROM on those consoles, isn't it? (TG-16 in particular only has 8 kB of work RAM)  Wait, doesn't the 7800 also pull graphics from ROM much of the time, and isn't Maria rather fast?

Quote

At this point, the only major thing you AREN'T copying is code.  Games of that era were programmed in assembly and there just wasn't that much code.  The "game engine" for a given stage was rarely bigger than 10KB...  Since code uses only a small part of RAM, and ROM runs the code slower, why not copy it into RAM too?  Your performance will be better!

Wouldn't there be exceptions to that though? (some ports or 3D games perhaps)

Quote

Obviously I feel that memory-mapped access is unnecessary.  But why not offer it as an option anyway?  Cost.

The Lynx custom chips have their DRAM interface baked directly into the chips.  This is a really nice cost reduction -- not only is there no need for external DRAM multiplexing or more-expensive PSRAM type memory, but you use less PCB space, fewer traces, fewer pins, etc.  The downside is that it's harder to memory map ROM devices.  I think it was a good design choice.

Would including an MMU on top of that have been too much? (more likely onboard the CPU, but that would have meant a custom, licensed, implementation, like Hudson/NEC and Ricoh/Nintendo did -though havin an all in-house chipset could pay off with volume production too)


Sorry for getting so off topic though...
I suppose it isn't that big of an issue either... (the Lynx didn't have the same issues as the Jaguar other than limited marketing/funding and perhaps management)



However, one other thing on the backlighting: the Lynx II had the option to disable the backlight (apparently intended for pausing as a sort of sleep mode), but in cases with people assuming it was actually meant for a lower-power gameplay mode I've heard comments that (while difficult) visibility wasn't unacceptably poor. (and that's without a reflective backing)
Then again the Lynx II was later, but still only 1991 iirc. (or was it 1990?) That and I still wonder just how unnecessarily bulky the Lynx was (supposedly it has a lot of empty space) given that the Lynx II isn't that much smaller.

One also has to wonder how Mike Katz might have done managing the entertainment division had he stayed, instead of Mead Ames-Klein, Larry Siegel, and Bernie Stolar. (though Klein and Stolar were only there for a few months each -not sure what happened to the Entertainment division after Stolar left in late 1992)

Edited by kool kitty89, Sat May 15, 2010 11:33 PM.

[kskunk]

kool kitty89, on Sat May 15, 2010 11:03 PM, said:

That would be mitigated by an MMU expanding the address range for the CPU wouldn't it? (like on the Z180, and HuC6280, and the Atari 8-bit MMUs)
But the N64 could use direct accessing too, couldn't it.

That type of MMU is just a bank switcher.  To use resources "in place" via "direct memory mapping" requires a lot of programming and linking overhead.  It's a pain, so you only do it when you have to -- for example, on old systems with little RAM.  Once you have enough RAM, direct access to ROM is just a pain in the ass.  Jaguar games are capable of direct access but they all copy anyway.

The N64 doesn't allow direct access.  It's a copy-only system like the Lynx or Nintendo DS.  "Streaming" is not dependent on direct access -- CD-based systems stream too!

I just don't see what you're getting at.  If you can find developer quotes I can respond more directly to those.  Or, give it a try yourself -- write some Lynx code, then write some 130XE code (using Freddy the bank-switching MMU) and see what you think.  It will be easier to discuss the Lynx design once you are familiar with the details.

Quote

Would including an MMU on top of that have been too much?

Not very expensive, but why add something with almost no benefit?

"Why not?" is a bad reason to add a feature, because there are literally thousands of $2 features just like the one.  If you add all thousand "just in case", you will never ship and the target price will be ridiculous.

Good engineers put in the bare minimum absolutely required, because any other strategy can easily double the price or add a year to release.  (As an aside, you can spot bad engineers in those "wonderful" featureful products that never ship at all -- I'm looking at you, Konix.)

- KS

Edited by kskunk, Sun May 16, 2010 1:31 PM.

[kskunk]

I'll try to sum up my contribution to the thread:

For me, the fun questions are, 'Why DID they do XYZ?' That requires a lot of exploration, historical context, teasing out secret memos and documents.

But most of the questions lately have been, 'Why DIDN'T they do XYZ?'

In all cases so far, the answer is one of the following:

1. It wouldn't actually help in general, just that one specific game/made-up scenario.
2. Atari's products were already way over cost (in their view), further raising costs needs stronger arguments.
3. It takes years to get from design to shipping product. The Jaguar design was locked down in 1990, the Lynx in 1986.
4. That technology didn't even exist or was too expensive or too risky, by the "lock down" date of the project.
5. That idea hadn't been thought of yet, or was only known to one or two people in the world, by the "lock down" date.
6. Games like that hadn't been invented by the lock down date, or were still in secret development outside Atari.
7. Atari didn't employ the kind of person (pro game developer, movie industry veteran, etc) to have that idea.
8. Atari lacked the reputation and the money and the sales and the marketing to attract that kind of deal/3rd party/retailer.

