BEcause all we know that Spikeout looks better than the arcade...(for instance) and no way it looks like crap on XBOX ¬_¬'Probably the same people who will be waiting for Xbox 3 and PlayStation 4 for arcade perfect Sega Model 3 conversions, but are going to complain that "they're not exact because Model 3 was 512x384 and the new consoles have 1080p (or better) output, it's just not the same!"
Lazy8s said:No one said anything like that, of course. But nebulous conceptions about how much more "powerful" one hardware is than another will definitely not be the critical factor in making a faithful port.
Well you know, just like Virtua Racing on PS2 was horribly crippled port of 3DO version. They took out the Hardcore flat shading and replaced it with plain flat shading, bastards.Argyle said:Wow. There are people who really think the PS2 will have trouble with a port of VF2? You've got to be fucking kidding...
doncale said:yeah I cant believe Sega allowed the PS2 Virtua Racing to be glorified 32x VR, even though its 60fps which is smoother than the Model 1 arcade VR. isnt it funny and sad that many wonderful, classic games get brought home to far more powerful machine, yet the original is still the better game.
Lazy8s said:Regardless of which version PS2 Virtua Racing was based off, the environment wasn't modeled as nicely as in the arcade original. Still looked really nice, though. I have access to an arcade machine locally, so I'm sticking to that one for its deluxe cabinet and perfect controls.
Argyle said:I saw Spikeout in motion at E3, it really didn't look that different than the original, but it's been many years since I've played the arcade version. I think it's more "fanboys forgetting what Model 3 graphics really look like" than "Model 3 is more powerful than Xbox!"
Fafalada said:Well you know, just like Virtua Racing on PS2 was horribly crippled port of 3DO version.
Well I was being sarcastic, but also pointing out that one can never underestimate what some people will find wrong with each conversion. The "hardcore shading" comments in particular had me fuming...Shinobi said:I assume you mean the 32X version. It might well have been based on that version, but it's a lot closer to the arcade version then the 32X version ever was (and in some areas beats it silly).
There are extreme viewpoints for every issue, and they exist on both sides of the argument... like the overlooking of actual differences that exist in a port, resulting from whatever cause, just because the target hardware "should" be capable.what we got instead was people making up downgrades and what not.
doncale said:yeah I cant believe Sega allowed the PS2 Virtua Racing to be glorified 32x VR, even though its 60fps which is smoother than the Model 1 arcade VR. isnt it funny and sad that many wonderful, classic games get brought home to far more powerful machine, yet the original is still the better game.
God, I can't wait until that one Model 3 emulator is released so all the Segabots can shut up about how godly the hardware is.
And you guys are right, Spikeout doesn't look the same, after all, I think that they're skinning their characters now. I mean, it looks to me that Sega has only recently gotten a clue about how to write a character renderer it seems - none of the Model 3 games with human characters seemed to use skinning (I know VF3 doesn't - and I'm pretty sure Tekken 3 does, which surprised me since both games came out at about the same time and T3 is on much lower spec hardware), and even on recent games the skinning has been sub-par (look at the joints in VF4, arcade or PS2, yuck!)...
It's not any more difficult to design for programmability. It's a trade-off. Going the other way toward fixed function, there is the ELAN co-processor from the same time as PS2 and part of a system with lower silicon expense which is not as flexible yet sustains a lot more complex T&L in exchange.Kind of amazing that a machine like the PS2 would only come out a couple of years later with not just one, but a pair of very programmable vector units, and at a consumer pricepoint no less.
You know, I wonder if it's a limitation of the Model 3 geometry coprocessor? Maybe the DSP on the board was hardwired (a fixed function chip is much easier to make fast than a general purpose one) and couldn't do the skinning at all? At least on the Gamecube, which has a GPU that does skinning poorly, the CPU is fast enough to take care of it...
Kind of amazing that a machine like the PS2 would only come out a couple of years later with not just one, but a pair of very programmable vector units, and at a consumer pricepoint no less.
Going the other way toward fixed function, there is the ELAN co-processor from the same time as PS2 and part of a system with lower silicon expense which is not as flexible yet sustains a lot more complex T&L in exchange.
