Consoles have never been top-notch tech. They've always been RAM-starved, corner-cutting machines. And the fact is that Nintendo's always kept pace with their competitors, more or less exceeding them whenever they launched later. So they cut some corners where their competitors didn't--big whoop, doesn't nullify the point. The Wii is a complete departure from that. I have no doubt that the old Nintendo would be launching a machine right now with a DX9-class graphics chip and a multicore CPU, and that further they would have found a way to do it and be profitable from launch day onward.
WiiGeePeeYou (Hollywood) what IS it ?
- Thread starter Megadrive1988
- Start date
- Status
Consoles have never been top-notch tech. They've always been RAM-starved, corner-cutting machines. And the fact is that Nintendo's always kept pace with their competitors, more or less exceeding them whenever they launched later. So they cut some corners where their competitors didn't--big whoop, doesn't nullify the point. The Wii is a complete departure from that. I have no doubt that the old Nintendo would be launching a machine right now with a DX9-class graphics chip and a multicore CPU, and that further they would have found a way to do it and be profitable from launch day onward.
Yeah, I think I have about five posts on various pages in this thread saying that too. Wii is a major strategy change (or an extreme of what they've been doing all along). And a bummer that is, too. Of course, they're previous consoles post-SNES haven't exactly been major successes, so they do need to do something different, and basically shifting audiences with their console vs. the competiton is definitely one way to do it.
It's not realistic to think that Nintendo could go head to head with Sony or Microsoft in a hardware war though. They just couldn't sustain the insane losses like those companies which are way more diversified with other major money makers.
I wouldn't call N64 a major loss, it was nintendo's most profitable console ever (unless the NES was more profitable, but I'm almonst 100% certain SNES wasn't and I'm unsure of the NES) and they'd probably kill for that kind of market share again. (what, around 35 million?) I think it could qualify as a major success, even if it wasn't market leader, just in absolute sales and profits.
N64 got raped by three things above anything else:
1. Memory latency
2. Texture buffer
3. Nintendo
The irony behind the cartridge is that accessing the cartridge ROM was actually faster than accessing main memory. In a way, one could say that the cartridge both damned and saved the N64.
The 4kb texture buffer was absolutely ridiculous. Any texture had to fit in its entirety in that 4kb buffer, including all applicable MIP maps.
And Nitendo only allowed most devs to use certain microcodes for programming the system. True low-level access was never allowed by Nintendo. Emulator authors have estimated that N64 could easily churn out some 500-700,000 "PS1-quality" polygons, but no dev was allowed to access the stripped down quality mode.
1. Memory latency
2. Texture buffer
3. Nintendo
The irony behind the cartridge is that accessing the cartridge ROM was actually faster than accessing main memory. In a way, one could say that the cartridge both damned and saved the N64.
The 4kb texture buffer was absolutely ridiculous. Any texture had to fit in its entirety in that 4kb buffer, including all applicable MIP maps.
And Nitendo only allowed most devs to use certain microcodes for programming the system. True low-level access was never allowed by Nintendo. Emulator authors have estimated that N64 could easily churn out some 500-700,000 "PS1-quality" polygons, but no dev was allowed to access the stripped down quality mode.
Not even close.
For small transfers the cartridge was actually slower than a CDROM drive.
Unlike the SNES or Genesis where it was common to run code out of the cart using the cart on the N64 for anything during gameplay was a lot of work. We used it to stream animations and audio data, but that was it.
The actual situation was worse, if you used trilinear, the cache had to be split in half and adjacent mip levels had to be in alternate halfs.
A number of devs were given low level access late on, and rare had it for most of the lifetime of the console (as did lucas arts if the rumours are to be believed).
It's pretty easy to work out which devs rolled their own, because their games don't run on the emulators.
I'm not even convinced very many N64 devs would have strayed too far from the supplied uCode even if they had access as bad as the original Fast3D uCode was Fast3D2 was actually pretty good. The uCode tools were crap, and writing workable uCode on N64 was at best time consuming.
I guess it really depends how you classify top-notch tech. Do you mean absolute best one can buy, Top 5% of PC consumers at the time of console launch, 10%, what? Taking the Top 5-10% approach I would say that current consoles (and maybe even the Xbox1) were top-notch tech in relation to what the top 10% of consumers had on the PC.
