Antialiasing... Do you think PD will do it?

  • Thread starter Thread starter Dracwolley
  • 104 comments
  • 8,177 views
EDIT: Forget it, thread is moving faster than I care to keep up with :p

The thing is a lot of current software was not initially developed with multi cores in mind, more the software has been made to adapt via patches etc..
When multicore becomes standard is when we will see close to 1:1 scaling.
 
Interesting maybe, but also no offence, irrelevant to current console technology. My point is that PS3 was not as focused to be a games machine like microsoft console hardware and software has been. I dont see any defence for that which leads on to the second point:

Predicting the future accurately is still impossible last time i checked, and gambling on something still being with the trend a decade down the line is once again not the best way to build a new console.

Did you not read the last part where it mentions consoles and the direction in which the industry is headed?

Edit: Damn double post:dunce:

Edit: You are also forgetting that PS4, XBOX 720 are most likely already in development, so whatever is the trend now will most likely be influential on the final product.
 
Did you not read the last part where it mentions consoles and the direction in which the industry is headed?

I am aware the direction we are headed. Again not my point. You are saying sony have got this right for the NEXT generation of consoles and not defending the fact that for this generation, PS3 has not been built focused.

Then im saying that no one can actually say what the best solution will be for the next generation in three years time. Not me, not you, and especially not sony when they designed CELL more than 10 years before next gen will appear!!!

Thus saying that CELL will pay off next generation isnt very useful to the here and now, and when the industry shuffles on a little bit more it may not actully be as useful as thought either. Who knows. Much like the billions of dollars poured into netburst the best answer to the current problems was to throw it in the bin and start again, because you then have options wide open.
 
You are also forgetting that PS4, XBOX 720 are most likely already in development, so whatever is the trend now will most likely be influential on the final product. Also 10 years is the life cycle PS3 will be supported, i don't think it will be ten years before PS4 is released.
 
You are also forgetting that PS4, XBOX 720 are most likely already in development, so whatever is the trend now will most likely be influential on the final product. Also 10 years is the life cycle PS3 will be supported, i don't think it will be ten years before PS4 is released.

Ahhh but indeed im not forgetting it at all. You wanna talk about the here and now then thats great. Im very aware that even presently, CELL is still not the answer to the question of whats the best design for gaming!! :) The architecture wasnt suited in 2005, and its not suited 5 years later.

The answer NOW is still a massively parallel GPU and a multi core, general purpose CPU. If you are designing NOW these are the answers still. Will it change for the future in say, 2015? yes i would expect so!! but for PC. Too late for consoles that as you say are being designed as we speak...

Either way it just sidesteps my point about ps3 is not focused like 360 for current games
 
The most important question is not how many cores you are running but for how many you have code for. If someone just comes up with converter to spread processing load efficiently without needing human input.(i mean higher intelligent programming language)
 
The most important question is not how many cores you are running but for how many you have code for. If someone just comes up with converter to spread processing load efficiently without needing human input.(i mean higher intelligent programming language)

Well the main complaint with developers is that CELL has to be coded for by hand, specifically to the minutest detail to get good performance. In laymans terms its like a child, you have to tell it exactly where you want it to be and exactly what you want it to be doing at any given time. A pain in the fricking ass essentially, compared to a PC or even 360s CPU, where you can just shove whatever you want into a compiler and sit back eating your pop tarts :dopey:

In a way the same goes for RSX- because it is split pipe if you dont load balance vertex and pixel yourself exactamundo, it'll run at a crawl. Whereas xenos being unified you just give it a list of stuff you want doing and it will sort itself out. A crude and basic explanation really but its not far from the truth.

This just emphasises the deal with 360 and ps3. One is designed to be friendly to games developers. Cheaper, and quicker for less programmers to knock stuff up and still get good results. Its what you would expect from a software company really isnt it? The other has been designed as a hardware engineers wet dream. Theoretical this, maximum that. Just that when he wakes up to the real world, it doesnt quite work as he had hoped!
 
