Game Sound, the 7800, is it really a big deal?

[kool kitty89]

One 2600 game that surprised me with music was BMX Airmaster. The title music rocks! I agree on gyruss, as well. I like the 7800's Ms. Pac-Man sounds better than the 2600 and both were tia.

The tricky thing with TIA (using the native tone/noise generation) is trying to cater attractively to the low pitch resolution (very specific sets of notes possible) to avoid sounding out of tune and only compromising that if it is really worth it, there's a few nice tunes on battleofthebits http://battleofthebits.org/arena/Tag/tia/ and various demos showing it well. (some music cuts it down to almost completely avoid the off-key stuff but often ends up super simple like Double Dragon and Journey Escape... though there's others where the much is really simple anyway like Dig Dug, and within the normal limits of the chip)

[BrianC]

One 2600 game that surprised me with music was BMX Airmaster. The title music rocks! I agree on gyruss, as well. I like the 7800's Ms. Pac-Man sounds better than the 2600 and both were tia.

The tricky thing with TIA (using the native tone/noise generation) is trying to cater attractively to the low pitch resolution (very specific sets of notes possible) to avoid sounding out of tune and only compromising that if it is really worth it, there's a few nice tunes on battleofthebits http://battleofthebits.org/arena/Tag/tia/ and various demos showing it well. (some music cuts it down to almost completely avoid the off-key stuff but often ends up super simple like Double Dragon and Journey Escape... though there's others where the much is really simple anyway like Dig Dug, and within the normal limits of the chip)

 

The funny thing about Dig Dug is how the 7800 one has the same music.

[kool kitty89]

One 2600 game that surprised me with music was BMX Airmaster. The title music rocks! I agree on gyruss, as well. I like the 7800's Ms. Pac-Man sounds better than the 2600 and both were tia.

The tricky thing with TIA (using the native tone/noise generation) is trying to cater attractively to the low pitch resolution (very specific sets of notes possible) to avoid sounding out of tune and only compromising that if it is really worth it, there's a few nice tunes on battleofthebits http://battleofthebits.org/arena/Tag/tia/ and various demos showing it well. (some music cuts it down to almost completely avoid the off-key stuff but often ends up super simple like Double Dragon and Journey Escape... though there's others where the much is really simple anyway like Dig Dug, and within the normal limits of the chip)

 

The funny thing about Dig Dug is how the 7800 one has the same music.

 

Same for Double Dragon.

 

 

 

 

 

 

 

And one more interesting thought on missed potential use for RIOT's timer (does the 7800 or VCS ever use it at all, even via polling?) would be adding a simple 1-bit toggle audio channel (fixed volume) mixed along with TIA for the analog audio that could be CPU controlled OR (critically) driven directly by RIOT's timer like the PCs do with the PC speaker and allowing a 3rd sound channel with much higher pitch resolution than TIA and thus more suitable for music generation, albeit limited to square waves at fixed volume. (having RIOT interrupts in 7800 mode on top of that would have been even better still -allowing CPU interrupt driven PWM on any of the 3 channels, but a very simple option for a rather significant boost in audio capabilities of the 7800 in any case with little to no added board space or R&D)

Edited January 24, 2011 by kool kitty89

[Rik]

Nope, no big deal.But it would be nice to have the sound it was supposed to have.I mean the 7800 was the upgrade from the 5200, and it has 2600 sound .It's such a shame.Some games have the sound upgrade in the cartridge itself, which is cool that they did do that at least.Many of the 7800 titles have pretty good sound still.Donkey Kong theme music is off key, and his walking is a bit hard on the ears, kinda like hearing a beginner playing the trumpet irritating.Most games are pretty good IMO, so i kinda forget about music and sound anyway.It's just too bad how great 7800 sound could have been, didn't happen because of a stupid IMO decision to cut costs, and the 7800 was the victim.They could have cut costs somewhere else i suppose, NOT on the damn systems sound for gods sake!

Edited January 24, 2011 by Rik

[kool kitty89]

Nope, no big deal.But it would be nice to have the sound it was supposed to have.

