Hmm...this coupled to MBX Pro might work well.
- Thread starter PC-Engine
- Start date
Lazy8s said:
Oh come on, hikaru was only like a dc + 50%. Probably similar performance to a kyro card.
It was impressive though, planet harrier(I believe that was the name) could hang with many gamecube games(wouldn't be the best, but it wouldn't but it wouldn't be thw worst, it was like a dreamcast game on steroids), and it ran at an impressive 60 fps, something that is seen too little on recent games on just about any platform.(gamecube seems to have a fair share, but most of its games fall more into the smooth 30 fps so it still isn't enough)
And it's so frigging hard to get a computer running modern games at 60 fps...it took my athlon xp overclocked to 2.45ghz before I could get stable 60 fps nearly 100% of the time in need for speed underground at high resolution with all effects on and some anti aliasing and anisotropic filtering.(I wonder why it took more cpu power at the higher graphical settings and not more graphics card power? guess my old 2.13ghz wasn't powerful enough to fully utilize the 9700 pro) Ridiculous that the xbox version ran at 60 fps(not sure if it was constant) yet my 2nd computer with a geforce 3 and a 2200+ can only maintain an unstable 60 fps at medium graphics settings.(tends to drop when computer controlled cars are around, especially if they're navigating a difficult turn) UT2004 seems impossible to get a constant 60 fps in onslaught and assault modes....halo it can be done on my main computer at low resolution, but my 2nd computer is stuck at 30 fps.
PC-Engine:
Along that line, though, compatibility with Atomiswave would make a little more sense, but even that wouldn't seem too useful considering the two platforms are aiming for opposite ends of the market.
I don't think backward compatibility will be much of a consideration this time around. By the time this new board comes out, development will have moved on from Naomi and Naomi2, and those older arcade units will still be around to support their own games (like Virtua Fighter 4: Final Tuned.) Backwards compatibility is really only useful to the arcade industry when there is continued development for the previous generation overlapping with the new generation. Also, part reuse from the earlier board isn't as likely for the new unit as it was when Naomi2 implemented the CLXs, so that pushes the potential for legacy support farther away.
Along that line, though, compatibility with Atomiswave would make a little more sense, but even that wouldn't seem too useful considering the two platforms are aiming for opposite ends of the market.
Fox5:
Some Firefighters and Star Wars shots:
Yeah it was impressive, like Star Wars Racer Arcade and Brave Firefighters which it also ran at a smoking 60fps in VGA proscan. What I was saying about NASCAR, though, was how it captured the effect where direct sunlight tends to create strong contrasts between the brightness of exposed surfaces and the almost blackness of shaded areas.
Some Firefighters and Star Wars shots:
Fox5 said:
Star Wars Racer arcade was indeed running on a Hikaru board.
Lazy8s said:
You reading too much into thoses shots, Lazy8s, imho.
NASCAR had a very simplistic lighting, if it looks good, for you, it boils down to the art direction, at best.
Hikaru's specialty was particles. It was designed to render fire and water more realistically as DC/NAOMI wasn't capable of those effects. You can see the cool effects from Hikaru in Brave Fire Fighters. Download this video.
http://www.segaarcade.com/movies/bff.mov
Polygons : 2 Million polys a sec
Shading : Phong Shading
Lighting : Horizontal, Spot, 1024 lights per scene, 4 lights per polygon, 8 window surfaces.
Effects : (at least) Fog, Depth Queueing, Stencil, Shadow, Motion blur
http://www.segaarcade.com/movies/bff.mov
Polygons : 2 Million polys a sec
Shading : Phong Shading
Lighting : Horizontal, Spot, 1024 lights per scene, 4 lights per polygon, 8 window surfaces.
Effects : (at least) Fog, Depth Queueing, Stencil, Shadow, Motion blur
PC-Engine said:
Eh, I'm still pretty sure hikaru was just a naomi with a higher clock speed and more ram, both things which would help for particle effects and fire and water.
2 million polys....that's for the hardware I suppose? If that's without tbr, that makes it twice the power of dreamcast, which makes it arguably more powerful than naomi 2.
Lol! That was a funny video! The fires looked 2d, but that's besides the point, I LOVE how sega makes every one of their arcade games(and most of their console games) overly dramatic! And did you check out the civ from house of the dead 2 who was at the bar at the end? I think he died in the same pose in both games!
CIN said:
Were the graphics as good as naomi 2? I've seen 1 hikaru game in person(maybe more, but I wouldn't have been interested and would have just passed over them, so I've only extensively played one), but haven't seen any naomi 2 games in person, and I don't even think any exist that look like planet harriers did.
