Sony Haters, come here, you're welcome here!

[kool kitty89]

But you are comparing a system add-on vs a completely NEW system. Plus the Wii while being cheap also offered a great novelty with the Wii Mote. The 32X didn't offer anything that was much of a worthwhile gimmick.

So dramatically improved color/digital audio/3D capabilities aren't worthwhile gimmicks? (albeit the Sega CD skewed things and the 32x wasn't nearly as cost effective as it could be -more hardware acceleration, less CPU grunt, and a better thought-out architecture could have made a big difference)

Then there's the fact that 1. it was an add-on only without "Neptune", 2. the Saturn conflict killed it (where the CD already complicated matters), and 3. the nature of the market was different at the time and the competition met that differently as well:
if the PS3 in 2006 had had the same clean hardware design, cost/performance, and followthrough with software support as the PSX, the Wii wouldn't have had a chance. (might have been forced into a budget/specialized niche)

So sell the Neptune and a gimmicky control - there you have it, the mid-90's version of the Wii. The Wii adds even less to the GameCube than the 32X did to the Genesis... it's perhaps twice the speed, not twice the ram, and a full size DVD instead of the mini-DVD.

Actually it adds more than 2x the RAM, it replaces the 16 MB general purpose/"buffer" DRAM with 64 MB of GDDR3, but the 24 MB 1-T SRAM and 3 MB video RAM stayed the same. (still 88 vs 43 MB)

The CPU and GPU are only 50% faster though, but there's the added 243 MHz ARM926EJ coprocessor as well.

The closest thing to the Wii is the Game Boy color in the sense of a straightforward direct "boosted" successor except the GBC added more. (if the Wii had had full double the clock speed of the GC, and some other tweaks -especially to video RAM and the GPU, it would have been closer to the GBC vs GB comparison) The Wii should also be capable of HD resolutions (not sure if the GC's GPU allowed it, but unless they physically locked the framebuffer size and digital video output, there's no reason it couldn't as well), but HD resolutions and "HD gaming" is not really the same thing as far as the mass market is concerned. (given PC games from 10+ years ago were HD in resolution)

Doing that to the Genesis would be more like the Supergrafx, or a direct update to the MD VDP plus a faster CPU, sound upgrade, and more RAM. (maybe more like the original Mars/Super Mega Drive concept depending on just what that really was)

The Saturn probably would have been a better candidate for a Wii-like design as such though. (cut way back with minimal RAM and CPU resource to be reasonably competitive, but far cheaper -and push heavily for high-level tools as efficient as possible within limits of the hardware)

3DO is a great design (for the time), but was licensed rather than produced by a single company. That model failed.

It was also overly conservative and limited by using 1 micron chips to minimize risk and ensure development stayed on schedule (albeit that would also mean dramatic consolidation later on), they did go for a very low-cost CPU and slow/cheap 80 ns DRAM for main memory, but the VRAM only used as a framebuffer and lack of GPU or CPU cache really limited things. (presumably the huge 1 MB dual-port VRAM framebuffer block was used for intended multimedia purposes -and perhaps high-res 24-bit images, but for what the system needed as a game/VCD platform was really just framebuffers the size of the Saturn and quite possibly using a simple page flipping/bus steering bank arrangement for 2 256 kB banks of 32-bit DRAM -ie 4 64kx16-bit DRAM chips)
Then they needed to avoid the bus contention, and short of a GPU cache (which the Saturn's VDP1 also lacked -or even Jaguar style line buffers, only practical due to the ambitious .5 micron process used), they could have added another block of RAM for textures and/or a CPU with a cache (like an ARM600 -adding an FPU and dropping the fixed-point matrix coprocessor would have been nice since it was quad based) would have addressed that. (1 MB main DRAM, 1 MB texture RAM, and 2x256k framebuffers might have been practical and topping that off with a CPU with a cache -let alone FPU- would have made it into a really competitive system in the long term -and cheaper in several areas; if they skimped on some things like the CPU/FPU upgrade, it very well may have been cheaper than the real 3DO due to the removal of the expensive VRAM and use of a smaller chunk of DRAM buffers in its place -with the added buses mitigating that to some extent; keeping the shared 2 MB main RAM/texture block but adding a faster CPU with cache would have been another trade-off to maintain lower cost)
Hell a 25 MHz 68EC020 could have been a better alternative to the ARM-60 and still much cheaper than any RISC CPUs with caches. (only a small I-cache, but enough to take a major bite out of the bus contention issues)

On top of that, they weren't pushing it in a low-cost form factor until the FZ-10 years into the system's life.



And yes, the market model killed it too, if they were going to push something like that, they needed something much lower-cost in general. (sort of like the above, except just the famrbuffer banks and limit main RAM to 1 MB with the same low-cost CPU and chipset and perhaps swap the CPU to a 68EC020 to reduce contention issues while providing a well-known/common assembly architecture with good C support -or more so if they were more aggressive and pushed for a fully consolidated graphics+I/O+audio+coprocessor ASIC on a smaller process like .8 micron or smaller -more so if they pushed even harder like the Jag and crammed it down to .5 micron and added line buffers for much more efficient bus sharing)
Hell, the Jaguar might have been cheap enough to push into that market model with some tweaks to the design (namely the funds needed to work out the bugs and swap the 68k for an EC020 or maybe an ARM-60 like the 3DO -no cache but high performance and C-friendly- or a lower cost x86 chip with a cache -Jag supported a variety of architectures by design and x86 might have eased PC ports too -especially with the rising PC market)

But that market model was an experiment doomed to fail: razor and blade was tried and true, and a tight partnership with Panasonic might have provided that. (maybe not on Sony's level of spending, but pretty capable)

The Jaguar was a real bitch to develop for, mainly because the tools were immature due to rushing to the market (very much like the 32X).

