Adinnieken said:
FLOP = Floating-point Operations Per Second. It's math. The ability to do 2.56TFlops of calculations by a processor has nothing to do with what it is able to do in terms of actual rendering. The act of pulling graphical data in, performing computational changes and pushing graphical data out. Not saying graphics don't involve lots of math but the calulations are internal to the processor with a small amount of data going to the memory for temporary storage. There are bottlenecks in computers. There always have been. GPUs are no different. Raw power has NEVER been the be-all-end-all determination of how well a system performs because at the very end of the day it's how that power gets used. The PS3 was a perfect example of this. Raw power wise the PS3, other than the GPU, was leaps and bounds ahead of the Xbox 360's Xenon processor. There is absolutely no debate about this. However, the fact that the Xbox 360 had unified memory, as well as the EDRAM meant the console could perform just as well as the PS3 if not at times better. Back in the day (I don't know if it still exists) there was a great discussion regarding the PS3 GPU and Xbox 360 GPU and initially the prevailing thought was based on the raw numbers that the PS3 would have a material advantage. Then someone came along and actually compared the two systems based on the amount of work the two systems would be able to do. Taking into account the memory bandwidth, the CPU, the performance of the GPU including the shaders. The assessment was that the PS3 would barely be able to do 1xAA if not 2xAA, and 4xAA would be out of the question. With the Xbox 360 the determination was it would be hard pressed to do 4xAA but actually could do 2x(2xAA) more efficiently. And the results were evident early on. PS3 games tended to be jaggy until developers were able to do MSAA on the PS3. And even still, MSAA isn't as good as AA but it required much less computation on the GPU. So, games on the PS3 tended to be less sharp than the Xbox 360 games. So to your question and point. Yes, I am saying the memory bus is a bottleneck. If Microsoft is saying for performance reasons it is a bottle neck, than it would be a bottleneck for Sony as well. Just not necessarily at the same point. Sony's recommendation (not requirement) for a balanced approach was to use 14 CUs for graphical rendering and 4 CUs for GPGPU operations. Sony hasn't denied this. The only thing they have said is it isn't a requirement, that developers can approach things however they want. But my guess is that's why you're seeing Kill Zone drop in frame rate at times on the PS4. Unless you have a plausable reason why KZ drops down to 30FPS? |
Killzone is unfinished and until the final product is released we won't know. GG claim 1080p 30FPS SP and 1080P 60FPS MP. They obviously want the eye candy in the single player and have went with 30FPS probably due to time constraints. I would assume it would probably easier to get SP to 60FPS than MP.
PS3 and X360 are wildely different architectures and can't really be compared. PS3 had the much ore powerful CPU but the X360 had the better GPU and unified ram and was way way easier to develop for.
PS4 and X1 are very comparable as they share the same CPU and GPU family. PS4 has the better RAM set up. How? Its faster with a higher bandwidth and its one unified pool.
Mark Cerny:
"We received very positive responses to the concept of 'Supercharged PC Architecture' from developers because we made a drastic improvement to performance and made it easy to develop games at the same time," said Cerney, who developed many games. "It is hardware for which games can be developed really easily."
As for the "supercharged" parts, or the parts that SCE extended, he said, "There are many, but four of them are representative." They are (1) a structure that realizes high-speed data transmission between the CPU and GPU, (2) a structure that reduces the number of times that data is written back from the cache memory in the GPU, (3) a structure that enables to set priorities in multiple layers in regard to arithmetic and graphics processing and (4) a function to make the CPU take over the preprocessing to be conducted by the GPU.
The GPU in the PS4 is more powerful in every way! So they simply haven't just increased the CU count but every part of the GPU. The Bandwidth of the GDDR5 is way more than the GPU in the PS4 will ever saturate. Just cause MS claim something doesn't make it so or right!
X1 GPU:
1.18 TF GPU (12 CUs) for games
768 Shaders
48 Texture units
16 ROPS
2 ACE/ 16 queues
PS4 GPU:
1.84TF GPU ( 18 CUs) for games + 56%
1152 Shaders +50%
72 Texture units +50%
32 ROPS + 100%
8 ACE/64 queues +400%
