"Programming the PSP"

[Fafalada]

It wouldn't surprise me if they are just using a DDA algorithm that's outputting strips. That would really only lend itself to fixed tessellation.

Ok, my english is bailing out on me again. Are you suggesting tesselation inside DDA?

You ought to get out more

Well I think he was just relating to existing portable products, in which case he's quite right
Granted, PSP isn't an existing product yet either, so comparison isn't really fair

 

[MfA]

Forward differencing is easier with fixed stepsizes ... but how about simply having a flag per edge to drop every second vertex to allow power of 2 increases in tesselation level without cracking?

 

[marconelly!]

You ought to get out more

Shame there doesn't seem to be any pictures from the "cutlery" curved surface demo for MBX. It has lots of gratuitous reflection mapping. The Utah teapot demo (for which there are pictures) also has reflection mapping but it's subtle.

Oh, I've seen pictures of MBX demos. It's nice, no doubt, but I can't really get excited about it, until I see some organized game developer effort behind it. Also, it is an up-and-coming hardware, just as PSP, not something that I can just 'get out' and buy today

Well I think he was just relating to existing portable products, in which case he's quite right

Yes. Don't get me wrong guys, I don't think PSP is some miraculous hardware, nor do I think that on a purely technological level it's much better than MBX or other upcoming mobile 3D chips. It's just that this one has pretty much sealed deal for the support of world's top developers - that's what gets me excited.

 

[V3]

Last time we had some question regarding how the MIPS core going to drive VFPU and FPU.

In light of this new info, can someone tell me what's going on with the VFPU and FPU ? Is it really a seperate unit or the same ?

 

[Panajev2001a]

Still at least they are able to support both Bezier patches and B-Splines which is better than nothing IMHO.

No, I am not taking glory away from PowerVR and the MBX, each design has its pro's and con's.

So what do you perceive the cons to be with the curved surface support on MBX?

I was talking in general, not necessarily about that feature's implementation in particular.

Perhaps speed will be ( not all HOS Tessellating engines have the same speed ), perhaps direct and indirect ( less transistors and power budget for other things ) effects on the performance of the chip as a whole.

I know too little about the exact nature of the MBX HOS support: is there a Hardware unit besides the VGP ? Is it done on the VGP ? etc...

I am just stating the obvious: I respect engineers in both companies, I think both of them have access to good technology and perhaps good books and I think that since I do not see major screwups in either design I take for granted that corners were cut and somewhere they were cut roughly.

Now that Modifier Volumes have been exposed on KallistiOS ( not trivial to use, but there is a way and we still need DOT3 ) I am quite happy about PVR and well, if the maker of my Kyro II board had some more brain I would have had to underclock it to keep it running.

Still those are small issues: I have respect and admiration especially for PVR2 CLX ( DOT3, Modifier Volumes, Hardware sorting for translucency, VQ Texture Compression, Deferred Rendering, etc... ) , but also for the path IMG Tech. has chosen.

I do not think any path is absolutely the best, but they have found so far a viable one.

I am trying to look for the patent I promised: this one contains a scheme which could work under the circustances we are talking about ( polygon sub-division ).

http://appft1.uspto.gov/netacgi/nph....&OS=AN/"Sony+Computer"&RS=AN/"Sony+Computer"



Yeah, I know they have a patent too about High Speed Bump-Mapping and we have yet to see it applied, but can I hope about any kind of DOT3 capability in the PSP's Pixel Pipeline or not ?

 

[gurgi]

ahem

 

[PC-Engine]

Island demo feature water effects that I've rarely seen even on PS2.

Well that jetski game did have similar water effects.

 

[Simon F]

It wouldn't surprise me if they are just using a DDA algorithm that's outputting strips. That would really only lend itself to fixed tessellation.

Ok, my english is bailing out on me again. Are you suggesting tesselation inside DDA?

My apologies. I should really have said "Forward Differencing". I blame it on sleep deprivation It's still DDA, I suppose, except that you are incrementing several derivatives.

Well I think he was just relating to existing portable products, in which case he's quite right
Granted, PSP isn't an existing product yet either, so comparison isn't really fair

Fair enough but these demos should probably be compared to PC or console systems of, say, 5 years ago in which case thy aren't that amazing.

Forward differencing is easier with fixed stepsizes ... but how about simply having a flag per edge to drop every second vertex to allow power of 2 increases in tesselation level without cracking?

There is a quite clever way of varying the step size but it's patented (and deservedly so). The idea of skipping vertices is interesting but you couldn't just drop every second vertex because (assuming FP maths) the edges of adjacent patches could drift apart slightly.

Oh, I've seen pictures of MBX demos. It's nice, no doubt, but I can't really get excited about it, until I see some organized game developer effort behind it. Also, it is an up-and-coming hardware, just as PSP, not something that I can just 'get out' and buy today

I believe "Edge" magazine may have had some information. As for not being available, well true, but the public demos use real silicon <shrug>

It's just that this one has pretty much sealed deal for the support of world's top developers - that's what gets me excited.

That's why I said "never underestimate the dark side".

I am trying to look for the patent I promised: this one contains a scheme which could work under the circustances....

I don't think you meant that patent - it's seems to be about a point of sale Roulette game?!!! (How is that patentable???!)

 

[Panajev2001a]

Simon, I swear that is not the patent I copied and pasted the url of.




WTF happened... bah...


I will look again and give you the patent number.