You mean the per-cart add-ons? I still that that was an odd route to go... a good feature to include, but not one to depend on for most games (and given TIA's limitations, the majority of games could have used better sound -more so if the 7800 had been better supported with many more games). The quickest solution would have been POKEY or some (possibly cheaper) off the shelf sound chip (the SN76489 was only a 16-pin narrow DIP, not as flexible as POKEY, but a very nice addition to TIA nevertheless -and quite possibly cheaper than POKEY in spite of Atari owning the IP). There's arguments of board space use but A. a riser board could have been an interim option until consolidation allowed a full single board, B. there were much smaller (probably cheaper) options than the bulky 40 pin POKEY (aside from the SN there was the AY-3-8913 in a 24 pin wide DIP), and C. if board space was so limited they could have cleaned it up a bit by making some pretty simple changes: removing the RF modulator for one thing (use an external RF box with an AV port including a 5V line to power the modulator -also would mean stock support for composite video out and potential to use an automatic switchbox -without the odd configuration the 5200 used), and replacing the 2 2k SRAM chips with 1 8k chip would have been another board space saving option. (a short-term cost increase -perhaps less so than a riser board- and definitely a very good move in the long term due to the consolidation it would allow and possibly lower component costs even as SRAM got cheaper -2 24 pin 2k chips would eventually be significantly more expensive than a single 28 pin 8k chip, and of course you'd have that nice added 4k of SRAM to use )

 

Hell, I don't see why GCC couldn't have used POKEY to accelerate in-house sound chip production. They could have requested that Warner make Atari Inc provide the schematics for POKEY and perhaps engineering assistance to help isolate the audio block, then take just that audio block (maybe retain the timer/interrupt functions) of the die and shrink it (given the newer processes used in '83/84) and slap it into a compact 16/18-pin narrow DIP and maybe later integrate it with MARIA. Short of totally re-engineering the POKEY die, maybe they could at least have cut the pin count down to 24 pins (any smaller and you'd need a narrow DIP which the die might not fit into) and later do a proper redesign isolating only the desired functions of the chip to a smaller die and package. Using an off the shelf chip (especially the smaller/cheaper SN76489) could have been a fall-back if (for political or technical reasons) using a redesigned POKEY was impractical. (given the stock of POKEYs on hand at the time, the manufacturing cost wouldn't have been so much a factor though and making a bit of space on the board for POKEY in the short term might have been preferable to a 3rd party chip -and then later using a customized redesign of POKEY to consolidate the 7800)

 

Then there's the potential RIOT timer hack I mentioned above for adding 1 high-resolution (but fixed volume) square wave channel via a simple 1-bit toggle mixed into the analog output from TIA. (one step further would be proper RIOT interrupt support for the CPU in 7800 mode to allow hardware timed PWM and PCM) Both of those things should have been cheaper and wouldn't have compromised board space at all . . . actually, it probably would have been a very good idea to add a simple 1-bit tone channel (CPU or timer controlled) as well as full RIOT interrupt utility for the CPU even if they also added another sound chip.

 

I mean the 7800 was the upgrade from the 5200, and it has 2600 sound

The 7800 wasn't supposed to be an upgrade over the 5200, it was supposed to be a "fix" to the flaws Atari had made with the 5200 (by GCC's logic at least) and namely in terms of low cost and backwards compatibility. (plus it was on a fairly tight schedule and aimed at a fall 1984 launch)

Atari Inc's 3200 design was actually more in line with the 7800 concept with a fairly low-cost and backwards compatible design (not so much implementation -MARIA is very different from the A8 derived STIA+ANTIC arrangement), but of course they dropped that in '81 to move on to the "quick fix" option with the 5200 using the A8 chipset (sans PIA) directly rather than reworking the custom STIA (TIA compatibility with GTIA like functionality), though it's a bit of a shame they didn't keep pushing the 3200 design concept in that "quick fix" format. (ie use separate TIA and GTIA chips rather than STIA -the later introduction of CGIA would cut board space as well, no added sound over GTIA's 1-bit keyclick toggle -and you could use RIOT's timer to drive a high res square wave through that or CPU/interrupt driven PWM through it or either TIA channel -also 4-bit PCM sample playback via TIA)

 

 

It's such a shame.Some games have the sound upgrade in the cartridge itself, which is cool that they did do that at least.Many of the 7800 titles have pretty good sound still.Donkey Kong theme music is off key, and his walking is a bit hard on the ears, kinda like hearing a beginner playing the trumpet irritating.Most games are pretty good IMO, so i kinda forget about music and sound anyway.It's just too bad how great 7800 sound could have been, didn't happen because of a stupid IMO decision to cut costs, and the 7800 was the victim.They could have cut costs somewhere else i suppose, NOT on the damn systems sound for gods sake!