And the bugs, bus contention (lack of caching -though heavy buffering for some graphics operations- and single bus that the 68k and JERRY accessed very slowly and heavily exacerbated the contention issues), and weak/slow CPU with no cache. (hence why it's difficult to code for in assembly and why it was difficult to make any decent compilers for -and why the first thing many programmers suggest should have been changed is 68k replaced with a more useful CPU with cache or addition of a separate bus for the CPU -actually, a separate slow 16-bit bus would have been great to offload both the CPU and JERRY/DSP onto -DSP could be 32-bits and has 32 data lines but was configured to match the bus width of the CPU -plus connecting only 16 data lines cut PCB cost and for the intended sound synthesis role, bandwidth wasn't a prime factor -using it as a general coprocessor is when it chokes the bus)
Performance could have been boosted significantly with faster RAM, but they wanted to keep cost low. (hence why they also only used 375 ns ROM)

There's a nice summary here:
http://www.atariage.com/forums/topic...1#entry1751151
(and that's not even taking the tools into account -or the difficulty of using the GPU as a CPU)

The N64 was a real bitch, but had excellent support from Nintendo.

Except for RSP microcoding tools. (albeit that was in part due to SGI not developing them, but also from Nintendo not investing them) Plus restricting use of the more PSX-like "turbo 3D" microcode, though that may have been less-used anyway due to the ROM space limits makign unfiltered textures highly undersirable. (and no middleground from the slower "fast 3D" and the "turbo 3D" code as far as official SGI microcodes -ie other than the handful of developers who pushed custom microcode like Factor 5/Lucas Arts and Rare)

The PS2 and PS3 have the same issue - they're easy to program for unless you need to use the vector accelerators, then they are only as easy or hard as the libraries available for the vector accelerators. Sony hasn't been as good at providing libraries as Nintendo, but are much better than Atari or SEGA were.

Hmm, I got the impression that Sony's APIs for the PS2 were cripplingly weak and that 3rd parties had to resort to assembly language programming and/or developing their own APIs/tools for the system.

The Dreamcast's vector accelerator was well supported in libraries, wasn't it?

That, and isn't the PS3 also limited by CPU resource in many cases? (the single PPE being a bottleneck for games pushing "conventional" CPU cores as with PCs and the 360 -let alone games fundamentally having heavy logic/AI related stuff that can't be accelerated by vector units/FPUs)

What are you smoking, man, and why aren't you sharing?

The Wii is NOT a masterpiece in anything, much less the three areas you mention.

It's a masterpiece in priting money!

Or more seriously, a masterpiece of good marketing and finding a gimmick to push into an increasingly neglected segment of the market with and milk for all its worth. (and also realizing when the general public will find the graphical capabilities "good enough" and the price poitn a major mitigating factor)

There's sooooooo much more to it than just a gimmicky controller. With the Wii Nintendo presented the real deal, a fully developed well thought through concept from start to finish. The Wii is and was a masterpiece in design, architecture and innovation, much like the Famicom/NES was in 1983/1986.

It's nowhere near as innovative/well designed as the Game Cube, but they managed to realize that good hardware doesn't matter: gimmicks and marketing do.

Nothing remotely like what made the Famicom in Japan (very good hardware at the right time with no real previous competition and weak contemporary competition -like the VCS in the US), but I will give you that Nintendo's marketing and gimmicks with the NES are hugely tied to its success in the US. (and their followup anti-competitive tactics plus preexisting Japanese support on a monopolistic level)

Sega was incapable of anything like that as they have proven many many times - aside of the early consoles maybe. Not even Dreamcast did not go online until months after it was officially discontinued.

Yes, Sega doesn't seem to have had the business sense and stress on profitability that has kept Nintendo alive. They never learned to milk products for the max profit and not go overboard on the hardware side. (let alone recognizing how to make compromises optimized for different market regions)
What Sega did made sense with the SG-1000 up to MCD to attempt to stay competitive (though their US marketing sucked up until Katz came onboard -Tonka was a lot better than Sega, but far from great).

SoJ and SoA made a combination of mistakes on the Saturn/Mars/32x/MD/CD/GG/Dreamcast in various different areas from 1993 onward (or earlier on SoJ's side with inconsistent support of the MCD that lingered through its entire life).


Not sure what you mean about the DC going online in 2001. (though I agree mistakes were made with the DC as well, especially if they were trying to optimize competition on the market in a sustainable manner given their financial situation -the pack-in modem, various extensive rebates, and price drops were not smart moves in that regard -and they cocked up marketing in Europe too, not sure Japan could have been helped with Sony pushing like they did, maybe they could have milked Saturn a bit longer though)

They didn't seem to push nearly as far as they could have into the PC market either. (tons of great/marketable Sega Published Genesis, CD, 32x, Saturn, DC, and arcade games that should have been released but weren't -and could have provided critical revenue and possibly strengthened the brand name as well -potential PC sales is one reason they probably should have completed Sonic Xtreme in spite of missing the Christmas '96 date)

I'd much rather have hardware designed by people that design and build electronic devices. Just look at the 3DO, Saturn and Jaguar; all 3 were a total mess inside and weren't designed to be programmer friendly either. On the other hand, the original Playstation was inexpensive to manufacture, easier for programmers to get the most bang out of and could be downscaled years later to be even cheaper to manufacture. Later, MS made a very compentent piece of hardware with the original Xbox, but they couldn't keep the manufacturing costs down to compete.

See below for the other consoles, but the PSX wasn't inexpensive compared to anything but the Saturn, it was only priced as it was because of Sony's vertical integration (more so with the ownership of CD-ROM tech patents and already producing them in quantity plus already owning a license for the MIPS R3000A) and deep pocked to allow selling well below cost is what gave them the advantage. (they could have made it lower cost by doing things like cutting out some of the RAM and/or consolidating the buses -putting video, CPU, and audio all on a shared 4 MB block perhaps with a bit more buffering could have been significantly cheaper, but would have weakened peak performance -memory use would have been more flexible though, just as on other single bus designs)

