Predict: The Next Generation Console Tech
- Thread starter Acert93
- Start date
- Status
But again, it's not going to matter! Unless Patsu's mom is a 23rd century cyborg with nanosecond sensitivity, she nor anyone else will notice if there's an extra few nanoseconds of delay between connecting her display via copper wires or via an optical cable. 
Only if those extra ns are used frequently, with loads and loads of random read/write requests as opposed to sustained transfer, will the difference between copper and optical connects have a tangible impact, and that'll only happen in some bizarro situation where software and hardware is poorly designed, such as sticking the system RAM in an external box. Putting it another way, sticking low latency DRAM in your PC doesn't create a massive improvement in performance. It'll help eek out an extra few percent frames per second maybe, but no developer is going to sweat over different DRAM timings in different boxes. And that's the most commonly incurred latency for processing, RAM latency every time you want some new data. Prefetching and caching solves a lot of that. OoO processing helps, using those few cycles of waiting on data to do something else. So if latency on RAM isn't making that much difference, tiny latencies on a peripheral will be far, far less of a concern.
There's no question that the difference in latency between optical and copper cables for electronic devices is going to add no problems at all to using devices. No next-gen console designer is going to have to fret about whether to go with costly optical versus cheaper copper ports on account of copper being slower. The concerns over incorporating these new standards will be entirely cost and relevance - is anyone actually going to want them? USB3 will be good enough for everything a console would want, surely. Unless MS or someone wants 1080p, uncompressed, stereoscopic depth cameras!
Only if those extra ns are used frequently, with loads and loads of random read/write requests as opposed to sustained transfer, will the difference between copper and optical connects have a tangible impact, and that'll only happen in some bizarro situation where software and hardware is poorly designed, such as sticking the system RAM in an external box. Putting it another way, sticking low latency DRAM in your PC doesn't create a massive improvement in performance. It'll help eek out an extra few percent frames per second maybe, but no developer is going to sweat over different DRAM timings in different boxes. And that's the most commonly incurred latency for processing, RAM latency every time you want some new data. Prefetching and caching solves a lot of that. OoO processing helps, using those few cycles of waiting on data to do something else. So if latency on RAM isn't making that much difference, tiny latencies on a peripheral will be far, far less of a concern.There's no question that the difference in latency between optical and copper cables for electronic devices is going to add no problems at all to using devices. No next-gen console designer is going to have to fret about whether to go with costly optical versus cheaper copper ports on account of copper being slower. The concerns over incorporating these new standards will be entirely cost and relevance - is anyone actually going to want them? USB3 will be good enough for everything a console would want, surely. Unless MS or someone wants 1080p, uncompressed, stereoscopic depth cameras!
Blazkowicz
Legend
umm I was talking about a "multiseat computer" idea, i.e. single computer, multiple users each with their keyb/mouse/display etc., each running a desktop, apps or full screen games with focus ; and just saying that microsoft software supports that as of now. (with 2D only USB peripherals on the hardware side, for now)
maybe I was unclear then sorry for that.
yes as shifty geezer say that kind of stuff will be useful on general purpose computers, even then it would be niche stuff. not much use on a console where all computing hardware is soldered in.
Every Thunderbolt copper cable has built-in circuitry to remove noise. That's why it is expensive. I don't know how much latency the cable firmware and chip add on top of the physical media.
As for your RAM example, prefetching and caching does solve some of the issues but not all. Essentially, sometimes software needs to be changed to workaround the problem. Depending on the actual latency, it may not be transparent for existing apps anymore.
There are h/w and s/w protocols running over the cable, their overhead would be higher. This may/should overshadow the raw physical media performance but it's interesting to see how far optical can go anyway. It will depends on what and how they use it.
From the Vaio Z perspective, adding a more powerful GPU makes a positive difference overall. But from a distributed computing perspective, knowing the latency will allow developers to tune their algorithm better. The smaller the latency, the more "general" the design.
Exactly, and when you choose cost over performance, you should be willing to accept the overhead and performance hit anyway.
The thing I didn't really understand was why was multiseat-computing brought up in the high-speed connections discussion. Pretty much only thing you might need high throughput there is for the video signal and HDMI is good enough for it.
Now if someone had proposed chaining together resources from multiple PCs (RAM, GPU, CPU, ...) over that line then that would have been a whole different thing that is in no way related to multiseat computing. Also, you could definitely not compare latencies there with the latency you get with onlive-like usages.
