What could have been for Dreamcast....

[akira888]

megadrive1988:
Considering that SDRAM at DC launch was US$1.50/MByte, the memory alone would cost 150 dollars. Not practical.

All in all I think Sega did a very good job with Dreamcast, and I think you're looking in the wrong direction if you think hardware could have saved DC, I don't think a P4/AthXP with GF4TI/Radeon 8500 would have beaten the PS2 marketing juggernaut.

If I had to guess clear out of the ether perhaps the most cost-benefit effective improvement would have been something like Elan (only not as fast) integrated onto CLX, but then again even that may have broken Sega's desired pricing structure.

You really have to admire the fact that DC can rival the PS2 at a small fraction of the silicon cost. I would like to see a "redesign PS2" thread once - now that would be enlightening.

 

[Johnny Awesome]

The $485 cost figure for launch PS2s is actually conservative. I wish I still had the link, but some reports had the figure up as high as $540.

Nevertheless, if Sega had Sony's resources they could have been extremely competitive with a $400 unit in March 2000.

They Dreamcast would probably have had even better texturing and filtering, a DVD drive, along with some dot3 and support for more light sources, but probably would not have been able to produce as many particles or polygons as the PS2.

The system would have been easily competitive and very cheap to manufacture at this point in time and also very easy to program.

They should have waited and partnered with Panasonic. They probably would have had EA support and that would have been enough I think to keep them in the running.

 

[Guden Oden]

$540 sounds like pure fairy tales to me. I can't see there's any way the machine actually cost that much to make.

It's so amusing to see all this "if Sega had done this and that, and if so-and-so hadn't happened, then DC would have been THIS way, and ruled now and forever... Sheesh, when will you guys call it quits?

 

[SegaR&D]

$540 sounds like pure fairy tales to me. I can't see there's any way the machine actually cost that much to make.

It's so amusing to see all this "if Sega had done this and that, and if so-and-so hadn't happened, then DC would have been THIS way, and ruled now and forever... Sheesh, when will you guys call it quits?

NEVER!

 

[thop]

NEVER!

That's the spirit! 8)

 

[akira888]

(snip) Sheesh, when will you guys call it quits?

NEVER!

Oh dear me....

 

[jvd]

$540 sounds like pure fairy tales to me. I can't see there's any way the machine actually cost that much to make.

It's so amusing to see all this "if Sega had done this and that, and if so-and-so hadn't happened, then DC would have been THIS way, and ruled now and forever... Sheesh, when will you guys call it quits?

well if sega had more money to throw at the dreamcast perhaps we wouldn't have the hype machines in the market and instead have gaming companys (nintendo and sega )

 

[archie4oz]

The $485 cost figure for launch PS2s is actually conservative. I wish I still had the link, but some reports had the figure up as high as $540.

It was $488... All the higher numbers were wild speculation...

well if sega had more money to throw at the dreamcast perhaps we wouldn't have the hype machines in the market and instead have gaming companys (nintendo and sega )

Technically they not gaming companies either... Nintendo is an "entertainment" company and SEGA is an amusement company...

The also, without the "hype machines" then we wouldn't have any gaming consoles...

 

[MODEL 3]

akira888:
"If I had to guess clear out of the ether perhaps the most cost-benefit effective improvement would have been something like Elan (only not as fast) integrated onto CLX, but then again even that may have broken Sega's desired pricing structure.
You really have to admire the fact that DC can rival the PS2 at a small fraction of the silicon cost. I would like to see a "redesign PS2" thread once - now that would be enlightening."

Naomi 2 was launched in Q4 2000.My guess is that Videologic wanted DirectX 7 (Q1 00) to be finalised and also to see the first implemetation of competitors efforts (GeForce 1 &2 , Radeon) before make an attempt at T&L.Also I heard that they had problems in their early effort at integrating T&L unit with CLX .So maybe in Q1 00 was possible an Elan processor (1 year before Naomi 2) but not CLX integrated.Elan in order to make the transformation and shading was supported by 32Mb ram, even if in redesigned dreamcast Elan had only 8Mb (not sufficient) we should have added another 4Mb in the CLX (3Mb for polygon storage and 1Mb for added texture memory), so it was not cost efficient solution at all.
Also the gain in performance I suppose is not much since Naomi 2 (With Quad memory in relation to DC and 32Mb supported Elan+2 CLX ) was capable according to SEGA for 10 million polygons.I guess Elan was GF2 class at best with 1/8 theoretical polygon transformation per clock GF2 200MHz 25million pps Elan 100Mhz 12.5 million pps.Even if it was GF3/Radeon 8500 class with 1/4 theoretical polygon transformation per clock then it would be at 25million pps (in my example the 0,25nm 400Mhz SH4 is capable of 22,5 million pps)
Concerning "redesigned PS2" only 3 changes (all perfectly possible) in the Graphics Synthesiser(changes in other areas was not really necessery):
1)single pass multitexturing
2)compressed textures
3)8Mb ram (since the graphics ram is embedded on the chip maybe the 8mb would affect the yields at 150Mhz but I would prefer a 250Mhz emotion engine + 125Mhz GS with 8Mb of ram since the pure polygon performance of PS2 is higher than the 1,5 year later Gamecube.I am pretty confident that with these changes the PS2 graphics would look much much better.
In another subject I haven't seen anyone in the forum reffering to the quality of the output of the consoles.I mean that in a primitive (basic) simple picture (like drawing a 2D picture or a simple no special effect single 3d model) the gamecube (like dreamcast) has superb color quality and level of antialiasing in contrast many PS2 games seems blurry with 16bit kind of quality textures and low-res and many XBOX games have washed out colors and a very non unified feel like cardboard graphics with no depth.
It is very important the output to have a very heavy like image, the colors to be fat.What I mean is that the image should give you the impression that there is hidden information inside every pixel that the eye can't determine (like in real life the level of detail is infinite).With this way you have a very attractive and pleasant image quality without the aid of other special shading effect.Model 3 was another good example of this.