Sony won the market in the 5th gen because they had the money/position/inherent advantages and the right management/marketing to pull it off. If you take any other hardware from the generation and trade that chipset for Sony's, things wouldn't have been much different. (Sony would have pushed stronger marketing, stronger software and tools, modified hardware -due to their funding, position, and ability to absorb cost for a final result being rather similar -with 3DO they'd either be on the market sooner or hold off release until '94 with some tweaks to sigificant modifications to the architecture -tweaks as in a faster CPU with a cache, more significant as in a GPU with a cache; Jaguar was almost completely limited by funding and low cost emphasis -Sony easily could have provided the funds to allow the bugs to be eliminated and pushed it to a configuration with near-PSX capabilities -with advantages and trade-offs- plus a CD drive and STILL have lower production costs than the PSX, and the N64 would have been switched to a dual bus design -probably 2 MB video, 2 MB CPU+audio, maybe use plain PC-66 SDRAM on 32-bit buses rather than the 9-bit RDRAM configuration, or limit the RDRAM to video- and add a CD-ROM drive and still might be cheaper to manufacture than the PSX as well as more powerful in pretty much every respect -and with better tools it could have really kicked ass)


The PSX's friendly architecture and good software support (the latter a critical move also related to buying Psygnosis and having them build the western SDKs) did not follow through for any later Sony consoles:
they horribly botched the PS2 in hardware design (the Dreamcast OTOH took every element that made the PSX hardware great and improved upon it while being truly lower cost as well), the PS3 wasn't as bad on the development side (still nowhere nearly as freindly as it should have been), but they totally f*cked up with the price point and cost/performance let alone efficiently integrating backwards compatibility. (granted, the PS2 hardware design would have complicated that, but a strong focus on efficiently embedding PS2 hardware compatibility and making as much integral use of the old hardware -or enhanced derivatives thereof- should have allowed that without sacrificing price point or performance unacceptably -it may have meant dropping an off the shelf GPU unless it provided close enough compatibility/features to allow it to emulate the PS2 GPU with a bit of added CPU grunt -cramming the EE in there efficiently would be another story, but might have been possible to employ as a useful coprocessor)

There's so many other areas where the PS3 added unnecessary cost and inflated the design (and obviously wasn't aiming at a conservative price point with high cost/performance), that it ended up an utter mess not unlike the Saturn in terms of cost/performance. (could have pushed for a single bus design, maybe DDR-2 give the cost advantages -allowing significantly more RAM and still saving cost plus the savings of a shared bus, the Cell derived CPU should have had fewer SPEs, maybe just 2, and at least 2 PPEs to better match contemproary standards -but still cut down the die size considerably over the full 1 PPE+7/8 SPE Cell used, etc, etc)


NEC has a chance to pull the same thing as the PSX with the PCE/TG-16, but they screwed that up. (didn't invest heavily enough, push aggressively enough, take advantage of their deep pocked and corporate clout, or get the management/marketing right -tying back into investment, for building up a comprehensive western marketing/distribution branch among other things-) They even screwed up with the SGX in Japan and progressive enhancements to the CD to some degree (albeit the 2 issues could b considered 1 in the same -basically SGX should have been an add-on and aimed at coupling with the CD), and obviously the PC-FX.

[gamevet]

I'd love to quote your beatdown of a post Kool-kitty, but I'm going to keep it simple.

The Playstation and PS2 might have started out as expensive pieces of hardware, but with Sony's ability to integrate many parts of the internals to those consoles, they could greatly reduce the costs to manufacture them. MS couldn't do it with the Xbox, Sega could barely make any changes to the Saturn design, and Nintendo pretty much went with the horse they rode in on.

Just look at how many revisions were made on the original Playstation and than the PS2. Even the PS3 has seen at least 4 different hardware designs since it's introduction, and Europe didn't even get the same hardware as North America at launch.

[Chilly Willy]

Too much text for quoting, kitty.

The PS2 and PS3 both suffered from the same exact problem - they were designed to make use of the vector accelerators. The problem with that is most games are written for the PC first, or at least with the PC in mind, so they use the PC philosophy of programming - throw memory and CPU speed at the problem.

Games are written around systems with virtually infinite memory and speed (for the period they were written). That speed usually comes from one source - the CPU. The CPU does EVERYTHING but the rendering.

Sony designed the PS2 and PS3 so that the CPU only manages the game logic and assorted housekeeping. For heavy lifting, you were supposed to write your game to foist the work onto one of the vector units. So that meant that cheap/bad devs were stuck as they had trouble rewriting their games around that architecture. If Sony didn't provide a library, they either had to do it themselves (good dev teams), or they resorted to assembly language optimization for the CPU to keep doing it without help (cheap/bad dev teams). The latter group also whined the most about how much work Sony "forced" them to do.

MS saw that with the PS2 as well as noticing the trend toward multiple cores, so they made their 360 with multiple CPU cores rather than vector units. Probably a wise thing as the continued issues with the PS3 show.

The PS3 is elegant and easy to use... if you design your program around its architecture. If you wrote your program for a PC and are trying to get the best speed on a PS3, you've got some work ahead of you in finding what parts to push onto the SPEs, and how to do that while keeping your program running with the fewest changes.

[kool kitty89]

I'd love to quote your beatdown of a post Kool-kitty, but I'm going to keep it simple.

The Playstation and PS2 might have started out as expensive pieces of hardware, but with Sony's ability to integrate many parts of the internals to those consoles, they could greatly reduce the costs to manufacture them. MS couldn't do it with the Xbox, Sega could barely make any changes to the Saturn design, and Nintendo pretty much went with the horse they rode in on.

Just look at how many revisions were made on the original Playstation and than the PS2. Even the PS3 has seen at least 4 different hardware designs since it's introduction, and Europe didn't even get the same hardware as North America at launch.

Every single system does that, at least if the parts are in-house.

The Saturn did that rather substantially, but it was still fundamentally more expensive.
The 3DO was, and would have continued to be far cheaper. (to manufacture, not to buy -totally different subjects there)
Hell, if not for the market model, a consolidated redesign (timely by 1995 for sure) with the entire chipset crammed into a single ASIC and a much smaller board, low-cost form factor, etc could have made the 3DO the definitive affordable console of the generation.

The Jaguar was extremely consolidated from the start, but could have gotten more so later on.

N64 started out extremely integrated from the start, but the board did get a bit cleaner as time went on.

VCS did it, NES did it, SNES did it, PCE did it, SMS did it, Mega Drive most definitely did it , A8, MCD, C64 did it, Amiga did it, ST, etc, etc.


