DCS on STADIA - Page 2 - ED Forums
 


Notices

Reply
 
Thread Tools Display Modes
Old 08-22-2019, 10:17 AM   #11
Wolve03
Junior Member
 
Join Date: Oct 2018
Location: New Delhi, India
Posts: 38
Default

I agree that it can provide massive processing power. And that's why it's used, but that processing power comes at a cost as well. Whether it's a subscription based service or a usage based service, do you really think it'd be feasible for gamers? I can think of a "cheap" option where you subscribe and get access to a large library of games (think Netflix) but will you have access to really old or obscure titles? And if yes, will they be available for eternity, or only until the provider retains rights? ... and what about EPIC exclusives



As for latency, how can you get over transmission speeds and processing time? Heck, serious gamers even count latency from the input lag of their input devices and the lag to their display devices. Those will still exist, and then you add network latency on top of it. As for implementation, we still haven't seen complete implementation of IPv6 across the world, and the demo runs you may have seen for remotely operated cars (like in that hill climb event) or for the tech demos of these new services, all have had dedicated links and servers nearby.
Now scale that to a city / state, and you can perhaps see the latency going up. If you have ~50 msec latency to your game's MP server, you need to add a bit more because now everything you see / do has THAT latency + the latency to Stadia's servers. And let's not forget that distributed systems would also have proxies in front which would add processing overheads (to route that request to the corresponding compute cluster / device).


And assuming that all of this works perfectly, it would certainly mean that even [the game featuring dragons]* becomes an online game.


*Sorry if I went off-topic for a couple of those, or referenced other games.
Wolve03 is offline   Reply With Quote
Old 08-22-2019, 10:25 AM   #12
1987kess
Junior Member
 
Join Date: Jan 2019
Posts: 30
Default

I hope not.

On a sim like DCS, or any other sim racing, we need very precise inputs and minimum input lag that we can. Add the latency of an internet connection and it results in a bad idea.

But the worst thing of Stadia is that is an absolutely closed platform: you buy the game, and play it, that's everything.

This is totally contrary on what sim and PC communities have ever been: Goodbye mods, goodbye editing luas, goodbye customized drivers, custom setups...

Could be great to play FIFA or call of duty or whatever on a mobile phone, but really can't see the point in DCS.
1987kess is offline   Reply With Quote
Old 08-22-2019, 01:06 PM   #13
ravenzino
Member
 
ravenzino's Avatar
 
Join Date: Oct 2018
Location: Australia
Posts: 253
Default

Quote:
Originally Posted by Wolve03 View Post
I agree that it can provide massive processing power. And that's why it's used, but that processing power comes at a cost as well. Whether it's a subscription based service or a usage based service, do you really think it'd be feasible for gamers? I can think of a "cheap" option where you subscribe and get access to a large library of games (think Netflix) but will you have access to really old or obscure titles? And if yes, will they be available for eternity, or only until the provider retains rights? ... and what about EPIC exclusives



As for latency, how can you get over transmission speeds and processing time? Heck, serious gamers even count latency from the input lag of their input devices and the lag to their display devices. Those will still exist, and then you add network latency on top of it. As for implementation, we still haven't seen complete implementation of IPv6 across the world, and the demo runs you may have seen for remotely operated cars (like in that hill climb event) or for the tech demos of these new services, all have had dedicated links and servers nearby.
Now scale that to a city / state, and you can perhaps see the latency going up. If you have ~50 msec latency to your game's MP server, you need to add a bit more because now everything you see / do has THAT latency + the latency to Stadia's servers. And let's not forget that distributed systems would also have proxies in front which would add processing overheads (to route that request to the corresponding compute cluster / device).


And assuming that all of this works perfectly, it would certainly mean that even [the game featuring dragons]* becomes an online game.


*Sorry if I went off-topic for a couple of those, or referenced other games.
I'm not disagreeing with you. If you look at my first post in this thread, you'll see my standing point. With the currently commercially available tech and average infrastructure in most places of the world, you are not wrong.

The point I'm trying to make here, is that while STADIA might not be suitable for DCS anytime soon, let's just don't rush to the conclusion of denying the merit and feasibility of using cloud computing for simulation in a longer term (say 5-10 years) and throw OP's idea into trash bin straight away, as 1) its power is unmatchable; 2) all the blocking problems are solvable via technologies either under development or already in early adoption.

Here're a few of them:
1) Latency
The current typical latency of accessing a DCS server, or basically any game server is somewhere between 30ms to 500ms or higher, give or take. To compare, local display lag is some 30+ms, input lag is some 4-8ms, current CPU/GPU processing lag is variable at about 17ms (60FPS) or above. Now, 5G latency is expected to be nearly 1ms. Realistically, probably a couple of ms. To assist understanding, that low latency is partially achieved via 5G network itself, and partially via edge computing, which is literally putting some computing power at your nearest mobile station/local exchange, which is typically a couple of km from you at most, if not some hundreds of meters. The next generation fibre transmission tech can also achieve similar performance. In fact, that is part of 5G as well (backhaul). I didn't say this will happen over night, but global telco operators are indeed on the way.

