Oh, neat, so it's not as bad as the Atari ST then. ;) (or VGA maybe -not sure if VGA has per-line H-scroll)Quote:
Originally Posted by Chilly Willy
The 32x video layer could thus have been simply used to generate a single 256 color high-detail bitmap scroll layer with line-scroll (but no column scroll) effects. A shame it didn't have a dual layer mode (like Amiga dual playfield -but bitplanes made that simpler on the logic design side), though that would have similar RAM limits as the highcolor mode (more updates or reduced screen size -could use genesis tiles to cover up garbage pixels in the horizontal boarder). OTOH it probably would have been much cheaper/simpler to add a bit of logic for blitting operations (even simpler line by line stuff beyond filling -like copy and affine line rendering). You've mentioned it before, but they seem to have wasted logic on some features like RLE mode when they could have added affine line rendering (for 2D blits as well as 3D texture stuff), or even dropping the 16bpp mode if it facilitated some of that.
And of course, the missed opportunities for various short cuts to accelerated hardware based on existing products or projects. (SVP, DSP from the Saturn, modified CD ASIC, cut-down derivative of the Saturn's VDP1, etc -actually, if there had been an earlier/simpler incarnation of the Saturn VDP1 that was somewhere between the MCD blitter and final VDP1, that might have been a good option to hack in too -the fact that the MCD ASIC was already designed for direct MD interfacing would mean simpler hacks in some areas and similar for the SVP -maybe tweak SVP to the 25 MHz with 12.5 MHz DRAM with the MCD type bus facilitated by the asynchronous MCD derived ASIC interface -maybe even have the SVP and blitter both attached to the same 128k RAM buffer with concurrent access for rendering and data sharing while saving component cost/board space)
If they just wanted to add more color for better 2D stuff (and not worry about 3D at all), then a direct derivative of the MD VDP (with expanded palette and logic to sync and mix analog RGB from the MD side) could have been a good option for an add-on, but that wouldn't have address the interests in pushing 3D. (they could have tacked on the SVP, but that would be limited still and a blitter approach would be far more flexible in general -especially since they had at least 2 existing designs to draw from for a fairly capable 2D/3D blitter)
But, finally there's the very fact that making the 32x more powerful still wouldn't have made it conflict any less with the Saturn... though making it cheaper/simpler/earlier and more limited very well could have been. (be it lack of color enhancement -like a plain SVP cart, or lack of 3D enhancement -like a plain enhanced MD VDP with more color/sprites/layers more in line with the Supergrafx, and more RAM in either case -probably some sort of audio enhancement)
I think I've gone over my thoughts on the SVP itself enough in the SVP thread (and other places). (ie in January of '94 SoA could have pushed for SVP as the basis for an add-on rather than Mars and also cancel the standalone VR cart in favor of the add-on released in its place -128 or 512k+SVP+DMA audio would have been nice and cheap enough to integrate as standard with newer Genesis versions too, like the Neptune but much cheaper/simpler/sooner to the extent of being practical to end production of the baseline MD in favor of that, though it would have meant lower profit margins on hardware late-gen -and not dropping to $50 with the model 3 like Majesco did)
OTOH, if we go by the claims that SoJ initially presented an enhanced MD for the mars project in January of '94 (vague, but seems somewhat supergrafx-like), the colors and 2D performance would have been the primary issues pushed, so a simplified direct add-on version of that (like one of the enhanced MD VDP options above) would have been a direct compromise rather than the tangent 32x design which pushed more in the direction the SVP had gone but with added color and direct bitmap/freamebuffer support.
A shame they left out any expansion via the internal VDP's digital pixel bus, but had SoA opted for the SGX-like approach, they could have used an analog hack for the add-on, and a fully integrated dual VDP ASIC for all-in-one models. (again, potentially displacing the MD in production)
In that sense though, they could have done a bit of both if they'd added such a MD VDP derivative AND SVP, or if pushing more for the blitting side specifically, add a derivative of the CD ASIC on top of the MD based VDP, or that AND the SVP. (or further integrated into 1 or 2 bigger ASICs to save cost/space -given the SVP was new ans already separate, 1 blitter+VDP+interface logic ASIC plus the SVP could make sense, and adding a pixel accumulation mode for 4x8 tiles and 8-bit pixels would have been really useful too -even if they didn't want to use a big block of CRAM for larger 256 entry LUTs, they could have used the added nybble to control 4 bits of RGB for the selected color of a 15 color palette -probably only on a per-layer basis and maybe not for sprites at all if they didn't want to modify the MD tilemap/sprite table formats)
Hell, for all we know, Sato's original Mars/enhanced MD design might have done some of that. (if it was proposed as claimed) Definitely something to ask if Sato ever opens up to an interview on the subject (along with various other things like elements of the Saturn design).