"SH-4" + "CLX"*2 + "ELAN" against "EE" + "GS" at the sizes they were respectively produced at when delivered in early 2000, per the original comparison.Under what conditions are you making this comparison?
Though they are two different architectures, the actual comparison was between the two geometry solutions justified in relation to their respective wholes (to indicate that the host machines were of a comparable class), and the actual significance was about monetary costs as dictated by the "pricepoint" context. RAM pricing would then come into play when considering the whole, and the two machines would then still end up being comparable context for the comparison of their respective geometry solutions.Are you considering framebuffer RAM silicon area on the Elan side?
Again, the PA quote wasn't out of context. It provided reference for characteristic PS2 polygon counts and wasn't compared to any other figure. The only comparison that was made was made direct: VF4 character polygon counts were reduced by half for the PS2 from the arcade version.Or are you just grabbing information from B3D and repeating it everywhere as fact without context? (i.e. PA figures that claim PS2 games sustain 2-5M polys/sec)
Yes, my whole point was that the trade-off was for more flexibility. Besides, ELAN's "lights" are not just more lights either; they're modifiers in a more general sense.As for more complex T&L, there are interesting things you can do with a DX9+ class 'vertex shader' other than just more lights.
Which would make it Elan+SH4 vs EE, not what you said.Lazy8 said:Though they are two different architectures, the actual comparison was between the two geometry solutions
"SH-4" + "CLX"*2 + "ELAN" against "EE" + "GS" at the sizes they were respectively produced at when delivered in early 2000, per the original comparison.
Yes, my whole point was that the trade-off was for more flexibility. Besides, ELAN's "lights" are not just more lights either; they're modifiers in a more general sense.
ot: I would've loved to have seen a 'Super Dreamcast' (not the same as a nextgen DC2)released, based on NAOMI 2. same CPU (SH4 @ 200 MHz), same audio processors (but more RAM), the 2nd PVR2/CLX, the ELAN, plus all the additional memory. kind of like the NEC SuperGrafx compared to the TurboGrafx. release an arcade identical Virtua Fighter 4, VF4 Evo, Beach Spikers, King Of Route 66, Sega Driving Simulator, Soul Surfer, Virtua Striker III and VF4 Final Tuned (just a few exclusive games) and naturally be compatible with DC games. yet also, enhance all DC ports of NAOMI 1 games back upto NAOMI 1 level...the texture res and whatnot. cool idea, no? just for the hardcore, like the SuperGrafx in 1989-1992.
No, the the VUs were brought up, and a contrast was made with the ELAN. T&L can be either run or offloaded from both the R5900 core or the SH-4.Which would make it Elan+SH4 vs EE, not what you said.
No, accounts indicate that it was delivered to SEGA in the first half of 2000. Application takes precedence over technological availability to a far greater degree with an arcade board launch than with a console. There's a delay to give time for the first AAA game to be polished and finished with an arcade board introduction, which is usually at least six months longer than traditional console launch software gets.ELAN wasn't delivered until 2001
No, the the VUs were brought up, and a contrast was made with the ELAN. T&L can be either run or offloaded from both the R5900 core or the SH-4.
No, accounts indicate that it was delivered to SEGA in the first half of 2000. Application takes precedence over technological availability to a far greater degree with an arcade board launch than with a console. There's a delay to give time for the first AAA game to be polished and finished with an arcade board introduction, which is usually at least six months longer than traditional console launch software gets.
No one said they weren't. The comparison, however, was between the VUs and the ELAN co-processors, not between the CPUs.You are not making any sense. Geometry is calculated on the PS2 on the EE (MIPS core + VU0 + VU1) and the geometry is calculated on the Naomi 2 on the SH4+ELAN. So they're comparable
Sony couldn't deliver it in 1999. They said they would, they tried, but manufacturing difficulty arose with the GS getting pushed back to a 0.25-micron process, I believe.Similarly, the PS2 chipset was complete in 1999
Lazy8s said:Argyle:
No one said they weren't. The comparison, however, was between the VUs and the ELAN co-processors, not between the CPUs.
