How to do Next-Gen Graphics

[A176]

Now that the 360 is approaching launch, the myriad of interviews with developers about the technical aspects of the console are becoming a dime a dozen.

While the initial comparisons against the PS3 seemed to have died off, one prevailing thought seems to have remained, that being the 360 is just plain easy to use.

As a useless reminder to those out of the loop, this is in stark comparison to the comments about the PS3, touted as being a "nightmare" to adapt to.

However, I must beg to ask the question: are there merits in both of the approaches?

Its a tried and true concept that programmers and insistently lazy; MS of all companies is aware of this ( ).

If something takes 10 seconds to complete, giving it enough power to complete that same task in one second instead without changing the code is a no brainer. (Heck, the way MS talks about XNA, they make it seem like it'll write the code for you!)

But how about taking the fresh new look at it? Again, its a tried and true concept that programmers are insistently looking for new, better ways to do things (*salutes shader branching*). Taking that same 10 second task and making it run faster by changing the code is even better than the above example (imo), albeit being more difficult to code for in the end (sorry for the weak example).

My horrible examples aside, I'm interested to hear other people's thoughts...

 

[jvd]

well an easy to program for console will see better games quicker from a visual standpoint and cheaper games .

A difficult system will take along time to max out and be more expensive


SO it all depends on console generation. Ms is launching in 2005 , if they keep to thier cycle we will see an xbox 3 in 2009 giving this unit 4 years . If sony sticks with thier 6 year cycle being harder to program for might hurt them alot as they will be up against a new xbox and still have naother 2 years with thier ps3 . Or worse they will have invested alot of money for a platform that only lasted 4 years before being upgraded and never got fully used

 

[Titanio]

Well I'd ponder how easy X360 dev has been for those rushing for launch, but that's perhaps more a factor MS's management than hardware difficulty..

In short though, to a point there is of course merit in both approaches. Part of the reason Cell is so complex to dev for is a) there's a lot of parallelism in there, but on the flipside that's also what gives it its power and b) in order to get that level of parallelism/power they have to transfer more responsibility to the programmer (or "a" programmer, be it a middleware dev or straight down directly to the game programmer). So there is a bit of a power/ease tradeoff in that, certainly, and that's an appreciable approach.

edit - also, on jvd's point above, another 4 year cycle would IMO signal a failure on MS's part. MS would LOVE nice long cycles to make money off of, if they were as successful as they like. If X360 is satisfactorily successful, they won't cut it short.. I am very sure many publishers have also looked to clarify that point with MS.

 

[Guden Oden]

While the initial comparisons against the PS3 seemed to have died off, one prevailing thought seems to have remained, that being the 360 is just plain easy to use.

As a useless reminder to those out of the loop, this is in stark comparison to the comments about the PS3, touted as being a "nightmare" to adapt to.

I don't believe this to be accurate even on a fundamental level.

PS2 may have been nightmare-ish for devs to get a grip of considering they had to hand-write assembler programs for weird subprocessors, deal with an in-order level2-cacheless main CPU and whatnot, but PS3 has high-level development tools from day one. They may not be as evolved and integrated as MS's, but I think most would agree the situation is different compared to PS2 launch. Besides, PS3 is still 6+ months away, time for devtools to evolve as well.

And x360 plain easy to use? With six logical in-order CPUs sharing the same meager cache resources, getting even close to practical ( and much less theoretical) max out of a beast like that is going to be a tremendous challenge, requiring a fundamental change in how games are programmed and laid out. It's all well and good running the game engine on one CPU, physics on another and sound + misc stuff on a third, but how much are each of them getting used?

 

[PiNkY]

edit - also, on jvd's point above, another 4 year cycle would IMO signal a failure on MS's part. MS would LOVE nice long cycles to make money off of, if they were as successful as they like. If X360 is satisfactorily successful, they won't cut it short.. I am very sure many publishers have also looked to clarify that point with MS.