Joe miller should also know at least an overview of details on that proposal (and would be able to confirm if SoJ indeed brought that to the table in Januar of '94), though he did mention he'd only do an interview with Melf if the 32x was left out of the topic. (not sure if that would include questions about SoJ's initiation of the Mars project too or if it's more about other issues on the 32x he's tired of rehashing)
There are also claims that before the dual SH2+framebuffer "hotel napkin" 32x design was decided upon, there were actually several other possible on-paper designs (perhaps some being direct add-on derivatives of Sato's proposal or building onto the Sega CD ASIC hardware), but there's no details on that other than Sam Pettus's comments on "SH1, dual 68000s, and barely better than genesis" hardware. (and his tech stuff is horribly distorted most of the time, unless it's direct unadulterated quotes -at least with most of his commentary on political things you can sort through what's biased and what's direct quotes or paraphrasing OK -the lack of internal citation hurts that though- but the tech stuff is more of a mess unfortunately)
Joe Miller (being SoA/STI's head engineer and manager of the Mars/32x project) definitely should know more details on that too, but again, that may be in the "don't want to talk about 32x" category unfortunately.
Yes, but also remember that Sega had massive revenue for amusement parks and arcades and also didn't seem to focus as heavily at conservative spending and building up cash reserves.Quote:
Originally Posted by Da_Shocker
Of course, the all-out marketing of the Genesis left little room for error in terms of cash reserve (profits, but still often deficit spending in-between from what I understand), but A. Sega's issues with lack of reserves seem to have predated that, and B. Sega squandered their hard earned market with the Genesis by dropping out before they could reap the late-gen reward with high profit margins.
Some of "B" ties in with the "not smart" issue (being too hardware oriented and clashing management), you could also argue that Kalinske should have tempered spending a bit more and still had a strong market (Katz seems to have been a bit more conservative in that regard -and defintiely experienced in how to manage a market if you did end up in a negative/weak position -ie as he managed Atari Corp's entertainment division from '85 through '88 with a very tight budget and surprising success in sprite of that -much higher market share than the vastly better funded Sega, he was also smart enough to keep a fair line of new games going to the 2600 in its fairly strong niche as a budget console)
However, going beyond that there's a good chunk of other issues: Sega's arcade market collapsed in the mid '90s (exacerbated by expensive new arcade hardware Sega was making with the model 1/2/3 and no lower-cost 3D board other than the ST-V which was NOT emphasized as such -the hardware could have been tweaked more for being a good lower-end multipurpose 2D/3D board though, like adding an FPU and perhaps more video RAM, and/or a later revision swapping the SH2s for an SH3 -Naomi changed that big time, but they could/should have pushed earlier -even then the arcades were weakening though, Sega's expensive hardware just made it worse for them).
Then there's the myriad of conflicts/problems/bad decisions on the Saturn and 32x from various areas (hardware design, cost, marketing, launch timing, software support, development tools, management/political issues, etc) and those mounded on top of eachother with mistake on top of mistake rather than deft maangement to soften the blow of said mistakes and push ahead.
Then there's the truely poor decisions to not push hard with the Genesis for a high-profit lower-profile budget market and not to push newer/competitive revisions of the GG. (by '95 at the very least they could have offered a consolidated/cost-cut version of the GG with a deluxe model still using a backlit screen and a more compact Jr model dropping to a reflective color screen -with contrast levels finally getting acceptable for unlit color LCDs- and thus partially correcting both the more moderate flaw of bulky size and fully correcting the absolutely critical flaw of battery life with the unlit model -and lower cost at that- so they'd have something better than the GB but close in price point and battery life -the chipset should have been consolidated into a single CMOS ASIC plus separate DRAM chip for video -given the time, probably a single 32kx8-bit DRAM chip given the general pricing/volumes of densities at the time)
And after that, a true successor to the GG. (not too early though, focus more on cost reducing and consolidating support for the original system before moving on -unless perhaps Nintendo upped the game sooner to respond to sustained competition, otherwise they could safely wait until the late 90s to move on with something more definitively superior to the GBC)
That mistake was even more critical as it meant Sega missed out on a chance for one more strong market position: even if they stayed 2nd to Nintendo, they could have competed very strongly (Sega and Nintendo would have been pretty much the only players on the handheld market until the late 90s -and the later competition probably would have had even less of a chance with Sega in the game, so effective no one but Sega and Nintendo until Sony jumped in over a decade later).