It's only a matter of A. using custom/licensed components produced in-house and
B. investing in said consolidated redesigns.

MS couldn't do it with off the shelf hardware unless its vendors were willing to do so or if MS had been willing to invest in licensing the chipsets in the first place. (a major cost overhead, but obviously paying off in the long run -assuming they could license the components)


It's much easier to make a complete list of consoles that DIDN'T extensively consolidate their hardware than it is to find ones that did. (ie the exceptions are much easier to pick out than the rule -and many of those exceptions were due to lower production and/or shorter lifespans, etc)


It's not something magical that Sony could do that others couldn't.
They did have vertical integration (which NEC did as well -except Sony also had CD/DVD in-house without the overhead NEC had), but that's a separate issue from manufacturing costs: that's an advantage in raw component costs, like MOS gave for CBM. (actually a bit odd that Atari didn't push for a tight partnership/merger with a smaller chip vendor when they had the funds to do so, let alone other companies)

[kool kitty89]

Sony designed the PS2 and PS3 so that the CPU only manages the game logic and assorted housekeeping. For heavy lifting, you were supposed to write your game to foist the work onto one of the vector units. So that meant that cheap/bad devs were stuck as they had trouble rewriting their games around that architecture. If Sony didn't provide a library, they either had to do it themselves (good dev teams), or they resorted to assembly language optimization for the CPU to keep doing it without help (cheap/bad dev teams). The latter group also whined the most about how much work Sony "forced" them to do.

That's not much different from the Jaguar then, except for the bus contention issues. (namely the 68k eating up too much bus time)

But the idea was similar, offloading everything but the game logic/AI stuff onto the "GPU" (custom RISC coprocessor), but there were no good libraries for the GPU to use out of the box, just an assembler (and a few bugs to work around -only some of which had good workarounds provided in the SDK), and the better developers (like ID) DID push tight assembly as well as custom libraries to push the architecture.
Actually, one of the critical things is that the GPU CAN also offset game logic and minimize the 68k bus hogging (it's not ideal as a CPU, but is better for many tasks than the 68k -so good programmers would need to pare down the 68k to a bare minimum of what it was useful for)
The biggest bug issue was with the MMU (due to a weak power rail on the die iirc) not allowing the GPU to execure code from main memory (the workaround was to page all needed code into the scratchpad on the fly), but homebrew developers (and presumably carmak) found a much better assembly workaround that allowed direct execution of code from main memory. (of course, the scratchpad is faster, so it would be used for specific cases where faster than paging into SRAM or keeping code in the scratchpad permanently)


I've seen comments (and been active in a few discussions) with some who've had experience with the PS2 and the Jaguar, and place the PS2 at a considerably tougher level than the Jag in that sense.


You've got to take things in context though: the Jaguar was a design started in 1990 (core logic docs were laid out then) and completed on the preproduction level in late 1992, being frozen for production by early '93, and also heavily aimed at low cost. (both in the chipset and the configuration -the goal was $150, but they couldn't manage that until 1995, in large part due to the financial situation)
It was being designed in the context of how 3rd and 4th generation computers and consoles were developed for: almost entirely low-level. PC games were just starting to really shift towards C just before the Jag was released and the 3DO was the first console to push for library based development (albeit far too exclusively), and even after the fact, Atari's limited funds would have restricted how much they could invest in producing comprehensive tools. (they'd have to compromise between tools, advertising/marketing, and commissioning games to be developed -and the dev kits likely would have been more expensive if they'd invested more heavily in them)

So it was at least a full generation older than the Saturn and PSX (in terms of technology), the chipset was started and completed years before others except the much more conservative 3DO), they were pushing a .5 micron chip design in 1990 when commercial high-end CPUs wouldn't start using that until 1994 (the PPC603 was among the first). They made the largest ASIC ever created by a 3rd party at that time (also the largest ASIC ever produced by Toshiba up to that time) and possibly the world's first MSOC.

The things that held it back weren't related to design flaws, but almost 100% due to lack of funds and upper management issues (the former exacerbated the latter, but Atari Corp had been declining since Sam Tramiel took over at the end of the 80s). Many of the design flaws were inherently tied to the funding issues, and the design quite possibly could have been completed sooner or without any bugs (possibly some critical improvements -a unified cache would have been a massive boon to performance and ease of programming), and with a bit of losening up on the cost side if they had been well funded. (to the point of being stable and profitable such that they could afford to sell the hardware at cost, consolidate it further as time allowed, and invest in a massive initial production order to push economies of scale -as it was, they were producing jags in the 10s of thousands and by the time production halted in '96, less than 250k units had been completed)

And of course, with the necessary funding, marketing/distribution could have been critically improved as well. (plus, the very fact that the company was more stable and profitable implies that its reputation would be far better and developer interest would be stronger)

It's almost a miracle the Jaguar lasted as long as it did and got so many games given Atari's problems and the pitiful market share.



The PSX's friendly architecture was really just icing on the cake, Sony could have had bus contention issues of the 3DO and development tools on par with the Jaguar (or conflicts with conventional development platforms as much as the PS2), and it still would have dominated is Sony didn't screw up elsewhere (ie marketing, management, software investment, etc), and of course if the competition made all the same mistakes that they did historically.

[gamevet]

Every single system does that, at least if the parts are in-house.

Well, that's the problem, most of these systems used parts manufactured by outside parties.

The Saturn did that rather substantially, but it was still fundamentally more expensive.

The SH2's were manufactured by Hitachi, so Sega had to pay Hitachi whatever price they had decided upon. MS couldn't reduce the price of the original Xbox, since NVidia owned the rights to the custom GPU of that console.

The revisions to the Saturn were very minute. They removed the CD access light and moved a couple of jumpers on the board. Any reduction in chip size would have to rely on Hitachi and the other chip providers for the console. Sony knew this from day one, so they took advantage of it.

The 3DO was, and would have continued to be far cheaper. (to manufacture, not to buy -totally different subjects there)
Hell, if not for the market model, a consolidated redesign (timely by 1995 for sure) with the entire chipset crammed into a single ASIC and a much smaller board, low-cost form factor, etc could have made the 3DO the definitive affordable console of the generation.

