Worst system design flaws in consoles?

I always thought it wasn't so much the hardware with the Jaguar that held it back so much, it was that the CEO at the time, (Jack Tramiel?), ran it into the ground. They deliberately witheld development tools thinking they would hold their best cards until PSX and Saturn came out. Instead, no one could make a game for the damn system and it died well before PSX or Saturn had a chance to kill it. I'm told that some of the newer games, (there are still people programming for it, although they are mostly doing it as a hobby), rival early PSX and Saturn games.

Another failure, although more marketing than hardware, was Sega's failure to inflate its numbers on the Dreamcast. I truly think that hype from the PS2 was what killed it, because I personally know numerous people who saw the press release of the PS2's 60m tris/sec, and then didn't give a second thought to the Dreamcast until it was down to less than 100 bucks. One of these people had been literally counting down the days to the US dreamcast release for well over a year. If Sega had pushed more for "effective" triangle and fillrate counts, I think the blow Sony dealt them wouldn't have been nearly as bad.

Also, I believe the Sega Nomad, Handheld Turbo Grfx, and Atari Lynx were all backlit, although they sucked batteries like Clinton sucked cigars.

 

Jaguar was fully programmable, even MORE so than PS2.

If you really wanted to, you could probably have the "sound" processor manage graphics, the "graphics" processor manage I/O, the Motorola 68k (generally used for I/O) do sound...

Oh and Jaguar is even more freakishly powerful in 2D than the Saturn, and that's saying a lot...

 

I remember being blown away by Aliens vs Predator on the Jaguar... but even if the platform had taken off, the cartriges would have driven into the ground.

 

BZZT! Wrong! There is a CD-ROM add-in that plugs straight into the cartridge slot, and just like 3D0, Saturn, and PSX, reads at 2x.

Fun fact: Jaguar's CD-ROM was developed before the XA standard appeared, so it uses standard audio CD's with data on them instead - the interesting thing being that to prevent misuse, the header data looks like an error to anything other than a Jaguar FUN!

 

Shortly after my run in with the Jaguar QWTF took over my life, so any and all video game related info never reached my ears!

 

Everyone hypes things up...

Sega hyped things up as well...

http://www.computer.org/micro/articles/dreamcast_2.htm specs the dreamcast T&L at 10Million polygons per second and the drawrate at 7Million polygons per second - much more than achieved in many games...

SCEA's press releases rated PS2 at about 36Million polygons/second which is pretty comparable to the Sega rate - It was the magazines that ran with the 75Million number ( the same ones going with the 4GPixel fillrate for Xbox.... )


Atari release the Jaguar with tools at the start that remained constant throughout its short lifespan - it was just the wrong machine at the wrong time, a year earlier it would have been a brilliant competitor to the SNES/Genesis and 32X - but both Saturn and PSX completely wiped it in both 3D and 2D graphics.... ( The machines strongest point was the gourard shading, but having to essentialy scan convert triangles using the GPU really reduced the performance... One of the best ways to get performance was to transform geometry using the DSP, and rasterise with the GPU - and run game code and sound generation in the VBlank )

 

...and we all know Jaguar developers where in a rush to make all their games for this CD add-on, right? Why does this add-on suddenly become an advantage for the Jaguar? If cartridge-only format can kill an N64, it seems the Jaguar would be just as susceptible.

 

The CD add-on came out literally at the end of Jaguar's life cycle, but games WERE announced for it... though only three or four actually saw the light of day.

I only mentioned it because zurich said Jaguar is cartridge-ONLY however that is FALSE.

Oh and there is one VERY VERY VERY good thing about Jaguar's CD add-in... the truly music-based (i.e. not some stupid fixed patterns but a morphing algorithm) visualisation... music enthusiasts today still buy Jaguars for this reason alone.

 

And it also made your sleek console look like a potty training appliance.

The PC Engine was in the same boat...

I have to admit, Virtual Light Machine was pretty cool...

 

I probably shouldn't say this, there was an option to put in homogenous coordinate rendering (as is available in PCX1/2) for tris that otherwise would go through the front clip plane but were told that the SH4 was fast enough and not to bother. Admittedly, the number of times you get foreground clipping is relatively rare... it's just a pain when you do encounter it.

As for far clipping - the idea with PVRDC is that you only bother to do gross far culling, and whack in a background plane at the far clip distance.

 

That's very interesting. Somehow, I had the mistaken impression that people buy consoles based on games, brandname, and word of mouth. Thanks for correcting me, now I know that real console gamers only care about Big Numbers of polygons and pixels.

Last time I talked to my friends about PS2 vs. GC vs. Xbox, I never heard "fillrate" or "polygon count" mentioned once. Instead, it sounded more like, "Tony Hawk", "Grand Theft Auto", "Mario Sunshine", or "Halo". Even when talking about graphics, anyone who starts spouting "Gigatexels" or "Gigabytes per Second" or "Pixel Shaders" gets rapidly shouted down by cries of "Metroid Prime", "Resident Evil", "Silent Hill", or "Halo 2". I guess my friends are just weird, after all, real console gamers pay much more attention to Polygons/sec than to how good the games look.

I mean, how else can you explain the overwhelming success of the PS2, what with its gargantuan fillrate and polygon advantage over its "big number" deprived competitors.