Yeah, they should have added better onboard sound while keeping the added expansion feature (the Famicom used that very well in Japan, MSX to some extent too, Nitnendo broke that for the NES though).

There definitely should have been other options for design trade-offs and even some (like the simple timer controlled 1-bit toggle channel idea -let alone RIOT interrupts for easier PWM/PCM) where there were virtually no trade-offs at all, and still others where the trade-offs not only would help sound but also help cost reduction in the long run. (like using an 8k SRAM chip -the external RF modulator might never have saved cost but it would have allowed an automatic switch box and composite video out of the box)

 

I wonder if GCC ever considered some of those options. (especially the 1-bit channel thing... could have added that in MARIA like GTIA's keyclick channel and then allowed it to be driven by CPU or RIOT's timer directly -again, fixed volume only but high frequency resolution for very nice musical note capabilities and no "out of tune" sound, plus PWM on top of that if the CPU was used to drive it)

 

 

Also note: "PWM" is pulse-width-modulation, and as used above, it's referring to CPU driven variable width pulse waves. (pulse waves are what the NES and C64 often used -especially the "electric guitar" sound- and the "modulations" done on POKEY as well) PCM is pulse-code-modulation and refers to digital sample playback. (voice, SFXs, drum/music samples, etc -Jinks did it on the 7800)

[GroovyBee]

Hell, I don't see why GCC couldn't have used POKEY to accelerate in-house sound chip production. They could have requested that Warner make Atari Inc provide the schematics for POKEY and perhaps engineering assistance to help isolate the audio block, then take just that audio block (maybe retain the timer/interrupt functions) of the die and shrink it (given the newer processes used in '83/84) and slap it into a compact 16/18-pin narrow DIP and maybe later integrate it with MARIA.

 

You'd need :-

 

D0 - D7

A0 - A3

PHI2

AUDIO OUT

+5

GND

CHIP SELECT

R/W

IRQ

 

Which is more than 18 pins. The timers are pretty integral to the audio section so you can't remove them.

 

Then there's the potential RIOT timer hack I mentioned above for adding 1 high-resolution (but fixed volume) square wave channel via a simple 1-bit toggle mixed into the analog output from TIA. (one step further would be proper RIOT interrupt support for the CPU in 7800 mode to allow hardware timed PWM and PCM) Both of those things should have been cheaper and wouldn't have compromised board space at all . . .

 

... And in reality it wouldn't actually work. The CPU is halted displaying graphics when MARIA is the bus master and therefore Sally can't respond to any interrupt requests. This affects the quality of the audio.

 

RIOT is also an off the shelf part known as the 6532 so no changes there unless you want to fabricate an additional ASIC with the same functionality.

[kool kitty89]

You'd need :-

 

D0 - D7

A0 - A3

PHI2

AUDIO OUT

+5

GND

CHIP SELECT

R/W

IRQ

 

Which is more than 18 pins. The timers are pretty integral to the audio section so you can't remove them.

That would be for a direct re-use of the die and in that case I already mentioned using 24 pins as such. However, with a modified die (and tweaks to internal signals/logic as with 6507 vs 6502), couldn't POKEY be wired to be write only? (and if you were only using it for normal sound generation, IRQ be removed -which is the context of how POKEY is on-cart and thus also applicable to a potential low-cost derivative of POKEY for on-cart use -might have even made sense for ACorp to do that unless they never exhausted the stockpile of normal POKEYs)

 

And as for the timers, I meant more the facility to use them externally to drive interrupts or allowing the CPU to poll them.