Microsoft has outcycled competitors with less resources in the os market, the office market, the browser market, media players, ... I would be rather suprised if xbox360's lifecycle turned out to be 6 years. Cycles will lengthen once MS has their marketshare, see oses, offices, browsers, media players, ...

 

[3roxor]

While the initial comparisons against the PS3 seemed to have died off, one prevailing thought seems to have remained, that being the 360 is just plain easy to use.

As a useless reminder to those out of the loop, this is in stark comparison to the comments about the PS3, touted as being a "nightmare" to adapt to

I think the difficulty with Ps3(Cell) is/was about adapting since it's different from what devs are used to. This doesn't nessesary mean the Cell is also harder to program for than..whatever.

That "nightmare" comment came from some PC dev if I remember correctly but I haven't read such comments from other devs.

 

[Lord Darkblade]

PS3 should not be that hard to program for, look at the responses from the Linux community, lots of them are throwing themselves into the simulator and actually getting to grips with the CPU early on. The PS3 has high level development tools, and at worst you can fall back on it essentially as a 3.2GHz in order processor, adapting to use the SPEs can be done later on or for experienced PS2 devs its probably a simple jump (VU0 and VU1).

Cell is a nightmare for PC devs, but then most console development is pretty scary for them as they can't rely on a new processor or card to fix problems, updates are also unlikely to be widespread (how often has an expansion patched the original game?) so it truly is a different prospect for a PC dev. Kojima and co seem to like the PS3, and they are the guys I would really be looking at, experienced CONSOLE devs.

 

[Megadrive1988]

well an easy to program for console will see better games quicker from a visual standpoint and cheaper games .

A difficult system will take along time to max out and be more expensive


SO it all depends on console generation. Ms is launching in 2005 , if they keep to thier cycle we will see an xbox 3 in 2009 giving this unit 4 years . If sony sticks with thier 6 year cycle being harder to program for might hurt them alot as they will be up against a new xbox and still have naother 2 years with thier ps3 . Or worse they will have invested alot of money for a platform that only lasted 4 years before being upgraded and never got fully used

I fully expect Xbox2 aka Xbox360 to go for 5 years (if not 6) before we see an Xbox3.
the Xenon CPU seems to have alot of headroom that developers can explore.
I think it would be bad if Microsoft launch the next-next gen Xbox in 2009, since Sony is unlikely to launch PS4 before 2012. by 2012, an Xbox3 that had come out in 2009 would be horribly outdated.

 

[drpepper]

I don't know how the development environment has anything to do with next gen graphics... but if you take history as a lesson, then really. Difficulty in programming paradigms can be overcome and has little impact to the succes of a console.

I think I agree with the posters before me, the nightmare comments that devs are making are devs that are coming from the PC realm. No one likes change, I know I don't, so when a dev is approached with a new problem of course they'll complain. But I haven't heard anything substantial coming from tried and true console developers.

From what I read they seem excited at the prospoect of more power, regardless of complexity.

Furthermore, a 4 year X360 product cycle would definitely indicate that it is another 4 billion dollar failure. A console company shouldn't have such a short life cycle.

My measly 2 cents...

 

[Synergy34]

Ok, so maybe this gen will be different but after seeing the avg dev costs for this gen, I don't buy the "our console is easier to develope for" comments. Everyone says the XBox is the easiest but then why did it cost devs about a million dollars more than the PS2 and Cube to makes their games?

 

[Shifty Geezer]

But I haven't heard anything substantial coming from tried and true console developers.

To be fair it's likely devs on PS3 aren't allowed to comment much on how they're finding it to develop for, due to NDA's. We haven't heard Deano or others here saying 'I can't tell you what we're trying to do, but it's a damned nightmare and I hate these SPE's' or 'I can't tell you what we're trying to do, but these SPE's are the best thing to happen to processors in the last 50 years'. (Or anything in between for you sissy moderates out there )

The only comments we've had that I know of have been in relation to the E3 demos in terms of 'we wrote this/ported this code in 2 months'. It'll be interesting to here what the Cell community (Cell developers running on the open IBM dev kit) reports in working with the system.