Take the 360. I cannot find the 2005 results from the Valve Survey, but I will use the current 2006 ones as a frame of reference (even though they don't represent the 360 launch window). This is a survey of gamers, so it will obviously lean toward that market, which we can then look at what the Top 10% of gamers have.
The 360 CPU has some tradeoffs compared compared to what Intel/AMD offered. Valve doesn't track cache, but we do know that very top of the line PC CPUs had more cache (although I it seems pretty certain that the bottom of the top 10% did not have more than 1MB) and are more effecient; but Xenon had 3 cores, beefy VMX units, cache locking with the ability for the GPU to read from L2 cache. Even now less than 3% of gaming PCs have more than 1 CPU. You wouldn't want Xenon for your desktop CPU as of today, but would devs be better off in the long run with a single core 2.2GHz Athlon64 (i.e. the bottom of the Top 10% of AMD CPU users) or a 3.2GHz Tri-Core Xenon?
On the GPU side, looking at today, less than 5% of gaming machines have NV GeForce 7800 series (or better) or ATI Radeon X1800 series (or better) GPUs. Xenos is easily in the top 10% of 2006, let alone 2005. Looking at its featureset and other design features (like eDRAM) I don't think anyone could argue that in the 2005 launch time frame that Xenos was not a "top-notch" design relative to what the market offered. Further, the majority of the market has GPUs very much below this performance envelope.
Memory is an interesting one. On the PC there is a lot of unefficient use of memory. Bloated CPU and background tasks plus a lot of games end up storing the same content twice (e.g. lets say you have 300MB of textures on your GPU, frequently those 300MB of textures are in your system memory as well). So it is difficult to make a straight line comparison. Right now about 11% of users have 1GB or more of system memory. Interestingly back in March more people had 128MB of system memory than users who had 1GB or more (now 128MB users is down to 7%; so there has been some movement on both). I don't have fall 2005 numbers, but it does seem the 360 was on the boarder of the bottom of the top 10% of gaming rigs. And that is not considering system differences like OS size and platform stability/target.
While there are some trade-offs, I don't think we can say that in Fall 2005 the Xbox 360 was not "top-notch" technology relative to the market at that time if we are looking at a what the top 10% of gamers had and what was available in volume.
True, in fall 2005 you could have got something with 2GB of system memory, SLI 7800s or Crossfire X1800s, and an AMD Athlon64 X2 -- if money was absolutely no object.
But if we are going to look at such systems as "top-notch" when they are less than a fraction of 1% of a market, where do we end? Do we start comparing super computing clusters with thousands of CPUs and servers with 16GB of memory? What qualifies as top-notch?
All I know is that the hardware in the PS3 and Xbox 360 in the 2005 timeframe was excellent compared to the consumer market -- not only what was used but also what was available.
I guess the fact that looking at the GPUs alone, in Fall 2005, someone would have to think hard between Xenos/RSX or replacing it with an X1800/7800 (the best on the market at the time) in a PS3/360 in the 2005 timeframe is a good indication they do have top-notch hardware... at least in some areas.
Anyhow, the market has changed. Console ARE getting better hardware, relative to past generations; but likewise a much broader consumer market has opened up in the PC market where consumers can now stick 4 GPUs in their PC. So where we draw the lines on "top-notch" is blurred in this regards because outside of the VoodooII, very rarely have we seen such a situation where the top end GPUs can almost immediately be multiplied in performance by a factor of 2x or 4x by just tossing more stuff in. And now this is applying to CPUs as well...
tongue_of_colicab
Veteran
Snes games looked alot better and thats a fact. And framerate issues? never seen any. Even graphically intense games like the donkey kong series run smooth all the time.
Now you are making the mistake of blaming the hardware for the devs incompatance to make a game that runs smooth. Again, its a fact that GC hardware is better than ps2 hardware and is also capable of things the ps2 isnt.
With 3 seconds of googling you can find all the DS specs you want. And yes they will hide the Wii specs if they can because ever since they introduced wii they said they didnt want to fight the performance war, thats why they dont release numbers because its useless, they are trying to do something different with the Wii.
I never played a snes game with framerate issues.