 

 

And again, if it got to the point of POKEY being more trouble than it was worth to hack in for the initial release, an off the shelf part should have been an option. (like the AY8913 or more so for the SN76489 -which was probably nominally cheaper, maybe significantly so even after you take into account the overhead compared to the Atari owned POKEY IP -small die, small chip, broadly mass produced) At very least they should have been able to tweak POKEY more moderately or directly re-use the die unmodified in a 24-pin wide DIP like the AY-3-8913 (assuming the die was too large for a narrow DIP), but even then it might not have had a cost advantage over an off the shelf part. (at least without further integration in the long run, or if they took advantage of the 24 pin configuration to retain 4 analog lines and swap those with the TIA POT lines when in 7800 mode -so you got nice hardware POT polling and probably using manual polling or fast scan mode when the POT lines were used for buttons via pull-up resistors)

 

 

Actually, I'm a bit surprised Atari Corp didn't opt to use some cheaper sound chips in 7800 carts; unless they never used up their stockpile of POKEYs, it would have made a lot of sense to use cheaper chips (and possibly invest in cut-down POKEYs be it a 24-pin package or cut-down die and more modifications to an even smaller package -or potentially use of glob top surface mounting directly to the PCB).

And if ACorp never exhausted its POKEY supply (or at least come close to it), it makes less sense that more games didn't use it. (since the only added cost was increased PCB traces)

 

 

... And in reality it wouldn't actually work. The CPU is halted displaying graphics when MARIA is the bus master and therefore Sally can't respond to any interrupt requests. This affects the quality of the audio.

Doesn't the A8 video also have priority over the CPU (halting it for DMA)? (you have to deal with the same thing on the Genesis and various other consoles with DMA contention and no caching regardless of interrupts -the 68k in the Genesis is halted when the VDP asserts DMA in vblank so any continuous processes -like sound management- are absolutely critical to account for by interleaving DMA -ie leaving an open spot for CPU access every few lines; there's a ton of 68k based sound engines on the Genesis that would have to account for that but some wholely Z80 based ones can work around it more since that has work RAM on a separate bus even when the 68k is halted and its bus is locked out by the VDP -and even then you need interleaved DMA -an open slot every few lines in vblank- and buffering into Z80 RAM if you want even PCM playback -other than some small samples sometimes crammed into Z80 RAM like Sonic 1 and 2 do).

I don't mean interleave as in cycle timed interleaved bus sharing like Apple/Acorn/Amiga/ST/etc (which the Genesis does do for Z80/68k bus sharing, but not for the VDP as that asserts full bandwidth -like the Amiga chipset can choose to), I mean interleaving for selective intermittent concurrent accesses (distributing halts/waits more evenly) opposed to parallel interleaved accesses.

 

That is, unless MARIA doesn't allow specific intervals for DMA as such, that and the context that (like the A8 and 2600) it's a single bus design and thus needs DMA to the main bus during active display as well as vblank. (so any interleaved gaps would need to be managed differently)

Any such issues would equally compromise POKEY interrupt stuff too, if they'd used it.

 

Can MARIA generate interrupts for CPU assisted raster effects? (like the A8's DLIs or simpler scanline/hblank interrupts on some other platforms)

 

 

Anyway, that wouldn't bar RIOT's timer from driving a 1-bit toggle square wave.

 

 

RIOT is also an off the shelf part known as the 6532 so no changes there unless you want to fabricate an additional ASIC with the same functionality.

Not sure what that was in response to, but in any case Atari already had licensed (or reverse engineered) RIOT along with the 6507 (possibly 6502) core for the single chip VCS JAN ASIC to be used in the 2600 Jr which eventually got introduced by Atari Corp later on. (and I know the 6532 is an off the shelf and the name was MOS's, not Atari's -just as PIA, CIA, VIA, etc are MOS -or Motorola as with 682x PIA and IIA- acronyms)

 

If you're referring to using the timer to drive a square wave, I didn't mean changing RIOT, but having a tiny bit of added logic elsewhere to allow such a simple toggle (like the Spectrum, Apple II, PC, etc) that could be driven by CPU writes or the RIOT timer (like the PC speaker could be via the PC's interval timers). That logic could have been added to MARIA (like the GTIA keyclick) as a simple alternative to the earlier plans of full sound generation inside MARIA. (use RIOT's interval timer output to toggle the 1-bit "beeper" register that would then generate square waves at the timer rate and output as simple 2 level analog voltage to be mixed with TIA's audio output -like GTIA did with POKEY on the XL and later computers-)