2) Throughput
To stream 4K video, you'll need 40Mbps network bandwidth. This might be a problem in some places of today's world, but shouldn't be as problematic in the next 5-10 years. Especially when 5G becomes largely available, some rich hardcore gamers don't even need a fibre, if a fibre is still not available by then...

3) Reliability
While network itself is drastically improving its availability (which is witnessed over the past decade), we always have the option of authorizing mission critical decision to be handled locally, which doesn't necessarily negate the merit of getting extra assistant from the cloud. Today, most, if not all the demos we saw are conceptual with the focus on showing its feasibility, while some people might easily become picky and rush to a negative conclusion by looking at its reliability side, which those demo aren't really designed for. Again, let's just be rational and not go from one extreme to another.

4) Remote Rendering
Nvidia demoed remote rendering 5 years ago at GDC 2014 (https://developer.nvidia.com/sites/d...chnologies.pdf)
A possible architectural change would be shifting local GPU into cloud. Now we would be talking about using an array of GPUs. DCS needs to support multiple GPUs of course... It would be, right? (Looking at ED)

5) Cost model
IMO, there could be two options.
a. DCS becomes cloud native, and sell as SaaS. Yea, in this case, it would likely be a subscription. But tbh, it wouldn't deter too many DCSers, as long as DCS local version still available, which has to be as it would be silly to assume everyone would have a good network connection.
b. You no longer need to own a home PC, instead, you own a virtual "cloud PC". Your some $2000 per 3 years PC upgrade cost can well afford that IaaS. Think about this: in the last 24 hours, how much of your CPU and GPU capacity is actually occupied and at what percentage? Assume averagely less than 20% overall. So if same HW is shared by 5 people, and charge everyone for 50% of its HW cost, it would be $1000 per person over 3 years. The total revenue would be $5000. Well, you can debate from different angles and come up with your own calculation model, but you see the economics in it right?

Please ED don't delete this discussion. It's tech nerds' joy, but could still benefit DCS one day. Let's have some dream~
__________________
i9-9900K, G.Skill 3200 32GB RAM, AORUS Z390 Pro Wifi, Gigabyte Windforce RTX 2080 Ti, Samsung 960 Pro NVMe 512G + 860 Pro 1T, TM Warthog HOTAS, VKB T-Rudder, Samsung O+
F/A-18C, A-10C, UH-1, AV-8B, F-14, FC3, SA342 Gazelle, L-39, KA-50, CEII, (Almost abandoned: CA - VR support please?)
PG, NTTR

Last edited by ravenzino; 08-22-2019 at 01:17 PM.
ravenzino is offline   Reply With Quote
Old 08-23-2019, 07:00 AM   #14
Wolve03
Junior Member
 
Join Date: Oct 2018
Location: New Delhi, India
Posts: 38
Default

Quote:
Originally Posted by ravenzino View Post
I'm not disagreeing with you. If you look at my first post in this thread, you'll see my standing point. With the currently commercially available tech and average infrastructure in most places of the world, you are not wrong.
Right. I noticed your first post ... but we need a few points and differing opinions to have a discussion, right?


I am not throwing the idea in the bin either because considering businesses, it may happen eventually; Microsoft wanted Windows to be provided on a SaaS model.
But I am not a big fan of SaaS and prefer on-premises installations. Consider this, for instance ... ED came up with a new licensing system, but had to provide a few days' "backup" in case someone was playing without an Internet connection. No matter how advanced we get, the whole world does not move at the same pace and you will have islands / pockets where you have high bandwidth, low latency connections, and the regular ones. So, this solution may not be feasible for all and if they deploy it in high-tech societies, then the monetary benefits may not really help as such a society will probably have the purchasing power anyway.


To further discuss ...
1. Latency. Broadly agree on your points but I doubt Stadia will use edge computing. It will just increase the costs unless you see large-scale acceptance.


2. Higher bandwidth will become available, but we will need newer network protocols as well, for better throughput. Considering you are techie yourself, you'd know that low latency data generally uses UDP, but TCP sessions are known to result in slight packet loss on the line as they throttle. The simple solution is QoS but how many ISPs or home routers implement it properly? Then there's the point I was trying to make by mentioning IPv6; adoption will take a long time unless the businesses see this as an absolute money-making idea.


3. Reliability will increase; agree. But as we get more people, we get more traffic and competing flows. Probably your neighbour feels that his VoIP call should be higher priority than your games. Or 20 of your neighbours suddenly figure they should all stream super high-def content together. It works even right now as well but as bandwidth increases, so much the quality of the content and then the bandwidth requirements.



4. Remote rendering is good, but that removes mods we use for aesthetics ... unless you do some post-processing locally as well. I figure it could be a hybrid model with rather than the full rendered frames being sent, perhaps computed data could be sent. I remember that in the last decade, ATI had come up with an idea that if you had 100 boxes in a frame, just calculate it once and then copy it all over. Similarly, process the raytracing etc. and let the local device render it? There are render farms and computer clusters already, right? It's all essentially data in the end.