Lazy8s said:Sony couldn't deliver it in 1999. They said they would, they tried, but manufacturing difficulty arose with the GS getting pushed back to a 0.25-micron process, I believe.
Not my point. I was saying that as the T&L solution on a hardware comparable to PS2, the ELAN's performance advantage makes it clear that the VU's had traded off for programmability instead, and vice-versa on the trade-off for ELAN. The point was that a console with the flexibility of the VUs coming out three+ years after Model 3 wasn't too hard to imagine.So you're saying that the ELAN coprocessor takes up less silicon area than just the VUs on the EE die, or is cheaper somehow? *shrug*
Every chip has sample and risk production, which can be used for devkits and evaluation, well in advance of its readiness for volume production. ELAN was no different. What's really interesting is that SEGA and AM2 evaluated both a PS2 proposal and a Naomi 2 proposal back in 1999 when planning the new arcade platform to make for VF4. Performance was a deciding factor.Interesting, but also false. I wasn't personally working on a PS2 project in 1999 but I was working for a company that was...and we did get devkits in 1999. Full speed ones, too, people had the half-speed kits months before we got ours...
So my assertion still stands.
That's all fine to compare VUs and Elan, but I was replying to the post you made here:Lazy8 said:No, the the VUs were brought up, and a contrast was made with the ELAN. T&L can be either run or offloaded from both the R5900 core or the SH-4.
It's nowhere near as programmable as a Vertex Shader - basically it's similar to GCN T&L except the lights it uses are better featured, and computed faster...Argyle said:I wonder though, if the ELAN was so wonderful, why we never saw it used in more applications? It seems you could sell a board with one of those chips stapled onto it and it would at least have DX8 style vertex shaders (then again, is the ELAN so hardwired as to not even be programmable as a vertex shader?)
Lazy8s said:Every chip has sample and risk production, which can be used for devkits and evaluation, well in advance of its readiness for volume production. ELAN was no different.
Lazy8s said:No, accounts indicate that it was delivered to SEGA in the first half of 2000.
Lazy8s said:What's really interesting is that SEGA and AM2 evaluated both a PS2 proposal and a Naomi 2 proposal back in 1999 when planning the new arcade platform to make for VF4. Performance was a deciding factor.
Fafalada said:It's nowhere near as programmable as a Vertex Shader - basically it's similar to GCN T&L except the lights it uses are better featured, and computed faster...
And like Lazy8 mentioned - you use lighting pipeline to calculate other stuff with too, like BM setups etc... which is again similar to Flippers T&L.
Just a misunderstanding. What it meant was that the comparison was of one geometry solution to the other (ELAN vs VUs), where the two system wholes were of similar class in Argyle's original context of release date and pricepoint (for which I referred to silicon expense to indicate that I wasn't contrasting a renderfarm to a home console or anything like that).To which you refered the followup about "comparing geometry solutions to their respective wholes".
No, that was when it was delivered, as in available for volume production. SEGA had both the Naomi 2 and PS2 in at least evaluation form in 1999.I'm saying that it seems to me that the devkit and evaluation chips for ELAN came out in 2000. You said so yourself earlier in this thread:
No idea about other uses of Elan. Anyway, if they could have released something for PC in 2000 with it it should have been pretty competitive - it definately has more raw transform power then your vanila GF2. 2001 would be a bit late for highend though yeah...Argyle said:I remember they really didn't put out too many PC graphics cards around that era...they would have gotten destroyed by Nvidia at that time anyway, unless they were shooting for a mid-range or value solution (but with 1+ rasterizer chips and a coprocessor, they probably wouldn't be able to win on cost either, not compared to a value card like the GF2MX)...
Me neither, but then again, we mostly just saw Sega games, and considering how long they took to get around using skinning, which was around a lot longer then dotproduct based BM...I've never seen an N2 game use bumpmapping (and judging by the gaudiness of VF4 I'd think they'd use it if they could) so I'm wondering how practical it was on N2, even with dual rasterizer chips...I don't think I've ever seen it used on a DC game, either, to be honest.
I don't think the PS2 would be able to handle those...Dice said:No, we already got a perfect port of that on Saturn. We need your RACERS Sega, your ARCADE RACERS.