Gummy disks present new problem for Halo 3

Status
Not open for further replies.
TheEnd, as to your question on 360 memory speed, the bandwidth on the 360 between memory and processor (both GPU and CPU) is VERY wide, actually wider than the PS3. The 360 does not have problems with bottle necking. There was fear that the PS3 would have this problem, because the bandwidth was so narrow in comparison with what the the cell could do, but some ingenuity on the part of Sony allowed the cell processor to do many of the FP calcs and on the fly calcs in the cell processor itself without a memory dump. This info was also able to be verticed without sending to the GPU for display. The 360 also does this, to a point. However, as in what I've listed above, these similarities stop right there and that is where the GPU on the 360 cuts above the Cell, because the architecture is better thought out and more powerful. Now, Sony has something up there sleeves for this, though. The original idea was for the Cell to be both the CPU and GPU, ****... it's powerful enough to do just that. Problem with this is that development is hard enough without now using all those SPE's and single genearl purpose core to now do graphics, and was just not realistic so they got NVidia to make the PS3 card later in the creation of the PS3. This is a good thing, and a bad thing. A good thing in that programmers will have an easier time now, but a bad thing in that the card would not prove near as powerful as their rival and so to match this they would have to fall back somewhat on using both the cell AND the GPU in tandem to make up the power difference. This is bad news, because it is VERY hard to do this. That is why I say the PS3 has SO MUCH POSSIBILITY!!! However, the software development industry has to catch up and think new rules in the game of programming, and by the time this happens I have the feeling it will be around the time of new console arrivals, and possibly a new king on the block.


To sum it up, great system, true. Not practical in important ways, and much of what they are doing is to little to late.


Now, onto memory specs. Here is some info on how the core and other parts work... part II

Inside the Xbox 360, Part II: the Xenon CPU: Page 1

Like I said before, I'm not anti Sony, I just think they tried to much at once, and it will turn and bite them on the ***. The cell is very powerful, but to match the GPU capabillites of the 360 the cell needs to be used in combination with the PS3's GPU, which was added to remove the need of using the cell for graphics use in the first place due to difficulties. This makes the Cell not really built for gaming, but scientific use. Now, if Sony had waited for a more robust and powerful GPU, God this thing would of been unstoppable but with MS 1 year ahead already and Nintendo releasing soon they didn't have time to spare.
 
just on a side note, stop flaming theend when he is actually having a real conversation, lol, cause it is actually interesting and i went to the effort of reading all the posts, that and killians... you know too much, haha

that and, don't flame me, real question here, i heard that the cells have a fairly substantial fail rate, and programmers only make their games to use 3 or 4 of them incase the others screw up? sounds like a load of crap to me, haha, but is there any truth in this? and yes i know, xbox has high fail rate, lol, no need to remind me, haha, but genuine question here, any idea on the fail rate for the cells?
 
well i haveent been flaming Killians, last time i checked asking someone the memory specs on the 360 isn't flaming.

As for the Cell failure rate that is false, the cell's dont fail in the ps3, instead back when they were first manufactured, there were a lot of cells that werent usable in the ps3 since they didnt meet quality control levels, they were diffucult to make since it was not the same process as making a Conroe, but now the production line and engineeres have rapmed u production and now the amount that usable cell are produced is very high, I also read about that and found out it was most likely M$ FUD but FUD nonetheless since they tried to make it out to say that the PS3 itself was failing, but that wasnt the case it was the producton line not having many quality samples but that is fixed now.

As for the SPU's in the Cells there is 8 total, one is used for secrity so in reality there is 7 usable SPU's, i wont get into detail but, programming for the cell is really hard since you basicaly have to write it for 7 threads if you want to use it effectively , however since it is not the same thing as writing HT apps, it is increasingly difficult to use all SPU's effectively rather than use 1 or 2 *cough* EA, but now developer tools are getting better and more user firendly, more devs are starting to get used to the complex cell programming and it is showing with upcoming games like, Haze, Killzone,MGS4,FFXIII, includig the ports such as UT3. and thx to the dev tools sony has been giving out including libraries and programming tips and tools studios and devs are starting to use them more effectively to make better ps3 games and make ports on par with the 360, but yeah it will be a while before devs can utilize the SPU's to their full potential.
 
ah kk, and yeah i didnt mean you theend, i meant the other guy about the flaming thing, for once! haha

you guys both know too much, lol
 
You know, I couldn't say on the PS3 cell failure rate. I mean, it is possible and there are enough SPE's to handle the load, but you would need hardware that more than likely is not available to the general public to find this out, that or some third party or individual person write a program to test the reliability of the cell structure. Easiest way I can think of to do this is writing an algorithm that creates something akin to the memory checker, except it will check the integrity after swapping from memory and back to each SPE individually. Multiple passes in each address register and check to see if any discrepancies are apparent. This would be SO time consuming, though, because, as you are probably aware, a failing chip may not constantly give up erroneous data, so it's a matter of how often the data passed to and from is corrupt compared to the total number of passes.
 
Status
Not open for further replies.
Back
Top Bottom