Dude... All consoles nintendo released where faster than the other consoles from the same generation so Nintendo did use ''top notch'' hardware. That it might not always looks that way from the games it doesnt mean it actually is better hardware. ps3 games wont look alot better (ifany) than x360 games, now did sony released a low tech console because they wont have alot better looking games while launching a year later than x360?
i think you must be in denial. even from the start there were framerate issues on SNES. Super Castlvania 4, Contra III: The Alien Wars, Super Ghouls 'N Ghosts... i think if you listed all of the games with framerate issues you'd have to use multiple posts because you'd reach the character limit pretty quick. i'm not saying all games had framerate issues, but quite a few of them did.
again, it comes back to your definition of better. the PS2 is ahead of GC in more than a few ways as well. PS2 has better fillrate, higher color depth, larger storage media, a more programable polygon pipeline, and a technologicaly superior controller (all analog buttons). and before you think i'm some rabid sony ******, i prefer to play console games on my GC more than i do on any other console i own, all things being equal. my GC collection at one point was larger than my PS1+PS2+XB collections.
oh really? well please, with your googling skills find me the manufacturer of DS's GPU, it's full featureset, it's fillrate, texture cache (if any), and information on the memory controller.
Your version of "top-notch" is a moving target. The SNES beat the Genesis (which less powerful my 286 with its VGA graphics) in every single aspect except CPU clockspeed and capable of some pretty impressive graphics tricks compared to Genesis (transparent sprites, Mode 7, etc) and that makes it "disappointing." Yet when the X360 doesn't beat even a midrange gaming PC in main bandwidth, storage, screen resolution, and possibly achievable fillrate (judging by the lack of AA and AF in games), but now it's "top-notch." You can talk about PC RAM inefficiency all you like, but when the rubber meets the road, 512 MB just isn't all that huge.
I think we can just summarize this by saying that systems you like are top-notch, and systems you don't like are disappointing. Because you certainly don't have anything resembling an absolute standard.
I think we can just summarize this by saying that systems you like are top-notch, and systems you don't like are disappointing. Because you certainly don't have anything resembling an absolute standard.
Last edited by a moderator:
fearsomepirate said:
EDIT: Nevermind, no point discussing an issue with someone who results to immediate name calling, mischaracterizations, and inability and unwillingness to deal with the points laid out.
I never said, nor hinted, at what you say above. I used statistics and my own opinion (which I did not force on you or others) on how to look at a complicated topic. I am sorry you are unable to discuss the topic without personal insults.
Last edited by a moderator:
Name-calling? OK, I'll deal with the immediate points:
The original claim was that Nintendo has historically kept up and often exceeded the competition. You denied this, citing your disappointment with the SNES as proof that Nintendo has apparently never cared much about hardware. But considering the following two things, that is an unfair judgment:
1. The Genesis was never a very impressive machine. The 80286 processor hit speeds of 8 MHz in 1982; by 1989, the year the Genesis launched, this was no longer a "high-end" processor. For reference, my family got a 286 (I think it was 12.5 MHz, possibly 20?) in 1989 with VGA graphics (which were introduced in 1987), thus making the Sega Genesis look less than "top notch." Except for sound (no Adlib or Sound Blaster for us; PC speaker only for us), it beat the Genesis in just about everything. Genesis certainly wasn't a top-notch system.
2. With its 4 sprite planes, bigger sprites, vastly larger palette, 256 colors, scaling, rotation, transparency, bigger cartridges, more RAM, much superior sound chip, Mode 7, more sprites on screen (128 vs 80), and an infinitely superior standard controller, SNES beat Genesis in every single regard except pretty much framerates (which became less of a problem as programmers got better). SNES's superiority was immediately apparent with F-Zero and Pilot Wings.
Thus, it is entirely fair to say that Nintendo spent the R&D money necessary to keep pace with the competition in the 16-bit era. The same could be said with the N64.
And as I said, consoles have always been RAM-starved. Even taking coding inefficiencies into account, 512 MB just isn't a lot when it's total RAM, as anyone who's gone to make sandwiches during an Oblivion load screen can tell you. And looking at the Valve survey, the top 10% of users have 1-2 GB of system RAM and 256 MB of VRAM. So no, I wouldn't consider 512 MB to be top-notch at all, not compared to the 1280 MB the top 10% of gaming PCs are apparently playing around with. And a 20 GB hard drive these days? Yes, technological corners are being cut.
Further, what about that video chip comparison? Your objection is that market penetration of video chips that can compete with X360 is pretty small. And sure, high-end DX9 card don't make up the majority of the market (although I think at this point, X360 games don't look much better than what 6800's are doing). But when N64 launched, how many people were gaming with Voodoo graphics (consider also it was supposed to launch in 96)? When Gamecube launched, how many folks had graphics cards that could pull off the effects we saw in Rogue Leader? So I think it's pretty clear that Wii is the first machine where Nintendo hasn't been trying in the slightest for competitive graphics. Maybe their previous machines weren't as comparitively cutting-edge as X360 was when it launched, but they were a lot more impressive than the Wii is.