Had they used RIOT instead of PIA in the A8, there could have been potential to allow RIOT to drive the GTIA keyclick output as well. (or use it to drive PWM with interrupts on the CPU)

Edited January 25, 2011 by kool kitty89

[GroovyBee]

That is, unless MARIA doesn't allow specific intervals for DMA as such, that and the context that (like the A8 and 2600) it's a single bus design and thus needs DMA to the main bus during active display as well as vblank. (so any interleaved gaps would need to be managed differently)

 

MARIA is bus master for as long as it has Display List entries in the current zone to process. Once those DLs have been processed the CPU is allowed to run. Thus with sprites moving in and out of zones the DLs are variable in length and thus DMA duration is variable in time.

 

MARIA doesn't do anything during VBLANK. That's when you have 100% of the CPU bandwidth to play with.

 

Can MARIA generate interrupts for CPU assisted raster effects? (like the A8's DLIs or simpler scanline/hblank interrupts on some other platforms)

 

Yep the 7800 has DLIs too.

[Corby]

I know someone who has got out of the 7800 scene for good cause of the sound!! So I think its a big deal.

[GroovyBee]

If you're referring to using the timer to drive a square wave, I didn't mean changing RIOT, but having a tiny bit of added logic elsewhere to allow such a simple toggle (like the Spectrum, Apple II, PC, etc) that could be driven by CPU writes or the RIOT timer (like the PC speaker could be via the PC's interval timers).

 

Its still not going to work. RIOT will count down at the following rates PHI2/1, PHI2/8, PHI2/64 and PHI2/1024 (depending on which register you write the starting count value to). When it reaches zero it does not auto-reload and it begins counting at PHI2/1 on the next count no matter what rate it was set to at the start.

 

The only way RIOT can indicate that the time interval is complete is by the CPU reading a register or RIOT asserting its IRQ line (if enabled). There is no external ripple carry out to connect any buzzer toggle to.

[spiffyone]

You mean the per-cart add-ons? I still that that was an odd route to go... a good feature to include, but not one to depend on for most games (and given TIA's limitations, the majority of games could have used better sound -more so if the 7800 had been better supported with many more games).

 

POKEY should've been on the board itself, not an add on to the cartridges.

 

The 7800 wasn't supposed to be an upgrade over the 5200, it was supposed to be a "fix" to the flaws Atari had made with the 5200 (by GCC's logic at least) and namely in terms of low cost and backwards compatibility. (plus it was on a fairly tight schedule and aimed at a fall 1984 launch)

 

Eh...semantics really. In effect it would've worked as an upgrade of the 5200 regardless, particularly as there was an idea of a 7800 "booster" for the 5200 (I have no idea how that would've worked other than basically having 7800 hardware in an add on shell). In essence, those with a 5200 would upgrade either to an actual 7800 console or to a 7800 add-on for their 5200.

 

Then again, I've argued (and still argue) that regardless of Atari's PR going on about it not making 2600 "obsolete" and being a "high end complimentary model", the 5200 was indeed an "upgrade" to the 2600, and not only that but also it's next gen successor but eventual replacement had things worked out regardless of intention or Atari's initial PR. Same deal with the 7800. Atari wanted users of older consoles to upgrade to successor models made by Atari, and that's pretty much what everyone (at least everyone I know) means when they say "next gen", "successor", or "upgrade" as concerns home consoles.

 

Besides, from what I understand (and see with my own two beady little eyes lol) 7800 is more powerful than 5200. That's an "upgrade" right there.

[Corby]

Is it possible to replace the Tia chip with pokey on a 7800 motherboard?

[Bakasama]

Since TIA does other stuff besides sound. I don't think so.

[GroovyBee]

Is it possible to replace the Tia chip with pokey on a 7800 motherboard?

 

You could do, but then 2600 games won't work and you won't have sound in nearly all the legacy games apart from Commando and Ballblazer.

 

Its much easier to wait for the XM .

[Corby]

Is it possible to replace the Tia chip with pokey on a 7800 motherboard?