They tried reducing costs by removing the motorized CD tray and replacing it with a flip-top design, but the internal hardware was still a problem. Goldstar tried to make internal changes and as a result, a couple of games wouldn't work on that system.

The 3DO was a poorly concieved concept from the beginning. The manufacturers had to rely on console sales to make profits and even Panasonic failed in that area.

The Jaguar was extremely consolidated from the start, but could have gotten more so later on.

Atari console design has always been one step behind the competition. The Jag was the one console that Atari made the efforts to design a CPU and GPU inhouse. The other consoles used existing technologies in their design. The design was still a mess though and you might say it was about as messy as the Saturn. I'd imagine the Jag would have been pretty pricey with a CD drive built in.

N64 started out extremely integrated from the start, but the board did get a bit cleaner as time went on.

Nintendo's consoles have always been about efficient design. They were also designed to be profitable from day one, unlike the other consoles. Even the GC was pretty much the same, minus the digital port that was removed with the Platinum console.

VCS did it, NES did it, SNES did it, PCE did it, SMS did it, Mega Drive most definitely did it , A8, MCD, C64 did it, Amiga did it, ST, etc, etc.

The changes to the VCS were very minimal over the first 5 or 6 years. The CPU didn't change during that time and most of those changes were removing a couple of switches, and changing around the internal arrangements, but they were pretty close (the heavy sixer is considered superior [made in Sunnyvale, CA], since it is said to have better color.) to the original model. It was until around the Atari 2600 jr. that major changes were made to the hardware. The chip designs were still outsourced though.

The changes to the NES and SNES weren't that drastic either.

The changes to the Mega-Drive actually made the hardware worse. We all know about the inferior soundchip in the Mega-Drive 2.

Most of the changes to the C-64 were cosmetic. I don't believe much was changed with the internals. Tramiel investing/owning the companies that made the internal chips of the C-64, was a bigger factor in the price reduction of that computer. The Amiga and ST computers had changes to internal memory and CPU's, but the major changes to those platforms was not because of some magical reduction of the internal chips.

Beyond the external look of the Master System, I'm not that familiar with the internal changes made to that console. Were they significant enough to reduce costs, or were those reduced cost more to do with the price changes of the internal chips used within the console?

It's only a matter of A. using custom/licensed components produced in-house and
B. investing in said consolidated redesigns.

Which was my original point.

Sony has an advantage, because they are an electronics manufacturer and designer. The designs of the EE in the PS2 and Cell in the PS3, were designed with partnerships between Sony and Toshiba. IBM was also involved with the Cell, but Sony has its hands in the cookie jar with that and Blu-Ray. They can choose to reduce the scale, when they see fit, while a company like MS has to work outside of their area of expertise to do the same with the internals of the 360.

MS couldn't do it with off the shelf hardware unless its vendors were willing to do so or if MS had been willing to invest in licensing the chipsets in the first place. (a major cost overhead, but obviously paying off in the long run -assuming they could license the components)

Unfortuneatly, the original Xbox was pretty much off the shelf parts, but MS learned a valuable lesson from that.

MS does have more invested on the inside materials used in the 360, but I don't believe they have a experience, or knowhow of an electronics manufacturer like Sony. The 360 hardware has been a major money pit for MS, but LIVE revenues have more than made up for it. MS defineatly knows the software side of the business, and that's the real advantage for them in the console market.

It's much easier to make a complete list of consoles that DIDN'T extensively consolidate their hardware than it is to find ones that did. (ie the exceptions are much easier to pick out than the rule -and many of those exceptions were due to lower production and/or shorter lifespans, etc)

Any console that has lasted at least 4+ years is a good measuring stick. The PSOne and PS2 have had the most substantial changes to the internal hardware of any console.

It's not something magical that Sony could do that others couldn't.
They did have vertical integration (which NEC did as well -except Sony also had CD/DVD in-house without the overhead NEC had), but that's a separate issue from manufacturing costs: that's an advantage in raw component costs, like MOS gave for CBM. (actually a bit odd that Atari didn't push for a tight partnership/merger with a smaller chip vendor when they had the funds to do so, let alone other companies)

Sony didn't do anything magical, but they used the one advantage they had over all the other console manufacturers, the ability to design, patent and reduce costs through their own manufacturing plants.

Atari could have made those partnerships, but being owned by Time Warner pretty much killed that. When the Tramiels took over, they shifted the company towards the Atari computer line, which pretty much sunk Atari.

[Chilly Willy]

The 360 hardware has been a major money pit for MS, but LIVE revenues have more than made up for it.

No it hasn't. 2009 was the first year MS showed ANY profit on the XBox division at all, and it was slim at best. Last year did slightly better, but one year of slim profit doesn't erase ten years of pouring billions into a division. MS stayed in the console market solely on the profits of the monopoly on the desktop.

MS defineatly knows the software side of the business, and that's the real advantage for them in the console market.

They BOUGHT the software side of the business. I think Sony took a clue with that, doing a few purchases of their own.

[Zoltor]

No it hasn't. 2009 was the first year MS showed ANY profit on the XBox division at all, and it was slim at best. Last year did slightly better, but one year of slim profit doesn't erase ten years of pouring billions into a division. MS stayed in the console market solely on the profits of the monopoly on the desktop.




They BOUGHT the software side of the business. I think Sony took a clue with that, doing a few purchases of their own.

So true.

[gamevet]

They BOUGHT the software side of the business. I think Sony took a clue with that, doing a few purchases of their own.

I was talking more about LIVE and its online features.

[snume]

The Saturn did that rather substantially, but it was still fundamentally more expensive.

The revisions to the Saturn were very minute. They removed the CD access light and moved a couple of jumpers on the board. Any reduction in chip size would have to rely on Hitachi and the other chip providers for the console. Sony knew this from day one, so they took advantage of it.