There are infinite number of decisions they DIDN'T make... will this thread rehash answers 1-8 on every single one?  

- KS

Edited by kskunk, Sun May 16, 2010 1:46 PM.

[kool kitty89]

I'm not sure where I got the idea about the lynx's copy-only ROM interface being detrimental... but thanks for clearing it up.
Actually, using copy only with a decent amount of RAM opens up some other possibilities, like compressed data on the cart (which the Jag certainly used at times, not sure of the lynx); granted that would be most useful for loading major chunks of the game between levels, not on the fly. (unless you have the available resources to do so -which the Jag might depending on the game; though there might be cases where you'd want to store data compressed in RAM too)

kskunk, on Sun May 16, 2010 1:31 PM, said:

I'll try to sum up my contribution to the thread:

For me, the fun questions are, 'Why DID they do XYZ?' That requires a lot of exploration, historical context, teasing out secret memos and documents.

But most of the questions lately have been, 'Why DIDN'T they do XYZ?'

In all cases so far, the answer is one of the following:

1. It wouldn't actually help in general, just that one specific game/made-up scenario.
2. Atari's products were already way over cost (in their view), further raising costs needs stronger arguments.
3. It takes years to get from design to shipping product. The Jaguar design was locked down in 1990, the Lynx in 1986.
4. That technology didn't even exist or was too expensive or too risky, by the "lock down" date of the project.
5. That idea hadn't been thought of yet, or was only known to one or two people in the world, by the "lock down" date.
6. Games like that hadn't been invented by the lock down date, or were still in secret development outside Atari.
7. Atari didn't employ the kind of person (pro game developer, movie industry veteran, etc) to have that idea.
8. Atari lacked the reputation and the money and the sales and the marketing to attract that kind of deal/3rd party/retailer.

There are infinite number of decisions they DIDN'T make... will this thread rehash answers 1-8 on every single one?  

I see what you mean: the discussion gets far too open ended and off topic...

For the "locked down" design the amount of funding/budget would come into play for sure.  but in the case of the Lynx, the hardware (other than form factor) was solidified a fair bit before Atari was involved iirc. (Atari buying the design from Epyx in mid 1987 I believe)


7. made me think of the Jaguar's development tools in particular and then to id software again. It probably would have been good if they'd contracted some major game developer, not just to work on game software, but development software. (Atari did contract id for games)

Actually, it would have been really good to try and get input from developers industry for input on the hardware and software development system (poll/survey) if at all possible. (depending on the time any hardware changes would be more limited -by 1992 probably only the CPU and RAM configurations as well as deciding how long they could afford to spent addressing the remaining bugs)
I believe Sega did just this when designing the Dreamcast (katana) and the parallel 3DFX based "black belt" design as well. While contracting a game developer to design the software tools (which id later did by themselves to some extent) would likely have cost more than the minimal software tools they did invest in: arranging to survey game developers for input seems like ti could have been relatively worthwhile. (granted, the earlier it was done, the less insight on newer game types, but the later the lesser possibilities for hardware modifications -but still software)

[Gorf]

OPL = Object Processor Logic
so no, It's not a typo.

[kool kitty89]

Gorf, on Sun May 16, 2010 10:47 PM, said:

OPL = Object Processor Logic
so no, It's not a typo.

Heh, that's what I'd actually assumed the "L" stood for in the first place.

[Gorf]

kool kitty89, on Mon May 17, 2010 7:20 PM, said:

Gorf, on Sun May 16, 2010 10:47 PM, said:

OPL = Object Processor Logic
so no, It's not a typo.

Heh, that's what I'd actually assumed the "L" stood for in the first place.

Oh come now! But I'll give you that if...you finally kill this thread!!!!
Sheesh! This makes Beatle mania looks short lived!!!

[kool kitty89]

No, seriously, before I asked, I'd been assuming OPL was referring to the Object Processor, as I mentioned when inquiring about the abbreviation, but Object Processor Logic was the only logical (heh) assumption for the "L" I could think of. I almost mentioned it in that inquiring post too. Believe it or not, but I'm not being facetious.

And yeah, I went pretty off topic with the whole Lynx thing... Though I did find out a few useful things there too. (LCDs, lynx memoy copying not being a hindrance, and the 65816 memory accessing not being as bad as a previous discussion had implied)

Edited by kool kitty89, Tue May 18, 2010 8:12 PM.