[/sarcasm]

The failure of the Dreamcast has nothing to do with polygon pushing and texel filling. The marketing failure was simply a complete failure of advertising and hype-building. Many people werent even aware of the Dreamcast's existence until it was nearly dead; in contrast, even someone living in a hole for the past 3 years would've been exposed to PS2, Gamecube, and Xbox hype. The Dreamcast's graphics were leaps and bounds above anything else, but they failed to demonstrate this. During the launches of the N64, PS2, and Xbox/Gamecube, you could hardly step into a Wal-Mart without walking into display consoles demonstrating the most graphically impressive games. Dreamcast came and went with hardly any buzz at all. It's this lack of marketing, not any lack of deceptive pixel-speak, that killed off the DC.

 

Mega Drive / Genesis - the 64 on screen color limit - barring any special coding like HAM - really limited the colorfullness of most games (there are some amazing exceptions like Thunder Force 3 and 4) also hampered the accuracy of Sega arcade conversions. not that they were that accurate anyway, aside from color.


Super Famicom/ SNES - the slow as molassis 3.58 Mhz 16-bit CPU. the speed issue in games was overcome by some top notch developers but still it would have been MUCH better if SNES had a fast 68000.

PC-Engine/TurboGrafx: The 8-bit CPU, and sound chip of the base system (not counting CD-ROM systems) Though it's somewhat understandable that PCE had an 8-bit processor because of it's age (released 1987) it still could have had a 16-bit CPU, like the Amiga 500 which was even older - there was NO excuse for the SuperGrafx having yet the SAME CPU as the PCE, being that SG was 1989 hardware, newer than the Mega Drive.


NeoGeo - price!


N64 - lack of CD-ROM drive, lack of enough texture cache. sound processor saps CPU and bandwidth.

-PSX - 1 MB VRAM not really enough when Saturn and NEC PC-FX had more (1.5 MB) - small CD-ROM cache made for SLOW loading times. decent 3D graphics for the time. NOT Ridge Racer arcade quality as many would like to say. could have used perspecitve correction and filtering on the textures. unspectacular but adaquate 2D and audio. pretty good overall.

Saturn - lack of any dedicated 3D geometry and rasterizing hardware. It was a sprite pushing beast but it's 3D was FAKE - using distorted sprites to make it look like 3D polygon graphics. there for Saturn conversions of Model 2 (and even Model 1) were no where NEAR approaching the arcade versions, despite what anyone says otherwise.


Dreamcast - none that I can think of. sure we all want more memory and more CPU power and more graphics, but nothing was serious lacking in the Dreamcast hardware. awesome. a stunning achievment. brought home highend arcade quality graphics (near-flawlessly with the RIGHT developer and enough time) at a FRACTION of the cost of an arcade unit. God Bless Videologic PowerVR.

PS2 - all kinds of flaws, dispite the extremely powerful "super computer" CPU and "workstation like" graphics synthesizer. needs more VRAM on GS or at least texture compression to make the memory go farther. (effective texture memory like Flipper's 6:1 TC) the bus structure needs widening in certain areas (as someone mentioned)


Gamecube - nearly perfect - but could have used at least another 8MB system memory and a couple more MB on Flipper. And a stronger T&L unit in Flipper, but overall exellent design.

XBox - should have had either much faster UMA bandwidth (12.8GB/sec at least) or seperate pools of memory, with some being very fast. Also larger cache on NV2A or a real eDRAM scheme like GC/PS2. Also should have had a better CPU. like a P4 - or at least 1 Ghz P3 with more cache, not a cross between Celeron/PIII - Actually, would have benifited from a more heavily modifed Intel/AMD CPU like Gamcube's Gekko PPC.
XGPU should not have been scaled back from NV25 down to what it is today (inbetween NV20 and NV25) - The GeForce4 Ti4600 is pretty much what the XGPU should have been, that is, an NV25 clocked at 300 Mhz. Booooooo Hissssss M$ for being so damn cheap. (in truth Nintendo did same thing with Flipper) For these and other reasons, XBox is prevented from truly outshinning the Gamecube and even the PS2.

 

Ummmm... XGPU isn't a scaled back NV25.

It's a scaled-FORWARD NV20.

Before NV25's specs were finalised, nVidia called NV2A 'a kind of NV27.5' - though it is in reality more like NV22.5...

 

AFAIK the GF4 (particularly the LMA II) was built on Nvidia's and developers' experiences with the NV2A in realworld Xbox projects...

 

At 300mhz or even 250mhz wouldn't the Xbox reach its memory bottleneck long before it became fillrate limited? I don't see how cutting costs by properly balancing your design is 'cheap.'

 

I actually cannot really disagree with you here. when you say XGPU is a scaled forward NV20, you are right in a way. it depends on how you look at it I suppose. The glass is either half full or half empty.

The original annoucment of XBox had the GPU as an NV25. Then later it was said to be NV25 with some of the geometry features of the NV30, thus the "NV27.5" comment. But in the end, the actual XBox GPU would be LESS than even the NV25.

I also agree that in reality, it is basicly an NV22.5 - That's exactly where XGPU should be placed in Nvidia's NV2X line, IMO.

 

Look carefully at who said the XGPU was like a NV27.5; it was a marketing guy! He obviously got his code-numbers mixed-up either by mistake or design, saying NV30 when he should have said NV25.

Does the XGPU have any DX9 geometry features? No. So how could it have NV30 geometry features then? Simple answer, it does not.

*G*

 

Nobody's arguing that it is NV27.5.

And it was called that by nV marketing before NV25's specs were finalised (which is why it's NV20-scaled-up, not NV25-scaled-down - nothing to scale down from yet!).

 

IIRC at the time of the "NV27.5" comment the NV25 was supposed to be what ended up as the GF3 Ti500.