Megadrive1988
Veteran
Hikaru couldn't have been as powerful graphically as NAOMI 2, but better than NAOMI. Hikaru supposedly had some custom Sega or ImgTech graphics processor in addition to the PowerVR2DC. or maybe it was a custom PowerVR2DC. don't really know. but it was used in Star Wars Ep 1 Racer and I think Brave Fire Fighters. SW Ep1 racer had extremely smooth framerate, while BFF had water and fire effects that NAOMI couldn't do. might be wrong on BFF though...going to check around google then edit in/out whatever i find might be wrong with what ive said.
edit:
http://www.system16.com/sega/hrdw_hikaru.html
so this board could do phong shading? didn't know that. also, more lighting than NAOMI it seems, though not as much as NAOMI 2.
while System16.com has ALOT of arcade technology info, their accuracy is sometimes shoddy and/or incomplete.
edit #2:
http://www.segaarcade.com/press_htm_docs/bravefirefighter.htm
I would consider that official info on Hikaru.
edit:
http://www.system16.com/sega/hrdw_hikaru.html
so this board could do phong shading? didn't know that. also, more lighting than NAOMI it seems, though not as much as NAOMI 2.
while System16.com has ALOT of arcade technology info, their accuracy is sometimes shoddy and/or incomplete.
edit #2:
http://www.segaarcade.com/press_htm_docs/bravefirefighter.htm
I would consider that official info on Hikaru.
Sounds like all hikaru did was add raw power...so basically sega just overclocked some naomi boards?
BTW, I believe the naomi could do 1 million polys before tile based rendering is taken into account, if naomi2 is just two naomi2 chips than it would do 2 million, if hikaru does 2 million without taking tile based rendering into account than it is as powerful as naomi2 and possibly more so because of the efficiency of 1 chip versus 2.(would be very slight I suppose....anyone have any info on how the powervr multichip set up worked?) Well, it'd be weaker on the t&l and lighting I guess.
BTW, I believe the naomi could do 1 million polys before tile based rendering is taken into account, if naomi2 is just two naomi2 chips than it would do 2 million, if hikaru does 2 million without taking tile based rendering into account than it is as powerful as naomi2 and possibly more so because of the efficiency of 1 chip versus 2.(would be very slight I suppose....anyone have any info on how the powervr multichip set up worked?) Well, it'd be weaker on the t&l and lighting I guess.
PC-Engine said:
I dunno, if sega could design their own graphics hardware like that, I don't think they would have needed to go to anyone else for the graphics chips for their consoles. Doesn't seem worth it to design their own graphics chip either when powervr tech probably could have done the job.
Fox5 said:
Not sure what you mean by "the naomi could do 1 million polys before tile based rendering is taken into account". Naomi/DC could draw something like 7 million pps, and calculate more as far as I can remember. Sega gave the peak "in game" performance to be something like 3 million pps, though this was possibly never actually reached.
Sega claimed Naomi 2 could realisically do 10 million pps with six lights, though I remember the peak theoretical figure being about 12 million pps.
Even the bog standard Naomi 1 system, made using todays processes and clocked at 400 mHz for the GPU and 800mHz for the SH4 (or even 500/1000), would probably make for a cheap and pretty powerful arcade board with low development costs (that could easily maintain backward compatability with Naomi 1 and Atomiswave). Double or quadruple the memory (64-bit DDR 400 being very cheap these days), and I imagine you'd have a fine budget board.
Fox5:
The Dreamcast's fillrate is 100 MegaPixels per second. Because that only gets used on the visibile part of the image, it's effectively several times higher than that when you account for a game's overdraw - this is necessary when comparing with most other architectures since they draw over pixels already rendered. The DC's effective theoretical maximum fillrate is 3.2 Gpixels.
What you're thinking of is overdraw - not tile based rendering. And overdraw is usually mentioned in reference to fillrate (pixel drawing speed). Basically, PowerVR's performance is used exclusively on visible parts of the image, so none of the polygons it renders are later hidden by other polygons and lost. They're referred to as being fully-featured, and the CLX part (Dreamcast, Naomi) can do up to 7 million fully textured, lit and shadowed polygons per second.
The Dreamcast's fillrate is 100 MegaPixels per second. Because that only gets used on the visibile part of the image, it's effectively several times higher than that when you account for a game's overdraw - this is necessary when comparing with most other architectures since they draw over pixels already rendered. The DC's effective theoretical maximum fillrate is 3.2 Gpixels.
Naomi 2 uses an ELAN geometry co-processor to drive T&L, and it therefore sustains complex conditions around 10 million polygons per second with six fully-featured lights.
Megadrive1988
Veteran
the way I understand it:
Dreamcast / NAOMI PowerVR2DC / CLX gets 100 million pixels per second raw fillrate but it was also rated at 200 million pixels per secend because of the average overdraw. overdraw on that chip is usually 2 to 3 times, therefore making the 100 million pixel/sec fillrate equivalent to 200~300 million pixels/sec in fillrate of conventional rasterizers like Voodoo3 or TNT2.
NAOMI 2, with its 2 PowerVR2DC / CLX chips has a raw fillrate of 200 million pixels/sec and that would be like equivalent to 400~600 million pixels. still less than Gamecube's Flipper (648M pixels/sec) or Xbox's NV2A (932M pixels/sec)
Dreamcast / NAOMI PowerVR2DC / CLX gets 100 million pixels per second raw fillrate but it was also rated at 200 million pixels per secend because of the average overdraw. overdraw on that chip is usually 2 to 3 times, therefore making the 100 million pixel/sec fillrate equivalent to 200~300 million pixels/sec in fillrate of conventional rasterizers like Voodoo3 or TNT2.
NAOMI 2, with its 2 PowerVR2DC / CLX chips has a raw fillrate of 200 million pixels/sec and that would be like equivalent to 400~600 million pixels. still less than Gamecube's Flipper (648M pixels/sec) or Xbox's NV2A (932M pixels/sec)
Similar threads