 

You could do, but then 2600 games won't work and you won't have sound in nearly all the legacy games apart from Commando and Ballblazer.

 

Its much easier to wait for the XM .

 

Ohhh I am Waiting patiently! But I was just curious... is it possible to piggy back the pokey with the Tia chip?? like I said just curious, was just something I wanted to ask and the right thread came up!

[GroovyBee]

Ohhh I am Waiting patiently! But I was just curious... is it possible to piggy back the pokey with the Tia chip?? like I said just curious, was just something I wanted to ask and the right thread came up!

 

You could probably do an adapter PCB that fits in TIA's socket and supports both but you'd also need to pick up some of the other lines to address decode POKEY properly. The XBOARD sat in MARIA's socket and decoded POKEY at a different address. Only a few homebrew games support the XBOARD tho.

[kool kitty89]

That is, unless MARIA doesn't allow specific intervals for DMA as such, that and the context that (like the A8 and 2600) it's a single bus design and thus needs DMA to the main bus during active display as well as vblank. (so any interleaved gaps would need to be managed differently)

 

MARIA is bus master for as long as it has Display List entries in the current zone to process. Once those DLs have been processed the CPU is allowed to run. Thus with sprites moving in and out of zones the DLs are variable in length and thus DMA duration is variable in time.

 

MARIA doesn't do anything during VBLANK. That's when you have 100% of the CPU bandwidth to play with.

Is MARIA active during hblank as well? (or is there a bit of CPU time available there?)

 

Could you manually organize display lists to leave more CPU time during active display?

 

 

Does the A8 have that problem or is the ANTIC/GTIA DMA managed differently or with more fixed length DLs? (if it IS the same problem, you'd have to work around that to use POKEY timers to drive PWM sounds and PCM samples)

 

Whatever the case, it would seem like the A8/5200 could have done some interesting GTIA channel hacks with the help of an added timer. (like if RIOT was used in place of PIA -that and it might have been possible to allow the CPU to run code from RIOT's scratchpad when video DMA was asserted on the main bus)

 

 

 

 

 

If you're referring to using the timer to drive a square wave, I didn't mean changing RIOT, but having a tiny bit of added logic elsewhere to allow such a simple toggle (like the Spectrum, Apple II, PC, etc) that could be driven by CPU writes or the RIOT timer (like the PC speaker could be via the PC's interval timers).

Its still not going to work. RIOT will count down at the following rates PHI2/1, PHI2/8, PHI2/64 and PHI2/1024 (depending on which register you write the starting count value to). When it reaches zero it does not auto-reload and it begins counting at PHI2/1 on the next count no matter what rate it was set to at the start.

 

The only way RIOT can indicate that the time interval is complete is by the CPU reading a register or RIOT asserting its IRQ line (if enabled). There is no external ripple carry out to connect any buzzer toggle to.

Ah, so it's not possible to do something like the i8253 driving the PC speaker... or is the PC using a CPU interrupt to drive the square wave?

Edited January 26, 2011 by kool kitty89

[GroovyBee]

Is MARIA active during hblank as well? (or is there a bit of CPU time available there?)

 

MARIA is active for as long as it has DL headers to process. If you keep the DL headers to a minimum you'll have more CPU time on that scan line. I have done a tech demo using the equivalent of G2F on the 7800 here.

 

Could you manually organize display lists to leave more CPU time during active display?

 

Its all down to the game. If you have a game with not much going on then you'll have more CPU time during the active part of the display. The more complex the game graphics the less time you have for the CPU to do useful work during the display.

 

However, you don't have to wait for the hardware to indicate VBLANK. Providing you've done the display for the game you can enable a DLI to set a flag during the first zone to contain a NULL DL entry. Subsequent zones with NULL entries will allow maximum CPU time. This can gain you precious VBLANK cycles.

 

that and it might have been possible to allow the CPU to run code from RIOT's scratchpad when video DMA was asserted on the main bus)

 

There is only one data bus on the 7800. If MARIA is using it the CPU can't.

 

Ah, so it's not possible to do something like the i8253 driving the PC speaker... or is the PC using a CPU interrupt to drive the square wave?

 

The 8253 has several operating modes and also an external pin that is set by a count expiring and cleared by actions on registers.