 

[jvd]

Also the gain in performance I suppose is not much since Naomi 2 (With Quad memory in relation to DC and 32Mb supported Elan+2 CLX ) was capable according to SEGA for 10 million polygons.I guess Elan was GF2 class at best with 1/8 theoretical polygon transformation per clock GF2 200MHz 25million pps Elan 100Mhz 12.5 million pps.Even if it was GF3/Radeon 8500 class with 1/4 theoretical polygon transformation per clock then it would be at 25million pps (in my example the 0,25nm 400Mhz SH4 is capable of 22,5 million pps)

I'm sorry are you sugesting that the geforce 2 was capable of 10 million polygons with 6 light sources ? I don't think the geforce 3 or 8500 was capable of that .

 

[Simon F]

I'm sorry are you sugesting that the geforce 2 was capable of 10 million polygons with 6 light sources ? I don't think the geforce 3 or 8500 was capable of that .

I think you should also stress that it was with 6 spot lights, i.e. not just trivial directional/parallel lights. FWIW, the 10million figure was also a conservative, measured figure and not the peak.

 

[jvd]

I'm sorry are you sugesting that the geforce 2 was capable of 10 million polygons with 6 light sources ? I don't think the geforce 3 or 8500 was capable of that .

I think you should also stress that it was with 6 spot lights, i.e. not just trivial directional/parallel lights. FWIW, the 10million figure was also a conservative measured figure and not the peak.

thank you for adding that simon . 8)

 

[schmuck]

Honestly i dont see the need for an alternative dreamcast, i think dc was powerful enough. If it had more top developers developing for it i think the the differance between ps2 and dc wouldve been almost diminishing.

Personally i think a better solution would have been for microsoft to cancel the xbox and adopt the dc. It would have been a win win for both partners. Then ms wouldnt have to cut xbox lifespan and they would have a more tranceperant transistion between the generations, and they wouldnt be loosing money on hardware. Not to mention the fact that they would have 6-8 milion users from the get go. And sega wouldve have all the benefits of a first party developer.

 

[Panajev2001a]

I'm sorry are you sugesting that the geforce 2 was capable of 10 million polygons with 6 light sources ? I don't think the geforce 3 or 8500 was capable of that .

I think you should also stress that it was with 6 spot lights, i.e. not just trivial directional/parallel lights. FWIW, the 10million figure was also a conservative, measured figure and not the peak.

I would hope that a mostly fixed-function dedicated T&L unit with almost the same amount of Transistors as the Emotion Engine itself would be fast at T&L.

 

[Simon F]

I would hope that a mostly fixed-function dedicated T&L unit with almost the same amount of Transistors as the Emotion Engine itself would be fast at T&L.

I'd expect it'd be almost an order of magnitude faster.

Sorry... were you implying that Elan was as large as the EE?

 

[Panajev2001a]

I would hope that a mostly fixed-function dedicated T&L unit with almost the same amount of Transistors as the Emotion Engine itself would be fast at T&L.

I'd expect it'd be almost an order of magnitude faster.

Sorry... were you implying that Elan was as large as the EE?

Doesn't the Elan have ~10 Million Transistors ?

The EE has 13 Million Transistors (10.5 Million Transistors was the 1999 specification for the 250 MHz EE, most of the additional 2.5 Million Transistors were due to bug-fixes and also some other optimization [no new features were added]).

 

[MODEL 3]

jvd:
"I'm sorry are you sugesting that the geforce 2 was capable of 10 million polygons with 6 light sources ? I don't think the geforce 3 or 8500 was capable of that"
Simon F:
"I think you should also stress that it was with 6 spot lights, i.e. not just trivial directional/parallel lights. FWIW, the 10million figure was also a conservative, measured figure and not the peak"

Yes I suggested that Elan was geforce 2 or Geforce 3/Radeon 8500 class.But lets just say the my estimation of Geforce2 class was conservative in relation with Elan power, are you suggesting that my second estimation of Geforce 3/Radeon 8500 class (Q3 01) is also conservative?
Sega suggested 10 million pps in many interviews (they didn't mention anything about light sources).But In the Segatech site they report that: "Naomi 2 has a dedicated geometry coprocessor to handle transformations and lighting which is rated at 10 million polygons per second with 6 light sources".They don't clarrify the type of light (ambient, parallel, point and spot?).Even if Elan was capable of 10 million pps with six lighs sources (which I doubt having seen Naomi 2 games) then my suggestion is correct since a Radeon 8500LE (250Mhz) can to 10 million pps with 8 lights at 1600X1200 under Win and DirectX environment (not the best way to see the actual capability of the hardware level of 8500).Also in Segatech site the report that Naomi could do actual 3,5 million pps sustainable (but they are not reffering to light sources).In any case my second estimation of Radeon 8500 class I guess is correct.

 

[Simon F]

Model 3:
Let's just say that I'm pretty confident of my knowledge of what's in Elan.

 

[Panajev2001a]

Model 3:
Let's just say that I'm pretty confident of my knowledge of what's in Elan.

What do you know ?

Admit it that you are IMG Technology's Janitor once for all :lol.


















J/K .

 

[archie4oz]

I'm sorry are you sugesting that the geforce 2 was capable of 10 million polygons with 6 light sources ? I don't think the geforce 3 or 8500 was capable of that .

A GeForce2 GTS would do about 2.5-3million with 8 lights... Probably around 2mil with spots... I've gotten between 4-10million though on an 8500 with 8 spot lights... (batching and driver were the main hangups, one could probably extract more performance from a console type environment)