5. I'd rather not comment on the cost part.



A few merits ...
1. No more desync issues if everyone is coming from the same source
2. Perhaps some tasks could be offloaded to the cloud to have a hybrid approach, rather than full Stadia. I can think of CPU-intensive tasks for scripts, AI control etc. Maybe, some heavier GPU tasks could also be done remotely and then sent back. I recall a few research papers on Predictive models for CPUs, through ML. Maybe that could help prepare stuff before it is needed by the client, and that could be a good way to avoid latency.
3. It could be cheaper than maintaining a gaming PC and a games library (although Steam Sales help), and could see success similar to Netflix, Spotify and others. But as with the different stores / licenses thing ... how long will it last?



And additional points
1. ED would likely need to run the dedicated server if Merit (1) is to be achieved, in hand with Google. Although, I am not sure if we want to give a single entity as much power ... look at how a few games got blocked in Iran due to political reasons.
2. What about mods etc. we so fondly do for single player games? In case of DCS, many people use texture mods, or reshade for aesthetic reasons. That goes out of the picture (1987kess mentioned this)
3. I am not a big fan of subscribing, instead of buying stuff. With subscriptions (and I am talking of more games than just DCS), you are the mercy of your subscription service provider. Often times, I watched a movie or series on Amazon Prime / Netflix and then later found it removed. There was a famous case with Apple last year where a person purchased a song but then lost access to it because Apple figured they will not carry it further. We are gamers and some of us even play games from the last decade. Will these services maintain an exhaustive list of all games? Even GOG or Steam don't do that
4. Do we really need that much reliance on clouds in our lives? What if you have a freak el-nino and lose all your data as the clouds fly away
Wolve03 is offline   Reply With Quote
Old 08-23-2019, 08:13 AM   #15
epoch
Member
 
epoch's Avatar
 
Join Date: Dec 2007
Location: The Future
Posts: 388
Default

This won't work for DCS, for one simple reason.

Google is an advertising company and exists solely to sell ads.

In DCS, unless you're in a chopper (or playing Combined Arms) we're just gonna be too damn high up to see the billboards.

epoch is offline   Reply With Quote
Old 08-23-2019, 08:31 AM   #16
Xilon_x
Member
 
Join Date: Aug 2018
Location: Italy
Posts: 753
Default

I believe that wolve is right perhaps for the world of the Simmers and therefore for all the simulators like dcs, falcon 4 bsm, the -2 sturmovik, jane's simulation, microsoft flight simulation, x-plane, ecc.ec. it'll take another kind of cloud platform technology.
Xilon_x is offline   Reply With Quote
Old 08-23-2019, 09:08 AM   #17
Wolve03
Junior Member
 
Join Date: Oct 2018
Location: New Delhi, India
Posts: 38
Default

Quote:
Originally Posted by epoch View Post
This won't work for DCS, for one simple reason.

Google is an advertising company and exists solely to sell ads.

In DCS, unless you're in a chopper (or playing Combined Arms) we're just gonna be too damn high up to see the billboards.

Heh, reminds me of when SWAT IV tried showing ads in the game. The community was in an uproar on being served ads despite having purchased the game.
But who knows what these tech giants throw our way in the future. They would want and advertise cloud because they can easily make you reliant on them, in such a case. Just consider how many of us had our GMail accounts and now have Android phones updating contacts, photos and what not. Maintaining those services, by using more Android devices, becomes kinda important if we still need easy access to all that data. (this is just as an example)


As Ravenzino and others have also said ... it may make sense in some cases. But I fear how quickly it has the potential to ruin the demographics.
Wolve03 is offline   Reply With Quote
Old 08-27-2019, 10:00 PM   #18
MiG21bisFishbedL
Senior Member
 
MiG21bisFishbedL's Avatar
 
Join Date: Apr 2014
Location: 'Murrica
Posts: 1,510
Default

Ads in a game could definitely work, especially if they could be relevant to the game. For example, sports games. Also, I recall there being some fervor about ads potentially being in Deus Ex Human Revolution. Given the genre, ads would've been welcome as part of the critique it offered on corporate and consumer culture.

Quote:
Originally Posted by epoch View Post
This won't work for DCS, for one simple reason.

Google is an advertising company and exists solely to sell ads.

In DCS, unless you're in a chopper (or playing Combined Arms) we're just gonna be too damn high up to see the billboards.

Dohohohoh.

But, on serious note, Google also has the collective attention span of a five year old. I wish I could have read this on Google Glass.
__________________

Move MiG. For great justice. You know what you doing. Take off every MiG.
CH Combatstick, CH Pro Throttle, CH Pro Pedals

Last edited by MiG21bisFishbedL; 08-27-2019 at 10:03 PM.
MiG21bisFishbedL is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT. The time now is 10:14 AM. vBulletin Skin by ForumMonkeys. Powered by vBulletin®.
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.