 

[scooby_dooby]

There's no way MS will doa 4 year lifecyle. In 4 years they'll be producing their box for like $50 and making money off the hardware. The whole design of X360 seems to be aimed at keeping costs low and reducing them quickly, the decision to cut the HDD as standard is all the proof you need that MS's ultimate goal is a $99 system.

 

[drpepper]

To be fair it's likely devs on PS3 aren't allowed to comment much on how they're finding it to develop for, due to NDA's.

Ah yes, very true. I forgot about that.

 

[Magnum PI]

The whole design of X360 seems to be aimed at keeping costs low and reducing them quickly.

It's a basical principle of console design.
Every other console manufacturer already do like this.

The remarkable fact is that it took MS one gen to figure it out.

 

[seismologist]

I guess the point is that challenging architectures can inspire creative ways of doing things? If that's your point then I sort of agree.

It's always been like that on consoles. Compared to the PC where the attitude is to make development as easy as posisble and if it doesn't run well, then people will upgrade.

 

[jvd]

I fully expect a new xbox in 4 years .

Its the easiest way to keep presure on sony . Just keep putting out systems and keep taking more market share . That doesn't mean they need to kill the x360 like they did to the xbox , but they can have the x360 along with the xbox 3 like sony will have the ps2 along side the ps3 .

Just put pressure on sony who have enjoyed 5-6 year generations . Make sony either speed it up or get left behind using ms's size and money to thier advantage .



Anyway as i said the longer the generation goes on the bigger the edge for the ps3 will have as its a more complicated system , the shorter it is the bigger the edge for the xbox 360.

 

[Carl B]

I kind of agree with JVD. Using DirectX as their base, it almost makes sense for MS to keep the pressure up on Sony and shorten the traditional console cycle. Ensure full B/C - which shouldn't be hard this time around - and you're good to go with a smooth transition. If the architectures are more orless the same, then it shouldn't be too hard either for MS devs to somewhat painlessly transition over to the new system. Thus MS is able to apply the latest in terms of PC--->console functionality as well as not be left completely behind on things like HD media playback; and in fact, perhaps jump ahead in some areas.

Obviously though the counter to this argument is if the console is worth it's weight as a console, it will be a needlessly aggravated 'loss-leading' cycle for MS as well. But it does make sense from a certain angle, depending on how this gen plays out financially.

Anyway I'm sure MS already has an internal plan on how long they expect 360 to last until their next console; we're just guessing at what that is and why.

 

[scooby_dooby]

I'm all for 4 year cycles, with continued support for 4-6 years on the previous console. Consoles really start showing their age after 4 years(it was almost exactly ~4years ago today that xbox1 was launched) so it's a good time IMO to upgrade.

 

[Shifty Geezer]

I'd actually think Sony might be in a better position to pull off a short generation. It takes two years to get to grips with the hardware, and four to really suss it out. Now if PS4 is just a multi-Cell PS3, code ought to be portable and devs can jump straight in. Whereas unless MS somehow develop a varient of XB360, or keep devs on a high-level API only, devs will need to get to grips with new hardware once again. Releasing a new hardware in 4 years is SERIOUSLY going to pee off the developers as a result and I doubt anyone would take the financial hit to develop software for a new system so shortly after the key profitability of the current gen.

All console developers should IMO wait for the current gen to be fully tapped before releasing a replacement, leaving time for a greater leap in technology and longer revenue generation period for both software companies and the hardware develops who'll be investing some of that in designing the new hardware.

 

[Synergy34]

4 years for the XBox but rememeber it didn't start selling well into it's life span. i would guess most people have only had it a good 2.5 years. Right now it's the hardcore gamers wanting an upgrade, but they never let the casual gamers settle in. I think they rushed to market and it really could bite them in the ass with the casual crowd.

Say you were a parent who isn't up on the gaming scene and find out less than a year after you bought one for your kid for X-Mas that a new version is coming out AND the company is shutting down support for the old one. You'd be pissed. Now your kid wants the upgrade and you to drop over 500 bucks for it.