Not completely true. The major revision from oval button Saturn to round button Saturn also came with a change in CD control boards. The ribbon cable changed from 20 to 21 pins. And there were at least 3 different CD control boards in the round button Saturns as well. Though the actual audio/video hardware didn't change and I'm sure that would have been where the majority of any cost/size savings would have come so this not entirely relevant, just needed to point out there were some substantial changes - I would assume mostly to combat piracy.

[gamevet]

Not completely true. The major revision from oval button Saturn to round button Saturn also came with a change in CD control boards. The ribbon cable changed from 20 to 21 pins. And there were at least 3 different CD control boards in the round button Saturns as well. Though the actual audio/video hardware didn't change and I'm sure that would have been where the majority of any cost/size savings would have come so this not entirely relevant, just needed to point out there were some substantial changes - I would assume mostly to combat piracy.

That's still fairly small changes.

[kool kitty89]

Not completely true. The major revision from oval button Saturn to round button Saturn also came with a change in CD control boards. The ribbon cable changed from 20 to 21 pins. And there were at least 3 different CD control boards in the round button Saturns as well. Though the actual audio/video hardware didn't change and I'm sure that would have been where the majority of any cost/size savings would have come so this not entirely relevant, just needed to point out there were some substantial changes - I would assume mostly to combat piracy.

Actually, they did quite a bit more than that, and there were several other revisions in between. (there's over a dozen hardware revisions of the Saturn)

Aside form the above, they merged VDP1 and VDP2 (that happened towards the end of 1995), they merged the 68k with the SCSP, and I believe some other components were rearranged/consolidated. (they should have been using smaller, surface mounted DRAM and ROM chips later on, not positive though)


The main problem was the design was just more expensive to make, end of story: even with Sony-like integration and licensing it would have been significantly more expensive. (ie even if Sony was producing it and ha dlicensed the rights to the SH1/SH2 cores and most other off the shelf parts -as Sony did, or had prior to the PSX's design due to other reasons, and Sega did do for the MD and some things in the Saturn -the 68k was obviously off the shelf but got merged just as the YM2612, Z80+RAM, and 68000 got merged into the VDP ASIC on the MD -or how the off the shelf RIOT and 6507 got merged with TIA for late model 2600s)

Consolidation and vertical integration are 2 separate issues with different advantages. (Sony had both, Sega only had 1 and with a fundamentally more expensive design at that -ie they had to consolidate it a fair bit just to get it down to the cost of the original model PSX since it was a cheaper design)

There were a number of wasteful design aspects of the Saturn, but 2 of the biggest ones would be the CD-ROM subsystem (512 kB os fast SDRAM and SH-1 -both WAY overkill), and the other being the overall system RAM (much of it fast/expensive SDRAM at that). The soudn system was a bit overkill too (albeit lacking some features like hardware ADPCM, but including tons of other features that were never used -or almost never), but that's something that most were still making mistakes with at the time. (and would continue to with the next generation)
Had they re-used the MCD CD-ROM interface (modified to allow 2x speed and probably double the cache to 32 kB like the PSX/3DO) and have that 68k do double duty for the sound sub-system and CD-ROM, that would have gone a long way in making the system far more efficient. (no idea if they considered doing that early on, but it seems like an obvious option for cost savings as well as accelerated development)

Cutting back to 1 SH2 may have been worth the cost savings as well. (better tools for the DSP would have addressed some of that, and separate add-ons might have been possible later on -especially an FPU coprocessor given the quad rendering of the system)

The final thing would simply be cutting back RAM, and they could have done that and relied more on later expansion when needed (and when RAM got cheaper -especially after 1996), perhaps with a full 32-bit expansion port for the main bus. Probably cut back to 3 MB or less (cutting out the SDRAM from the SH1 and slow DRAM from main memory would have dropped to 3 MB of DRAM/SDRAM plus a bit more PSRAM/SRAM for the CD-ROM buffer and save memory -unless they went with memory cards).
However, the use of SDRAM was also a bit odd, and almost certainly tied to ease of development than any performance advantages. (the SDRAM in the Saturn is all rated for 66 MHz or higher -the lowest grade SDRAM was supplied for on the mass market AFIK- but only clocked at 28.6 MHz, thus easily within the performance range of less expensive EDO DRAM -still significantly more expensive than commodity FPM DRAM, but cheaper than SDRAM and getting cheaper quickly) If Sony used EDO DRAM for main memory in the PSX, that would be one more advantage. (I think the video RAM might be true dual-ported VRAM as in the 3DO, which is expensive, but "VRAM" could just mean video-RAM too)

There's a lot more they could have pushed beyond that as well though. (either to boost performance without much impacting cost, or reduce cost while retaining performance -one big thing would be more buffering/caching on the VDPs, especially VDP1 which seems to have unbuffered texture mapping -or unbuffered everything- and thus could have been much more efficient with 32-bit word buffers -allowing slow DRAM to be used at similar performance- let alone 64-bit and/or line buffers and added caching to allow a single RAM block more like the PSX without separate framebuffers -or you could buffer for higher speed externally on a narrow bus, like running the 16-bit SDRAM at 2x the VDP's speed -ie ~57.4 MHz, again with line buffers -or proper caches- adding potential to use a unified block of RAM without separate framebuffers)

The PSX could have been cheaper too, but Sony opted for performance at higher cost. (you could have it as a single bus design with contention, but still reasonably well performing due to the heavy caching/buffering in the system) The Jaguar is the exact opposite: it would have been far more powerful (and easier to program for) if it hadn't been configured in such a frugal manner.

[kool kitty89]

Well, that's the problem, most of these systems used parts manufactured by outside parties.

No, manufacturing is the vertical integration side, licensing/owning the IP of the chips is what mattered for in-house consolidation of components into ASICs. (as well as cutting out the overhead of fully off the shelf parts -custom IP you take to the chip vendor of choice like Nintendo with Ricoh or Sega with Yamaha or others -Atari Inc and Corp with several different manufacturers including Toshiba and Motorola for the Jaguar)

The SH2's were manufactured by Hitachi, so Sega had to pay Hitachi whatever price they had decided upon. MS couldn't reduce the price of the original Xbox, since NVidia owned the rights to the custom GPU of that console.