[Corby]

What else other than sound can the Tia/pokey chip do?

[GroovyBee]

What else other than sound can the Tia/pokey chip do?

 

TIA does :-

 

- Video signal generation including player, missile and ball graphics and horizontal position, playfield and colour control.

Collision detection.

- Paddle capacitor discharge.

- 2 channels of audio with tone/noise and volume control.

- Fire button detection.

 

POKEY does :-

 

- 4 channels of audio with frequency tone/noise and volume control. Channels can be combined for finer granularity.

- Random number generation.

- Asychronous serial data with IRQ generation.

- If not used for sound its timers can be used to generate interrupts.

- 8 paddle inputs.

- Keyboard scanning.

[kool kitty89]

that and it might have been possible to allow the CPU to run code from RIOT's scratchpad when video DMA was asserted on the main bus)

 

There is only one data bus on the 7800. If MARIA is using it the CPU can't.

 

Yeah, I was going a bit off topic with that one (also talking about it in the context of the A8 using RIOT instead of PIA), and it would have added to cost by requiring a separate bus (ie for I/O and the scratchpad) which would add to cost and be less attractive than other options. (in the A8's case it could at least have been used to run slightly faster than in DRAM due to lack of refresh -for whatever tiny amount of code you crammed into it; then again, maybe they could have designed CTIA/GTIA to use external CRAM via RIOT's scratchpad for expanded use of indexed colors in the character modes )

[DracIsBack]

Is it possible to replace the Tia chip with pokey on a 7800 motherboard?

 

This would eliminate 2600 compatibility. In 2600 mode, TIA is responsible for graphics and sound. In 7800 mode, MARIA takes over graphics.

[DracIsBack]

POKEY should've been on the board itself, not an add on to the cartridges.

 

Should have, yeah. At the time, though, they were starting to get into cost cutting and the 7800's motherboard was getting full with both 2600 and 7800 hardware. With the recycled case (2800) and the motherboard getting cramped, the thing was getting expensive and space was becoming a premium.

 

Still, POKEY does have nice sound and POKEY and TIA working together to generate sound is a neat idea. The only official 7800 game that did that was Commando, where the music was POKEY and the sound effects were TIA. 6 different sound channels in use.

 

Besides, from what I understand (and see with my own two beady little eyes lol) 7800 is more powerful than 5200.

 

Speaking of semantics, you've hit on the one that drives me NUTS! One 6502 running at 1.79 Mhz is not "more powerful" than another 6502 running at 1.79 Mhz!

 

There is a difference in processing power being - say - the Genesis and the 7800. Between the various 6502 based systems (ie. Apple II, Atari XL/XE, Atari 5200, 7800) etc, there isn't!

 

What you're really referring to is advantages in the graphics hardware and/or available memory etc.

[GroovyBee]

Speaking of semantics, you've hit on the one that drives me NUTS! One 6502 running at 1.79 Mhz is not "more powerful" than another 6502 running at 1.79 Mhz!

 

If you take the chip out of the system then I mostly agree (except when it comes to the NES's Ricoh 2A03 which doesn't have BCD instructions so these have to worked around).

 

However when you throw DMA into the mix to get your video its no longer a straight forward side by side comparison of the "CPU grunt" available to perform a given task.

[+wood_jl]

I'm with you for the most part Underball and I don't get what all the negativity is about when it comes to the 7800 sound especially. Considering it uses the same chip as the 2600, I think most games sound awesome. DK doesn't even offend and I know that's a point of contention with a lot.

 

Late to this thread.....

 

Here's why the 7800 sound was so disappointing:

 

1) Played 2600 games in the early days. Never thought anything negative about the sound.

 

2) Played Atari 400/800 games for *years* after that.

 

3) Heard the Commodore 64 sound (impressive!), for *years* after that.

 

4) Heard ST (nothing to brag about) and Amiga (nice!) sound for some time

 

5) Heard 7800 sound for the first time about 1987 to 1988, back to #1

 

Probably similar situation for other people. I understand the reason (TIA) but, that has little to do with the disappointment. If the 7800 had come out instead of the 400/800/5200 I would have never noticed, never thought about it, never been disappointed.