Now if someone had proposed chaining together resources from multiple PCs (RAM, GPU, CPU, ...) over that line then that would have been a whole different thing that is in no way related to multiseat computing. Also, you could definitely not compare latencies there with the latency you get with onlive-like usages.
Can we keep the discussion relevant to the next-gen consoles please.
But what has that got to do with including Thunderbolt/Lightpeak in next-gen consoles?
Unless you're suggesting someone's going to roll out a hardware platform that uses ultra-tight local-networked distributed processing where these tiny latencies matter, my point was always that whatever latencies are present, the difference between optical and cable will be irrelevant for next-gen consoles (and further irrelevant for almost all applications of these high bandwidth ports, but I don't want to stray too far). True or false?But what has that got to do with including Thunderbolt/Lightpeak in next-gen consoles?
If LightPeak is just another faster USB 3.0, then it may not make a big difference to the industry. But if it enables new form of PC/console config, then it may have a better chance. I have no preference either way yet. Once the differences/characteristics between an optical vs copper LightPeak implementation is known, then it's easier to talk about console influence and specific design.
At this point, Sony already broke a few LightPeak and USB 3 rules in the Vaio Z implementation. While it is compatible with both standards, it may not be able to market the link as Thunderbolt or USB 3. I am just curious what else is different about their implementation.
With a cheap enough LightPeak, we may be able to see assorted hardware that are detachable and can scale upwards relatively easily. With the proliferation of gaming devices, going after the crowded mass market is one way to succeed. At the same time, it is also possible to roll a uniform, specialized, and closed-box design that can scale upwards to deliver a very unique experience. I'm an intrigued observer in the mean time.
What real-world difference will there be in terms of real use connecting a PSVita2 to a PS4 over a Lightpeak or Thunderbolt connection? Why would Sony choose optical over copper because of latency? Unless the console is directly accessing the portable's RAM instead of the portable working as a node on a processing cluster, a few extra ns latency should mean nothing.
May be Vita 2 or even a laptop can use PS4's GPU, RAM, Blu-ray, HDD, TV/monitor all at the same time, just like Vaio Z ? It would work for my office PS3 setup. Would be interesting to see if we can swap Z's optical cable out for a copper one. When Sony broke the USB 3 and Thunderbolt rules, it seems that they are only interested in the characteristics of the technology and nothing else (e.g., save space on the laptop, performance ?).
Sorry all ... and about NDA etc...but i have to try...
Rys... is there any news(thoughts,theory etc) about "sinergy" between Rogue PowerVR 6 and other cpus ( like cell or even powerpc A2)?
Very interesting info here ...PowerVR is growing... series 6 road to sucess?
http://www.nordichardware.com/news/...ries-of-mobile-gpus-gaining-wide-support.html
(im still believing ps4 coming with customised powerVR 6+ MP32,600/800MHz,28nm,200mm^2...)
Last edited by a moderator:
(((interference)))
Veteran
You know what I was wondering - whether it'd be possible to have the next-gen equivalent of the N64 expansion paks or the SNES enhancement chips.
For example this gen the biggest problem is the lack of memory so if you could insert a 128 MB RAM module into your 360 or PS3 to improve load times, LOD, texture resolution, player/AI counts - I'd think it would be quite worthwhile.
For example this gen the biggest problem is the lack of memory so if you could insert a 128 MB RAM module into your 360 or PS3 to improve load times, LOD, texture resolution, player/AI counts - I'd think it would be quite worthwhile.
The problem with such expansions is they don't get properly used. Devs have to design for the minimum spec, and then spend extra to support a higher spec console. I suppose for texture fidelity a RAM pack is an easy-to-accommodate upgrade, but that's the only expansion you could comfortably fit, and you be better off putting such an expansion on a proper internal bus designed for the purpose. No need for cross-device portable connectivity such as Lightpeak for that! We haven't had a RAM expansion or anything close for 2 generations, and I doubt we'll see a return. The closest we got was Kutaragi's hopes of networked boxes over Gigabit, that went nowhere.
Multi-console via network are already used like in FM2/3 and GT5?, but yes not used for render 720p games in 1080p, GT5 demonstrate some work but don't see a real use in the final game. Market is probably to small for this type of render… May be with console under 200$ you can have more people interested in?
- Status
Similar threads
- Locked
