ARM's Cell processor, is Nintendo behind this thing???

[Panajev2001a]

On another note, this processor seems a bit far for GBA 2, assuming GBA 2 releases around Christmas 2004.

The PR talks about a long-term collaboration which will include the co-development of this multi-core processor.

I think this connects with GCN 2 and NEC's 65 nm processor: they could have it running at even higher frequencies than 2 GHz if they pipeline the ARM cores enough eanbling quite a few FLOPS for their CPU ( Gekko after all still does 1.9 GFLOPS while the EE does 6.2 GFLOPS... so if the CELL CPU in PlayStation 3 does 1 TFLOPS I would not be surprised if GCN 2's CPU does 256-300 GFLOPS: considering that GCN 2 is another system which is more GPU oriented [it does all the T&L and Pixel Shading] this would be very good news ).

 

[nondescript]

Yeay how about GBA2!

Yeah, a SMP ARM chip for GBA2 makes sense. Good call chaphack. (Could you stop with the floppy-flop comments please? I think people would take you more seriously that way...)

I meant that seriously ... I didn't think that it might be in GBA2, but after you (Chaphack) mentioned it, I began to see how that really made some sense. I think it's too early to be counting chickens at this point, but one possible application of a SMP ARM chip is GBA2, I agree.

(Why am I always misunderstood? Just because I like CELL doesn't mean I hope everything else fails miserably... or is it because my English is that bad?)

On another note, this processor seems a bit far for GBA 2, assuming GBA 2 releases around Christmas 2004.

The PR talks about a long-term collaboration which will include the co-development of this multi-core processor.

I think this connects with GCN 2 and NEC's 65 nm processor: they could have it running at even higher frequencies than 2 GHz if they pipeline the ARM cores enough eanbling quite a few FLOPS for their CPU ( Gekko after all still does 1.9 GFLOPS while the EE does 6.2 GFLOPS... so if the CELL CPU in PlayStation 3 does 1 TFLOPS I would not be surprised if GCN 2's CPU does 256-300 GFLOPS: considering that GCN 2 is another system which is more GPU oriented [it does all the T&L and Pixel Shading] this would be very good news ).

I agree that this would be very late for a 2004 GBA2 release - but how do we know it will be in 2004? It's not like N-gage is posing a threat to GBA. If GBA launches in 2004, it will inevitably be compared to PSP - a comparision that may not bode well for GBA2. If I were Nintendo, I would wait for the PSP storm to blow over and head for a launch in 2005.

There will be huge power issues to contend with if its meant for GBA2. (The article says nothing about quad-core or 2GHz - who dreamed that up?)

Anyways, if this ARM chip is related to Nintendo at all, then it is most likely meant for hardware in 2005-06.

 

[Panajev2001a]

Going into a multiple core set-up with ARM would be a good choice as it would make it easier to have software backward-compatiblity with the GBA software.

It would be funny if for Handhelds SCE goes for a GCN like architecture for the PSP and goes with the massively parallel CELL for PlayStation 3 while nintendo does the opposite ( their portable uing multiple cores on the same chip and software T&L ).

Dual core, 8 FP ops/cycle combined ( each Vector unit would do two parallel MADD operations, a bit more than what Crazyace extrapolated in his post [2 GHz, 4 cores = 16 GFLOPS or only 1 MADD per FPU] ), 300 MHz = 2.4 GFLOPS which would not be bad ( pretty much the same as PSP's CPU ) and the next thing to worry about would be GBA 2.

The GCN, opportunely scaled here and there ( no A-RAM for example ), could be made into a GCNP or a GCN Portable and the performance of this system would not be outclassed by PSP: they would be quite close to each other in performance and features ( over-all ).

The GCNP road could have started R&D quite early in time ( around GCN's introduction ) and would be quite doable and attractive: easier to develop for and not as late to market as GBA 2 with this kind of new ARM solution would be.

The new ARM SMP processor is not completed and I do not think the software side of things I do not think things are doing better than that: games need at least 1.5 years after the set of finalized specs of the machine and structure of the development environment and programming API is given to them ( and 5 months later they should receive some sort of emulators/simulators ).

If this project finishes by mid to late 2004 ( giving them about 1 year ) then we would not see a launch line-up complete before Q1 2006.

In that case, even if the specs of Nintendo's machine are way over the specs fo PSP ( as the extra Hardware R&D time would allow you to have ), you would see PSP's SoC being shrinked to 65 nm and its over-all cost reduced by a good amount to allow the PSP to be profitable waiting to use the price drop card to force losses on their competition and to increase thier sales.

 

[Tahir2]

Why can't Nintendo do what they did with GBA and base the GBA2 on Nintendo64. Sure it won't be as powerful as PSP but it would be more than adequate for a handheld.

A handheld with so many FLOPS and whatnot might make it a liability, I mean imagine if China got hold of so much power (joke...some people do not get blatant sarcasm and satire on these boards).

 

[Li Mu Bai]

The software production costs is what worries me for the PSP. If it's boasting mostly better than PS2 specs., in order for a dev. to make a profit & truly exploit the hardware, how much must be invested? Profit margins on the GBA are huge, what will the next-gens' look like? And games selling at a lower price point no less. Unless it becomes a PS2 shovelware machine, with tons of re-releases, why dev. for it if it isn't economical?