Last edited:
Well the main complaint with developers is that CELL has to be coded for by hand, specifically to the minutest detail to get good performance. In laymans terms its like a child, you have to tell it exactly where you want it to be and exactly what you want it to be doing at any given time. A pain in the fricking ass essentially, compared to a PC or even 360s CPU, where you can just shove whatever you want into a compiler and sit back eating your pop tarts :dopey:

In a way the same goes for RSX- because it is split pipe if you dont load balance vertex and pixel yourself exactamundo, it'll run at a crawl. Whereas xenos being unified you just give it a list of stuff you want doing and it will sort itself out. A crude and basic explanation really but its not far from the truth

Doesn't unified shaders lend itself better to multicore processing ala latest GPU's from Nvidia and ATI
 
Doesn't unified shaders lend itself better to multicore processing ala latest GPU's from Nvidia and ATI

Sorta. But CPU multi thread optimisations usually happen within the operating system and API, the software itself. Read the bit in your own post where it says patches have been issued to make PC games more multi thread friendly.
 
Sorta. But CPU multi thread optimisations usually happen within the operating system and API, the software itself. Read the bit in your own post where it says patches have been issued to make PC games more multi thread friendly.

Yeah, but that is because most software is not written with multicore being the standard, they usually add support later on, which isn't very efficient.
Surely if your developing a game soley for PS3 you will use multicore as a basis for your code. I can however see the problem with ports.
 
Yeah, but that is because most software is not written with multicore being the standard, they usually add support later on, which isn't very efficient.
Surely if your developing a game soley for PS3 you will use multicore as a basis for your code. I can however see the problem with ports.

Multicore development is standard these days for games. Console, Pc, even phones have multicore processors. The challenge is always within software to utilise them. The problem with ports forgetting the memory issues on ps3 is when you cross from Pc to 360 or vice versa, you can use a compiler to massively reduce the amount of work you have to do. With Ps3 if you dont get your hands dirty you end up with junk performance.

Developing from the ground up for ps3 is the ideal situation, but then that goes back to market conditions and publisher expectations. Something like GT5 looks the way it does because it is intricately designed to run on Ps3 and Ps3 alone. Besides the fact it has had so much time, with a massive technical budget. Internally sony throw every resource they have at it much like killzone 2, the reputation of the machine almost depends on it. 360 being cheaper, easier, faster and having a bigger install base means more games are going to be built for it. Ones that arent still typically run as good, because developers havent yet got the handle on ps3.

Ergo round in circles back to the idea about market conditions and you have a catch 22 for ps3 development that developers wont get the handle on it, unless they prioritise it over and above anything else. Throw money and people at it.
 
Last edited:
Multicore development is standard these days for games. Console, Pc, even phones have multicore processors. The challenge is always within software to utilise them. The problem with ports forgetting the memory issues on ps3 is when you cross from Pc to 360 or vice versa, you can use a compiler to massively reduce the amount of work you have to do. With Ps3 if you dont get your hands dirty you end up with junk performance.

Developing from the ground up for ps3 is the ideal situation, but then that goes back to market conditions and publisher expectations. Something like GT5 looks the way it does because it is intricately designed to run on Ps3 and Ps3 alone. 360 being cheaper, easier, faster and having a bigger install base means more games are going to be built for it. Ones that arent still typically run as good, because developers havent yet got the handle on ps3.

Ergo round in circles back to the idea about market conditions and you have a catch 22 for ps3 development that developers wont get the handle on it, unless they prioritise it over and above anything else. Throw money and people at it.


Well i think i'm going to end it here, not to mention we have taken this way off topic:lol:, but it's being an interesting exchange of thoughts:D
 
This is exactly why a machine with three cores and a custom GPU somehow manages to outperform a machine with 9 cores and a bunch unusable by the dev.
This is a badly stated remark. Plus it's an opinion.

The reason that some people think this is because multiplatform games coded for the 360 first are harder to port to the PS3 because the architecture is so different. Plus, optimizations require a bunch of rewriting, which is additional cost many developers just don't want to mess with. A few games in the past ported poorly to the PS3, but even in the past, a developer could get similar results between both machines with just a bit more effort, which they began to invest when they realized their games sold like poo if they didn't. But this year, multiplats are almost uniformly the same game, which might have a few graphical issues on either machine, and if the game is coded for the PS3 first, results are much better. Plus, there's the added benefit that the game tends to look comparable to PS3 exclusives rather than just another multiplat. This argument is now obsolete.
 
I've seen numerous threads about these differences on Playstation and game boards, and when screengrabs are made, the differences are almost non-existent these days. Like one might have fuzzier textures on the 360, or slightly less AA on PS3. Or the PS3 version might be better in just about every way, but you have to look closely. But multiplats look lame compared to PS3 exclusives. If Prologue was a multiplat game, it would probably look like FM2. :p
 
Back