Digital Foundry Article Technical Discussion Archive [2015]

Status
Not open for further replies.
I would say distributed computing, ie making cloud and online a requirement, would change gaming as we know it. That's the only way they can get enough computing power into these small low wattage affordable boxes to be able to try new things and move forward with new broader ideas beyond the scope of what a single box can ever hope to do.

Making the Cloud a requirement would also make a console's future a bit more at risk to DDOS and other more hackerish obstacles. I agree that there is a future for a cloud based game solution that takes advantage of a server's abilities but as we have recently seen you might need a new 'Internet" to go along with it. It is one thing to have authentication services pushed around a network in response to say a DDOS but it will be another problem entirely when every frame of a game may be impacted by some more fine grained tomfoolery by hackers. I would assume all data between a gaming console client and it's host server will be encrypted ( adding some performance issues at say 60fps maybe ) so some basic filtering can be done along with a secure connection but having very time dependent data open to some extent to the whims of an open internet environment may prove more problematic that at first blush.

Streaming gaming services would also be affected but steaming is a known issue. Having various "intimate" parts of a game ( where and when polygons are placed in a scene as a fer instance ) could have some unforeseen pitfalls.
 
Interesting is the fact that the good utilization of PS3's cell yields results that rival those of PS4's and XO's CPUs.
 

Glad to see MS is continuing to section off the performance bleeding aspects of the Kinect. Wonder if someone has pulled together the data to say how much die space was ultimately set aside for the kinect ? Not saying that at the time it was the WRONG thing to do but it is kind of fascinating to see how design choices were adjusted based on the potential windfall of 'winning' the living room with the all encompassing ONE strategy. The fraction given to kinect is of course just that ... fractions at this point although the 'what-ifs' do start whirling around. At the end the PS4 async compute bet may not pan out either or HSA in general for both consoles, although I personally prefer that kind of bet rather than a multifaceted marketing one.
 
Interesting is the fact that the good utilization of PS3's cell yields results that rival those of PS4's and XO's CPUs.

Wasn't 2015 supposed to be the year that the full power of the Cell would be unleashed so we may not fully be aware until later this year :LOL: 3.2 ghz and an efficient way to connect multiple processing cores is still a very nice thing to have even if your tech isn't the newest thing in the world. I don't know if we will ever see such a crazy beast again but I am glad we did.
 
Wasn't 2015 supposed to be the year that the full power of the Cell would be unleashed so we may not fully be aware until later this year :LOL: 3.2 ghz and an efficient way to connect multiple processing cores is still a very nice thing to have even if your tech isn't the newest thing in the world. I don't know if we will ever see such a crazy beast again but I am glad we did.
_1413434492.jpg

I know what you mean, cell was a beast, way in advance of its time
 
Interesting is the fact that the good utilization of PS3's cell yields results that rival those of PS4's and XO's CPUs.
Cell (7 SPEs + one PPE) at 3.2 GHz actually has twice the theoretical peak FLOPS of a 8 core 1.6 GHz Jaguar. I am suprised that it doesn't beat the Jaguar in this particular scenario since cloth simulation (heavy linear vector crunching) is a pretty much a perfect use case for SPUs. This result also points out the fact that Cell loses very badly in everything more complex (branches, memory pointer chasing). 600 cycle memory latency and no branch prediction makes complex things hard. Jaguar on the other hand has nice vector units with super low latency, allowing you to mix vector math with other code easily. Jaguar can be easily beat Cell by 5x or more in more complex code.
 
They're not freeing up a core by disabling voice commands. They're freeing up half to 80 percent of a core by disabling NUI, which is the skeletal tracking, IR feed, depth feed, custom voice commands. Those are all now opt-out features with some restrictions, like not being allowed to disable them on menu screens.
 
Last edited:
They're not freeing up a core by disabling voice chat. They're freeing up half to 80 percent of a core by disabling NUI, which is the skeletal tracking, IR feed, depth feed, custom voice commands. Those are all now opt-out features with some restrictions, like not being allowed to disable them on menu screens.
I didn't say anything about Voice Chat.
 
So DFs findings are a game 9-12 months from release has framerate and resolution issues. Absolutely nothing else of substance in that article. That info might be a little more interesting if it were 3 months from release.
 
Well, you hope 343i is aware of the framerate hiccoughs at least, and perhaps the animation issue in that pre-battle sprint thing.
 
Well, my previous expectations, considering the limited jump in hw power and the additional resource requirements of 720p+ resolution and double framerate, were that 343 is probably mostly going to just scale up the core of the H4 engine; removing limitations like disappearing shadows and corpses and such and scaling up geometry and texture detail.

The actual demo suggests more advanced lighting, more advanced materials, more complex post processing (which includes SSAO) and some small extras. So far so good, IMHO.

The really interesting question is of course that old rumor about making the game more open-world like. That would require some really serious engine architecture work, but it'd probably not effect the renderer features that much and thus it wouldn't be visible in multiplayer maps...
 
I still consider AO to be the work of the... well, the devil, for anything better, so I don't care how it's implemented, it's still wrong :)
 
Technically there's no such thing as 'occlusion' and your lighting should handle this stuff on its own ;)
But I do understand the need for hacks and approximations, I just don't bother learning more about them - eventually they'll be able to do it right... right? :)
 
Status
Not open for further replies.
Back
Top