You sure? I've been told otherwise - that N64's RDRAM has rather abyssmal latency and good bandwidth while the catrdige has lower latency but weak bandwidth.
Wait. You're ERP, of course you're sure, you've actually worked on the damn thing. =) Gonna have to talk to a few people about this one.
Did any N64 games actually use trilinear filtering though?
Not many games have significant problems on emulators, though. The only ones that really, REALLY hate emulation are Rogue Squadron and Battle for Naboo. Even RE2 will at least start up in almost complete HLE, and run somewhat.
Did you ever see the "low quality" performance I mentioned? Was it really as high as I've been told? And do you think low latency SDRAM would've improved performance, assuming everything else stayed the same?
On cart speed yes I'm sure.
It was basically free from a performance standpoint so yes we used it for items where we could live with the reduced resolution, and it helped. I believe Rare "used it" to get free "multitexturing" since it was really just a blend between what was in either half of the cache with one set having half the resolution.
Most people just used what was there, so they run fine.
Rare only made minor changes to the lighting calculations if what I was told is true.
Neither World Driver or Stunt Racer64 run, because they have radically different uCode, and use a different synchronisation mechanism.
Many with game with slightly modified uCode work because the command packet layout wasn't dramatically changed by the developer.
Depends what you're measuring.
WDC transforms about 100-150K tri's per second, and it take a very small percentage of the RSP time to do that. Triangle setup including perspective correction etc. on the other hand easilly eats the rest of the RSX time. We actually clipped the tris on the main CPU to give us more time for tri setup.
FWIW these we're significantly higher quality polys that on a PS1, and I doubt any PS1 games actually hit close to that number. The ones I've run through emulators show much lower numbers.
The Turbo 3D uCode which devs were not allowed to use would push about 500K tris, but it's below PS1 quality tris.
It was basically free from a performance standpoint so yes we used it for items where we could live with the reduced resolution, and it helped. I believe Rare "used it" to get free "multitexturing" since it was really just a blend between what was in either half of the cache with one set having half the resolution.
Most people just used what was there, so they run fine.
Rare only made minor changes to the lighting calculations if what I was told is true.
Neither World Driver or Stunt Racer64 run, because they have radically different uCode, and use a different synchronisation mechanism.
Many with game with slightly modified uCode work because the command packet layout wasn't dramatically changed by the developer.
Depends what you're measuring.
WDC transforms about 100-150K tri's per second, and it take a very small percentage of the RSP time to do that. Triangle setup including perspective correction etc. on the other hand easilly eats the rest of the RSX time. We actually clipped the tris on the main CPU to give us more time for tri setup.
FWIW these we're significantly higher quality polys that on a PS1, and I doubt any PS1 games actually hit close to that number. The ones I've run through emulators show much lower numbers.
The Turbo 3D uCode which devs were not allowed to use would push about 500K tris, but it's below PS1 quality tris.
JacksBleedingEyes
Regular
You know what fearsome pirate, you're whole babble just falls apart with the fact that you dont really know what you're talking about. And the most ovious example is the 256 vs 64 colors argument. Firstly, the snes had 15-bit color depth (RGB555) for a total of 32,768 possible colors. The Genesis/Megadrive on the other hand had 512 colors (1536 using shadow/highlight mode). For the sake of simplicity lets just say its was 16 bit vs 8 bit. And not 8 bit vs 5 bit color.
Example of Nintendo cutting costs effectively on their consoles however is the fact that the Snes, N64 and Game Cube all shared the same Video cables. I lost my 64 cable and just use the GC instead.
That's what I was thinking of.
Actually, SNES is limited to 256 colour pallettes, but they can be changed on a per line basis, whcih is what allows the system to display several thousand colours. The 15-bit colour depth is mainly for add/subtract - aka transparency, of which Genesis was outright incapable.
I'd say you're the one who doesn't know what he's talking about.
For reference: http://www.zophar.net/tech/files/snesgfx.zip
From Zophar's description of the file -
For comparison: http://www.zophar.net/tech/files/Genesis_Tech_Overview.html
Quote from that overview -
Hence, the "256 vs 64 colors argument".
Last edited by a moderator:
- Status