 

[Panajev2001a]

Profit margins on the GBA are huge

Apparently not... not for a lot of 3rd parties except the very big ones and even then it could be a bit better ( unless your name is Nintendo ).

PSP will have a more developer friendly environment than PlayStation 2 did: it will be more similar to the high level libraries approach PSOne followed.

 

[Li Mu Bai]

Profit margins on the GBA are huge

Apparently not... not for a lot of 3rd parties except the very big ones and even then it could be a bit better ( unless your name is Nintendo ).

PSP will have a more developer friendly environment than PlayStation 2 did: it will be more similar to the high level libraries approach PSOne followed.

Pana you develop for the GBA correct? This was from Jarrod over at GA:

"Not sure about margins, but development costs are next to nothing. A GBA developer I spoke with said a typical GBA game can be made by a 5-6 man team in just 6-8 months, usually for $80-100k. Big budget games can cost 3-5 times that though, but those tend to be Japanese releases. He also mentioned that Nintendo charges about $5 for a 8MB cart with battery save currently, and costs go down when higher volumes are ordered."

Is this not correct?

GBA can be pretty hit or miss for publishers though. The SNES/Genesis/N64 risk of overestimating demand is still there, but things are better. Turn around times have been essentially eliminated thanks to new writing techniques developed by Oki Technologies. And Matrix Semiconductor's newly developed 8 layer stacked cell 3D ROM will allow cart to sizes to radically expand (with zero volume increase) at significantly lower costs in the first half of next year. Carts won't ever be as cheap as optical media (evn custom formats like GOD or UMD) but they'll it's far from the extreme prices that scared developers away from N64...

 

[V3]

A GBA developer I spoke with said a typical GBA game can be made by a 5-6 man team in just 6-8 months, usually for $80-100k.

No kidding. It shows I guess. Most 3rd Party GBA games are atrocious.

 

[chaphack]

Unless you are a big developer with a big title, most GBA games gained little to no profit at all. That is the truth. 8)

 

[notAFanB]

Unless you are a big developer with a big title, most GBA games gained little to no profit at all. That is the truth.

er I'd like to find some figues on that beforehand if you don't mind. btw does anyone know how the business model works in the handheld (and by that I obviously mean GBA 8) ) segment?

 

[Panajev2001a]

"Not sure about margins, but development costs are next to nothing. A GBA developer I spoke with said a typical GBA game can be made by a 5-6 man team in just 6-8 months, usually for $80-100k. Big budget games can cost 3-5 times that though, but those tend to be Japanese releases. He also mentioned that Nintendo charges about $5 for a 8MB cart with battery save currently, and costs go down when higher volumes are ordered."

Is this not correct?

GBA can be pretty hit or miss for publishers though. The SNES/Genesis/N64 risk of overestimating demand is still there, but things are better.

Those small team games developed in 6 months often end up being big misses and what is not good is the low attach/tie-in rate for the GBA: the systems are sold, but the software sales for most games are not that great.

Even on GBA the costs can add up and the market does not seem to buy enough software, not enough if you look at the Hardware sals and the current GBA use-base.

 

[notAFanB]

Those small team games developed in 6 months often end up being big misses and what is not good is the low attach/tie-in rate for the GBA: the systems are sold, but the software sales for most games are not that great.

again I don't dispute this but does anybody have any details on the kind of cost/revenue curves we are looking at?

 

[chaphack]

That is why my insiders tell me. It is whats happening in the GBA fray, oh yes, so very true.

 

[DeadmeatGA]

...

GBA is not going away anytime soon. At $70 and a steady flow of inexpensive games it will survive anything.

This new ARM chip is supposed to be quadcore and be available for public release in 18 months. Why quadcore? Isn't this a bit of an overkill for cellphones???

 

[notAFanB]

That is why my insiders tell me. It is whats happening in the GBA fray, oh yes, so very true.

I am going to ask one more time, does anyone have a breakdown (with some supporting links/articles/etc) of the how handheld devlopment costs are planned and recouperated.

thank you

GBA is not going away anytime soon. At $70 and a steady flow of inexpensive games it will survive anything.

yup and to get ppls to upgrade anything besides better visuals may be required.

This new ARM chip is supposed to be quadcore and be available for public release in 18 months. Why quadcore? Isn't this a bit of an overkill for cellphones???

god knows, maybe it's for ngage mk2?

 

[Urian]

¿Arjuna?

 

[Tagrineth]

And Matrix Semiconductor's newly developed 8 layer stacked cell 3D ROM will allow cart to sizes to radically expand (with zero volume increase) at significantly lower costs in the first half of next year. Carts won't ever be as cheap as optical media (evn custom formats like GOD or UMD) but they'll it's far from the extreme prices that scared developers away from N64...

8)

 

[MfA]

As long as you need traditional litography to produce the memory the cost differential wont change much.

CD vs. PROMs ... UMDs vs stacked-PROMs ... same old same old.

Now the memory Rolltronics is producing ... that is exciting.

 

[zurich]

Someone here mentioned a $5 cartridge fee for each 3rd party GBA game.. is that on top of the ~$5 Nintendo licensing fee? If so, that's a pretty massive bite out of a $29.99 GBA game.