That's vertical integration, not consolidation, and more so: Sony licensed the R3000, a lot of overhead for that (but Sony could suck it up . . . though had already licensed it for other projects -being a massive corporation as such).

Consolidation is a separate issue, and Sega DID push that too, the 68k was off the shelf but later merged into custom chips like in the MD. (also doen with the Z80 and YM2612)


And vertical integration is a relatively rare thing and also favors monoply on the market. (very few have the opportunities to have such -Atari Inc missed a chance to do that in their peak -quire a few smaller vendors they could have merged with, possibly Synertek being the most obvious; most others never had a chance)
NEC had it, but failed to exploit it as Sony and Commodore had (and failed to exploit their corporate resources as well -otherwise they could have been the Sony of the 4th gen).
Sony had it even more than NEC since they owned many CD-ROM patents and were one of the biggest manufacturers. (albeit NEC used an in-hose CPU while Sony used one previously licensed from MIPS)

Microsoft probably could have invested in licensing the chipset in the Xbox and re-engineering it for further consolidation, but that didn't happen.

The revisions to the Saturn were very minute. They removed the CD access light and moved a couple of jumpers on the board. Any reduction in chip size would have to rely on Hitachi and the other chip providers for the console. Sony knew this from day one, so they took advantage of it.

Umm, no, there were considerable changes just as with the MD and others. (but it didn't live long enough to see anything as extreme as the VA4 or model 3)
All the custom/licensed chips could be consolidated, only the off the shelf parts wouldn't have been until those were licensed. (Sega had already licensed the 68k for the MD ASIC, so that was one less thing)

The VDPs, custom I/O and memory management ICs, sound system, CD-ROM system, etc, etc were consolidated. (as above, the 2 VDPs were merged into a single new custom ASIC, the SCSP added the 68k internally, etc)


The problem was that Sony had a cheaper system from the start, had vertical integration (and special advantages with CD-ROM), and could dump the price withotu destabilizing the company. (Sega would have needed cheaper hardware to compete on any of those areas)

They tried reducing costs by removing the motorized CD tray and replacing it with a flip-top design, but the internal hardware was still a problem. Goldstar tried to make internal changes and as a result, a couple of games wouldn't work on that system.

Yes, that's a problem with non-standard licensed production: had they unified the partnership with Panasonic and standardized hardware revisions, that wouldn't have been an issue. (the first step would have been consolidating the custom chips, then probably licensing the ARM60 core, using surface mounted RAM chips, etc, etc)

The 3DO was a poorly concieved concept from the beginning. The manufacturers had to rely on console sales to make profits and even Panasonic failed in that area.

Yes, that was the experimental market model that proved unworkable.

Atari console design has always been one step behind the competition. The Jag was the one console that Atari made the efforts to design a CPU and GPU inhouse. The other consoles used existing technologies in their design. The design was still a mess though and you might say it was about as messy as the Saturn. I'd imagine the Jag would have been pretty pricey with a CD drive built in.

No CPU in-house, just the GPU, but the GPU (DSP like chip with CPU-like programmeability) could be hacked to supplement ot replace the 68k.

The problem with the Jaguar was cheap configuration for a low-cost design as well as Atari's weak position at the time. The Hardware is exceptional for the time though and extremely innovative. (more due to Flare than Atari though, same with the Flare 1 and Slipstream in the late 80s)

The changes to the VCS were very minimal over the first 5 or 6 years. The CPU didn't change during that time and most of those changes were removing a couple of switches, and changing around the internal arrangements, but they were pretty close (the heavy sixer is considered superior [made in Sunnyvale, CA], since it is said to have better color.) to the original model. It was until around the Atari 2600 jr. that major changes were made to the hardware. The chip designs were still outsourced though.

No, the motherboard and discrete component counts were cut down quite a bit and by 1983 they had a single chip JAN ASIC merging TIA+RIOT+6507 (they also had CGIA merign ANTIC and GTIA into a 68-pin LCC), but that got delayed from the planned 1984 releae due to Warner's horrible management of the sale/split. (the VCS ASIC was later used on the Jr -later models- while CGIA never entered production -not sure i that was due to loss of the design or overstock of GTIA+ANTIC chips)

The changes to the Mega-Drive actually made the hardware worse. We all know about the inferior soundchip in the Mega-Drive 2.

That's just idotic and ignorant: the consolidation did nothing but improve the system: it was the external analog components that were the problem. (and the sound is the sole issue, which was fixed on late models with better external analog circuitry -mainly the amps)

The NES had some notable changes, but it was co consolidated from the start, it didn't have that far to go. (like the N64)
The SNES got far more consoldiated though. (merging some chips, and especially putting the SCSP into an ASIC on the mainboard rather than the bulky plug-in module)

Beyond the external look of the Master System, I'm not that familiar with the internal changes made to that console. Were they significant enough to reduce costs, or were those reduced cost more to do with the price changes of the internal chips used within the console?

Yes, most definitely, there were increasingly consolidated designs that eventually merged the whole thing onto a single ASIC (not sure if Euro models did so, but the Brazillion ones did -prior to the TecToy sale iirc).


Hell, even the Atari ST saw significant conslidation of the custom chips. (MMU, GLU, SHIFTER, DMA, maybe the BLiTTER later on too -but its lime span prevented sme further changes -the DMA sound was merged with the SHIFTER, MMU+GLU were merged as well, among other things)

Unfortuneatly, the original Xbox was pretty much off the shelf parts, but MS learned a valuable lesson from that.

Yes, so licensing the hardware would have been the solution to that. (liek Sony did for many of their components)

Any console that has lasted at least 4+ years is a good measuring stick. The PSOne and PS2 have had the most substantial changes to the internal hardware of any console.

I think the MD could be considered greater, though it's simpler hardware from the start, so less straightforward. (I don't think the PS2 ever got the entire system merged into a single ASIC though -just the main RAM and vidoe encoder being external)

Sony didn't do anything magical, but they used the one advantage they had over all the other console manufacturers, the ability to design, patent and reduce costs through their own manufacturing plants.

