Home | Forums | What's new | Resources | |
What the PS2 can do |
tsumake - Oct 23, 2004 |
1 | 2 | 3 | ... | 6 | Next> |
antime | Oct 24, 2004 | |||
Google for "ps2 architecture". The SCEA... and SCEE... sites also have plenty of interesting presentations and papers. |
Des-ROW | Oct 24, 2004 | ||||
I would talk about the PlayStation2 architecture, the Emotion Engine and the Graphics Synthesizer for hours, but I will make it easier for you - Here... you have a very nice and technical article that compares how a standard PC works to how the PlayStation2 architecture functions. - and here... you have a technical overview of the PlayStation2's main processor, the wonderful Emotion Engine. I hope you get the answers to your questions. |
Cloud121 | Oct 24, 2004 | |||
*Reads on in comfusion, completely lost by what the articles are saying* :huh It's late, maybe I should read 'em later.... |
Alexvrb | Oct 25, 2004 | |||
It was never THAT impressive. Early PIIIs on the market. Jam a couple of those in a board and you're off. Sony knows what it's doing, but they're not process-defying gods. |
Des-ROW | Oct 25, 2004 | ||||
Oh, you are right, maybe you should mail the people at Toshiba who designed it, Jon Stokes, Ken Kutaragi and SCEI, and let them know how things should be done. Anyway - |
Alexvrb | Oct 25, 2004 | |||
Well there's always early Athlons to consider for solid FP as well as integer performance in that time period too. Either one creams the EE easily on integer performance, the PIIIs have MMX and SSE, the Athlons had MMX and an updated 3dnow!. The EE wouldn't even need the vector units so badly if the GPU was more advanced. What I was implying was that they are unable to break the laws of physics any more than the other guys, and they are just as bound by the chip production process. |
Dyne | Oct 25, 2004 | ||||
"sparkle effect"? i think you mean the lack of any AA in a lot of the early games (and some recent games as well). its not an "effect" and is pretty undesireable. |
Berty | Oct 25, 2004 | ||||
Those specs are shit. Remember how every apple is a supercomputer???? maybe toshiba use the same benchmarking program... Biasedmark 200X Also they were only benchmarking floating point operations, a benchmark made specically for RISC processors. I even would go so far as to say that they have used specific machine commands for the EE. The emotion engine was good, but it wasnt that good. Don't even start comparing RISC and CISC without showing ALL of the benchmarks. ... Back on topic, The emotion engine is a processor speciffically designed for 3d games. As such it has a RISC instructions set with hardware commands made specifically to deal with haze effects etc. This is the reason for the low overheads when doing this on a PS2. However to do this on a PC with a CISC chip takes a number of smaller hardware commands and has to be implemented via software. Think if it this way, If i was to write "one" using the english alphabet it would only take 3 moves, o-n-e. To find thos characters, i would have to only look through 25 symbols. This is RISC But if i was to write "one" using the chinese system then it would onlyl take one move but i would have to sort through 30,000 characters. This is CISC and how a PC works. Furthermore, i could streamline my risc process further towards only outputting numbers. So if i wanted to only display numbers, i would reduce my character set to 10 digits,... 0,1,2,3,4,5,6,7,8,9. So if i wanted to output "1" then i would only have to look through 10 characters and the process would only take 1 step. This is what sony did in essence to tailor the EE towards games, but this comes at a cost. If the EE wanted to do more complex equations outside of games then it would have to use a large combination of its characters where as a CISC chip, even though it takes longer to find the character may have just one that does the same thing. |
ExCyber | Oct 25, 2004 | ||||
The first graph is not even a benchmark at all, it is just a naive comparison of the major theoretical numbers. It completely ignores things like cache performance, random read latency, and the need to perform both processing and I/O. IOW these numbers mean next to nothing as an overall performance measure. The second graph is, as you hint at, illustrating operations that are highly implementation-dependent, and which cannot be usefully evaluated without access to the code, compiler(s), and build scripts. They look very impressive, but that's about all they accomplish. :beerchug |
yasminkov | Oct 27, 2004 | |||
[looks blankly at all the technical jargon and faints] I wont even pretend I understood all of that but heres a thought, why is the system still running when the graphics it outputs often aren't as good as many of the DC's games 3 years ago. Also there seem to be obvious faults with it when you look at say Soul Caliber 2, across systems PS2 had by far the worse graphics and sound. In a bored moment i was playing the original SC on the DC and found that the graphics and sound we better than on the sequal on a better machine. Has the PS2's graphics system been used to the fullest or are developers not bothering to as they think it will sell anyway? |
Des-ROW | Oct 28, 2004 | |||||||
No.
Neither. |