Nintendo had the handheld AND console markets to support them, Sega had their arcade/other markets heavily decline in the mid 90s and screwed over both the handheld and console market positions.
They did make a decent attempt in the PC market, but that probably could have been much stronger too. (many great Sega published Genesis, CD, 32x, Saturn and arcade titles -and later Dreamcast- that didn't make it to PC but could have catered well to the market)
Hell, SoA was planning on several Saturn games with parallel PC releases, among those being Sonic X-Treme. (a shame they canceled it, it was delayed and probably over budget, but nearly complete -and had a chance of making back a good chunk of that at least if not being profitable as a PC+Saturn game -by '97 more so a PC release)
Overlapping products are bad, but diversifying the market is very healthy.
So let's see Sega missed out partially or wholely:
-stronger/continued 16-bit late/budget market
-stronger PC market
-stronger/continued handheld market
On top of the mess they made in the 32-bit market and problems in the arcade.
And if you go further you've got lots of other missed opportunities:
-no low-cost CD duo
-SVP add-on could have been a much cheaper/non conflicting alternative to 32x (and much sooner); whether or not it was seriously considered historically is not clear
-weak support for MCD with lack of various arcade ports (especially scaling), more 1st/2nd party unique games, and many more parallel MD/CD developed games or ports/compilations of MD games (with varying degrees of enhancement and price-point advantages allowing the CD to literally pay for itself) and you'd need broader marketing of the CD to advertise all those advantages with a wealth of games. (FMV was a good gimmick but too emphasized -even compared to other multimedia elements)
-various simpler/cheaper/earlier enhancements on-cart (without add-on) from a little chunk of RAM expansion (like a 128k DRAM and small interface ASIC), various sound enhancements (simple DMA sound chips, on-cart YM2612 -or even smaller ASIC YM2612 with built-in interface logic, etc), and some other possibilities. (like low-cost ASICs with dedicated logic for lossless decompression). One problem to be worked around would be V-DMA still locking-out the 68k bus and that would apply to any RAM/sound/logic on-cart as well, so interleaved DMA segments would need to be managed for smooth audio updates. (for the decompression ASIC, you could possibly have a small internal SRAM buffer to work in for better interleaved use with DMA -and a simple DMA sound chip could have the same or a larger external buffer that could be toggled between the 68k bus and a dedicated on-cart audio bus, but a smaller on-chip buffer would be cheaper/simpler -or if it was a more powerful PCM chip doing hardware sampling/mixing like the ricoh chip and not basic DMA like the Amiga/STe/32x/Soundblaster/etc, you could use a dedicated ROM chip filled with uncompressed samples for the chip to access directly rather than the CPU mixing/buffering samples and managing loop times for a simple DMA circuit)
Hell, in Nintendo's case, the virtual boy even had a positive impact in the short-run like the 32x (possibly more so) as it temporarily boosted PR/interest when Nintendo had delays with the N64 and had already canceled some late gen SNES games that could have also helped fill the gap. (though in that respect, it would have been smarter to push those SNES games -namely Super FX titles Nintendo felt would conflict with the N64 or keep it from making a "fresh start" with Nintendo 3D, or for that matter, Nintendo is the one who could have used a 32x/SVP/SFX add-on adapter to bridge the gap with the delay of the N64 :mrgreen: -something like hacking the SFX2+SA1+RAM into a cart adapter -they also could have probably adapted the virtual boy hardware but that wasn't designed to interface with the SNES already and used a more expensive off the shelf NEC RISC CPU and faster/more expensive DRAM/PSRAM/VRAM)
But yeah, Nintendo managed a short term boost in PR and a useful smokescreen for the N64 delays (and cancellation of the SNES CD and some games like Star Fox 2) and then deftly downplayed the failure of the Virtual Boy after the fact (in spite of it impacting more Nintendo users than the 32x had Sega), which is exactly what Sega should have done with the 32x. (though given the different market position, they would have needed to manage that differently than the Virtual Boy context)
