bigtakilla said:
|
I also think there should be way more to nexgen games than frame rate and resolution. Stuff like that really doesn't bother me.
The PS4's power seems... | |||
Better, but not THAT much better anymore... | 241 | 15.42% | |
Are you crazy?! The PS4 is GOD! | 349 | 22.33% | |
The Wii U is clearly unde... | 741 | 47.41% | |
The PS4 is selling better... | 36 | 2.30% | |
I think I'll be buying a... | 191 | 12.22% | |
Total: | 1,558 |
bigtakilla said:
|
I also think there should be way more to nexgen games than frame rate and resolution. Stuff like that really doesn't bother me.
T.Rexington said:
I also think there should be way more to nexgen games than frame rate and resolution. Stuff like that really doesn't bother me. |
bf4 is probably the most nextgen game out its makes the curren gen play and look like crap in comparison
starworld said:
bf4 is probably the most nextgen game out its makes the curren gen play and look like crap in comparison |
There's definitely been a boost in characters on screen and players in the same game.
curl-6 said:
-Marcan clearly noted Espresso should "win big on IPC on most code." -So in other words, you code in a way that benefits cache function, so in effect you are managing cache indirectly. -You were still wrong about it being symmetrical. |
Yes but he also expressed concerns about the ibm espresso's SIMD engine. Just because it can win in most codes doesn't mean that it will have more potential considering that it's capped at 15 Gflops. Where it won't win is in hpc performance which is also a somewhat relevant metric too considering the fact that games are performance critical.
It's true that you can fix your code to suit smaller caches and such but never does the programmer handle the cache.
Big woop ... Like that'll make a huge difference considering all of it's other weaknesses.
JoeTheBro said:
Oh I'll still talk with ZyroXZ2, but I'm done with this discussion with him. I like debating these things, not arguing them. With lines like "I said I'd stop, and yet here I am kicking you while you're down. I will stop from this point forward, regardless of your response, for your sake"he shows he's not expressing his points, but rather is making it a personal battle lacking any commitment to the truth. You're far from innocent yourself, but at least I haven't noticed any childish remarks like this in your comments. |
It may be true that I gave him a terrible attitude while being reluctant to explain anything but it's probably for the better since he's far from capable of interpreting the mechanics of performance. Well atleast I don't go around making conspiracy theory claims.
fatslob-:O said: Yes but he also expressed concerns about the ibm espresso's SIMD engine. Just because it can win in most codes doesn't mean that it will have more potential considering that it's capped at 15 Gflops. Where it won't win is in hpc performance which is also a somewhat relevant metric too considering the fact that games are performance critical. It's true that you can fix your code to suit smaller caches and such but never does the programmer handle the cache. Big woop ... Like that'll make a huge difference considering all of it's other weaknesses. |
-Did Marcan ever confirm 15 Gflops? And besides, that's why it has a GPGPU.
-That's my point, the way you program still effects how well the caches function. To get good performance out of Espresso you'll need to code for its larger cache
-But it does make it different to working with Gekko or Broadway, not to mention Cell and Xenon, which the multiplatform engines are based on.
curl-6 said:
-Did Marcan ever confirm 15 Gflops? And besides, that's why it has a GPGPU. Do I have to when marcan himself just said that it was just three overclocked ibm broadways ? If we scale the numbers up from broadway we get 15 Gflops for espresso. -That's my point, the way you program still effects how well the caches function. To get good performance out of Espresso you'll need to code for its larger cache However, you DON'T program or manage the cache itself. ALL OF IT IS DONE BY THE CPU ITSELF. The point of cache is to get SUSTAINED performance. It is not some magical part that can increase theorectical performance. -But it does make it different to working with Gekko or Broadway, not to mention Cell and Xenon, which the multiplatform engines are based on. Your grasping at straws here ... Just because one core has 1.5mb more L2 cache than the other cores doesn't mean that it will largely affect the functionality of the other cores. One core having less cache misses just means that it will stall less than the other cores. Once again the programmer doesn't ever touch the cache, all of it is done in the background of the application so little can be done about it. |
starworld said:
|
Wii U has GPGPU compute shaders and DX10/11 equivalent features which PS3/360 don't have.
Also, Xbox didn't arrive 6 years into PS2's lifespan after all engines were tailored intimately to PS2.
fatslob-:O said:
|
- That's only accounting for core count and clock speed, not the 12-fold cache increase which will boost performance.
- But you have to manage your code so that the cache can work well with it.
- Assymetrical cache amounts means the "main" core has to be prioritised, as Shin'en did with Nano Assault Neo, running essentially the whole game on that one core.
How the hell does this thread manage to keep going based on a couple of vids of two unreleased games?!