Cheep-Cheep
Posted 06 June 2013 - 10:55 PM
Well the PS4 isn't using Cell, so I don't know what couldn't be touched.
people like to say sony lied about the systems 1.8gflops.... but that was not the gpu number.. that was the cpu. Which was hardly touchable by devs.
I have not been able to look at any dev machines for the newer stuff (with an exception) but it worries me with what they are saying
Thwomp
Posted 07 June 2013 - 03:29 AM
Let us be serious, everyone here is going to say that the Wii U isn't underpowered when the thing costs $299 on release and has a huge ass screen on the controller. You realise that doesn't leave much for power? I'm just saying, I'm fine with the Wii U being underpowered compared to the other next-gen consoles, and if you aren't, then you've bought the wrong system.
Hammer Bro.
Posted 07 June 2013 - 08:38 AM
people like to say sony lied about the systems 1.8gflops.... but that was not the gpu number.. that was the cpu. Which was hardly touchable by devs.
I have not been able to look at any dev machines for the newer stuff (with an exception) but it worries me with what they are saying
Cheep-Cheep
Posted 07 June 2013 - 09:43 AM
No. http://www.ps3rules.com/ps3-specs.htm It's the GPU. And its 1.8 TFLOPS.
they may have reported that, but the Cell CPU is rated for 1.8 gflops by IBM
Lakitu
Posted 07 June 2013 - 09:51 AM
people like to say sony lied about the systems 1.8gflops.... but that was not the gpu number.. that was the cpu. Which was hardly touchable by devs.
I have not been able to look at any dev machines for the newer stuff (with an exception) but it worries me with what they are saying
Cheep-Cheep
Posted 07 June 2013 - 10:15 AM
No, check the tape, they said rsx was capable of 1.8TFLOPS, not cell
Im not talking about what they said, im saying they ... mistakenly transfered the number from one part to another.
Lakitu
Posted 07 June 2013 - 10:34 AM
Im not talking about what they said, im saying they ... mistakenly transfered the number from one part to another.
No, they meant it. If tallied all together, the RSX had 1.8TFLOPS, but almost none of it was accessible, and it didn't have the resources to use much of it at any point in time in real world operation. That's the problem with theoretical maximums, they are hard targets to hit anyway, and Sony should have known better than to boast about RSX in that way, to that extent.
Hammer Bro.
Posted 07 June 2013 - 11:15 AM
Im not talking about what they said, im saying they ... mistakenly transfered the number from one part to another.
Hammer Bro.
Posted 07 June 2013 - 11:25 AM
they may have reported that, but the Cell CPU is rated for 1.8 gflops by IBM
Cell isn't a bad design. Just not super efficient.
Edited by Socalmuscle, 07 June 2013 - 11:26 AM.
Cheep-Cheep
Posted 07 June 2013 - 12:33 PM
There is no documents saying what you are saying. So until a source is quoted, which shouldn't be hard since Sony is open on specs, please show us.
I could be wrong, but everything I remember and have looked at says something different .
http://savannah.gate...EC08_Stef_2.pdf
shows the cell getting about 1.8gflops
im just assuming that they took that and applied it to their gpu and called it tflops.
even tho that cpu could easily do far more that was the rating of 1 spu. seems to much of a coinidence that they came out saying that same 1.8 abotut such a weak gpu that could never have came close.
the gpu from tests i have done i was able to squeeze around 170 gflops tops.
but aparently its been reported that it had alot locked away as it was rated for 400glops (top possible)
so innefficent and locked parts... either way 1.8 tflops was far off what was possible.
Edited by Cloud Windfoot Omega, 07 June 2013 - 12:46 PM.
Hammer Bro.
Posted 07 June 2013 - 01:53 PM
http://savannah.gate...EC08_Stef_2.pdf
shows the cell getting about 1.8gflops
im just assuming that they took that and applied it to their gpu and called it tflops.
even tho that cpu could easily do far more that was the rating of 1 spu. seems to much of a coinidence that they came out saying that same 1.8 abotut such a weak gpu that could never have came close.
the gpu from tests i have done i was able to squeeze around 170 gflops tops.
but aparently its been reported that it had alot locked away as it was rated for 400glops (top possible)
so innefficent and locked parts... either way 1.8 tflops was far off what was possible.
Cheep-Cheep
Posted 07 June 2013 - 02:16 PM
1.8GFLOPS isn't the same as 1.8 TFLOPS. Which is what Sony said the RSX and PS4 GPU can achieve.
http://playstation.a...csDetails_3.htm
I said it was a mistake not that they are the same. There is no way that both parts even at their maximums will come close to 1.8tflops together
Hammer Bro.
Posted 07 June 2013 - 03:30 PM
I said it was a mistake not that they are the same. There is no way that both parts even at their maximums will come close to 1.8tflops together
Cheep-Cheep
Posted 07 June 2013 - 03:54 PM
That's our point, but that isn't what Sony is saying.
my point is they probably mistook something and said it wrong.... its anyones guess.
either way, dont epect being able to use all of the PS4's resourses.
Hammer Bro.
Posted 07 June 2013 - 04:42 PM
my point is they probably mistook something and said it wrong.... its anyones guess.
either way, dont epect being able to use all of the PS4's resourses.
Hammer Bro.
Posted 07 June 2013 - 04:53 PM
my point is they probably mistook something and said it wrong.... its anyones guess.
either way, dont epect being able to use all of the PS4's resourses.
It wasn't a mistake, they were marketing the theoretical peak power of the RSX, which is 1.8TFLOPS, as real world operations. In practical results, though, the RSX didn't come close to those numbers at all.
Cheep-Cheep
Posted 07 June 2013 - 04:55 PM
It wasn't a mistake, they were marketing the theoretical peak power of the RSX, which is 1.8TFLOPS, as real world operations. In practical results, though, the RSX didn't come close to those numbers at all.
i do not know where they think they got those peaks.... seeing as the numbers ive found dont come close to that, and they are theory
Hammer Bro.
Posted 07 June 2013 - 06:54 PM
http://savannah.gate...EC08_Stef_2.pdf
shows the cell getting about 1.8gflops
im just assuming that they took that and applied it to their gpu and called it tflops.
even tho that cpu could easily do far more that was the rating of 1 spu. seems to much of a coinidence that they came out saying that same 1.8 abotut such a weak gpu that could never have came close.
the gpu from tests i have done i was able to squeeze around 170 gflops tops.
but aparently its been reported that it had alot locked away as it was rated for 400glops (top possible)
so innefficent and locked parts... either way 1.8 tflops was far off what was possible.
Cheep-Cheep
Posted 07 June 2013 - 08:26 PM
The cell as spec'd by IBM is decent. And THEORETICAL performance is interesting depending on how it's "calculated." The ps3 cell beats itself up. It's impossible for it to even think so lofty.
Even so, the cell was designed for 8 spe's. ps3 has one gimped due to early bad yield. So it's taken down quite a bit on the vector front as well.
ya cell is said to get 1.8 per spe.... so they say
Whomp
Posted 13 June 2013 - 01:55 AM
ya cell is said to get 1.8 per spe.... so they say
Edited by 3Dude, 13 June 2013 - 01:57 AM.
0 members, 0 guests, 0 anonymous users