And more so due to the money and influence to buy up and license 3rd party hardware. (same thing NEC did with the PCE hardware -though they hardly exploited their advantages as they could have -probably would have dominated North America and possibly Europe)

Atari could have made those partnerships, but being owned by Time Warner pretty much killed that. When the Tramiels took over, they shifted the company towards the Atari computer line, which pretty much sunk Atari.

Warner made Atari what it was and kept the company from falling apart as the mess it was under Bushnell, they made mistakes, but were largely resplonsible for the much better business decisions that made the company so strong. (Bushnells poor business managment is the reason they needed the help from Warner in the first place)





EDIT:
I forgot to touch on one other subject specifically:
even for in-house projects, there's always the potential (and often used) option for outsourcing, something that even Sony does to a considerable degree: and yet another thing that having tons of cash allows to be more efficient:
ie a massive company can totally cut-out any long term overhead by investing hugely early on and licensing/buying the IP. (in some cases it's much less costly than others though -especially if you consult/commission a relatively small firm -or less than that even- vs licensing an existing design -As MS would have had to do for the Xbox chipset, what Sony and MS would have had to do for the CPUs and GPU designs of their current consoles, etc -even then it depends on favorable contracts as well -as Nintendo almost certainly had with the N64 and GC and Sega with the Dreamcast -some of those cases may have been royalty agreements rather than full licensing along with some fully off the shelf parts in other cases)

That, and off the shelf components will outstrip the advantages of in-house designs eventually if production volumes are high enough and the market is competitive (pushing low prices), unless of course you happen to be one of those mass market companies producing said commodity components. (like RAM and ROM chips, very common CPUs with numerous second sources, etc, etc) OTOH, once production volumes of any one custom product get larges enough, the licensed IP would be preferable. (it's a matter of volumes -and also how favorable the licensing is)

[adamx]

just wanted to say that I love me some sony. # 1 in my book.

[gamevet]

Umm, no, there were considerable changes just as with the MD and others. (but it didn't live long enough to see anything as extreme as the VA4 or model 3)
All the custom/licensed chips could be consolidated, only the off the shelf parts wouldn't have been until those were licensed. (Sega had already licensed the 68k for the MD ASIC, so that was one less thing)

The VDPs, custom I/O and memory management ICs, sound system, CD-ROM system, etc, etc were consolidated. (as above, the 2 VDPs were merged into a single new custom ASIC, the SCSP added the 68k internally, etc)

The 2 VDPs were never merged into a single custom ASIC, not by Sega. Are you referring to systems like the V-Saturn?

You can look up the different hardware configurations of Sega's version of the Saturn; the changes were very minimal.

The problem was that Sony had a cheaper system from the start, had vertical integration (and special advantages with CD-ROM), and could dump the price withotu destabilizing the company. (Sega would have needed cheaper hardware to compete on any of those areas)

Again, this backs my point that an electronics manufacturer like Sony can reduce costs, by having their hand in the design of the chips and parts within their consoles. The CELL backfired on Sony though, because IBM and Toshiba couldn't use it as they had intended.

No, the motherboard and discrete component counts were cut down quite a bit and by 1983 they had a single chip JAN ASIC merging TIA+RIOT+6507 (they also had CGIA merign ANTIC and GTIA into a 68-pin LCC), but that got delayed from the planned 1984 releae due to Warner's horrible management of the sale/split. (the VCS ASIC was later used on the Jr -later models- while CGIA never entered production -not sure i that was due to loss of the design or overstock of GTIA+ANTIC chips)

You're talking over 6 years later, which pretty much falls into the design of the Atari 2600 jr. that I noted in my earlier post.

The NES had some notable changes, but it was co consolidated from the start, it didn't have that far to go. (like the N64)

You're repeating what I'd said earlier. Most of Nintendo's consoles have had very little changes over the coarse of their lifespan.

Yes, so licensing the hardware would have been the solution to that. (liek Sony did for many of their components)

Again, this was my original point. Sony could reduce costs, because they had an invetment in the internal hardware from the design stage.

I think the MD could be considered greater, though it's simpler hardware from the start, so less straightforward. (I don't think the PS2 ever got the entire system merged into a single ASIC though -just the main RAM and vidoe encoder being external)

Sega removed the headphone jack and the internal RF. The rest of the changes were pretty minimal. I don't exactly know what Sony changed between the PS2 and the PStwo, but the expansion bay was removed in the later version. Still, the PStwo is a 3rd of the size of the orinal console.

Warner made Atari what it was and kept the company from falling apart as the mess it was under Bushnell, they made mistakes, but were largely resplonsible for the much better business decisions that made the company so strong. (Bushnells poor business managment is the reason they needed the help from Warner in the first place)

Warner had the money to market the console, as did Magnavox with the Odyssey, but they were not a company that had ideas about what the market was. Magnavox had no idea what they were doing with the Odyssey, and because they priced it like an appliance, the system failed. Warner pretty much marketed the VCS like Coleco did the Cabbage Patch Kids; they knew they had a hot item, but they didn't know where to go once it became dated. It was Bushnell's objections to what they were doing, that eventually led to him being pushed out. He definealtly knew that change was needed, to sustain the home video market.

EDIT:
I forgot to touch on one other subject specifically:
even for in-house projects, there's always the potential (and often used) option for outsourcing, something that even Sony does to a considerable degree: and yet another thing that having tons of cash allows to be more efficient:

MS has a lot more cash that Sony, yet they've shown that they can't create hardware that is profitable. They've shown very little, if any, efficiency on the hardware side. We've seen in the past, that a company like Commodore can reduce expenses by owning the companies that manufacture the parts within the (C64)hardware, but couldn't do the same with the Amiga, since most of the parts were from outside parties.

That, and off the shelf components will outstrip the advantages of in-house designs eventually if production volumes are high enough and the market is competitive (pushing low prices), unless of course you happen to be one of those mass market companies producing said commodity components. (like RAM and ROM chips, very common CPUs with numerous second sources, etc, etc) OTOH, once production volumes of any one custom product get larges enough, the licensed IP would be preferable. (it's a matter of volumes -and also how favorable the licensing is)

You'd think so, but MS couldn't get any reduction in costs from NVidia, for the custom GPU in the original Xbox.