Jump to content

SB Pro PE 4.0 Hardware Discussion


Parachuteprone

Recommended Posts

  • Replies 91
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Posted Images

On 7/14/2016 at 4:39 PM, Ssnake said:

right now turning legacy code into parallelized code in a non-destructive way is a very difficult job if you're expected to continue developing on the feature side of the application at the same time. But what good, for example, is a factor two or four when we increased the line of sight calculation requirements by (potentially) a factor of twelve when we introduced multi-party capability?

 

And that answers the question I was going to ask perfectly. And that answer points definitively at what will be an extremely difficult decision eSim games will eventually have to make.

 

From a purely business point of view based on limited information, I submit that the current code should stay at 3.0 in it's current state simply because it is legacy code that can no longer be adapted to an increasingly evolving environment - to wit - it can neither be rewritten to take advantage of multiple cores nor can it be rewritten so that currently produced GPU's would benefit much from increased graphic routine optimization.

 

That would, I am sure, be a very unpopular decision with most SB players (I doubt if the military customers would give a hoot though). However, what would be the PR ramifications of releasing a version 4 if its performance is such that your typical PC just won't produce the FPS that contemporary gamers have come to expect. Even with expert optimization of the current code Ssnake has as much as admitted that the performance will not meet the 60 FPS which is rapidly becoming the normal expectation of contemporary gamers.

 

Which leaves a complete rewrite of the code as the one option for 4.0 that would provide a solution with the downside (a major one to be sure) being it may take 2, 3 or even 4 to 5 years to get there.

 

There's a saying, Ssnake, that goes like this: It's GOOD to be the KING.

What it fails to say, though, is that along with the good comes the bad and sometimes, sometimes, it sucks to be the KING.

 

This may very well be one of those times for eSim.

 

Link to comment
Share on other sites

21 hours ago, Werewolf said:

 

And that answers the question I was going to ask perfectly. And that answer points definitively at what will be an extremely difficult decision eSim games will eventually have to make.

 

From a purely business point of view based on limited information, I submit that the current code should stay at 3.0 in it's current state simply because it is legacy code that can no longer be adapted to an increasingly evolving environment - to wit - it can neither be rewritten to take advantage of multiple cores nor can it be rewritten so that currently produced GPU's would benefit much from increased graphic routine optimization.

 

That would, I am sure, be a very unpopular decision with most SB players (I doubt if the military customers would give a hoot though). However, what would be the PR ramifications of releasing a version 4 if its performance is such that your typical PC just won't produce the FPS that contemporary gamers have come to expect. Even with expert optimization of the current code Ssnake has as much as admitted that the performance will not meet the 60 FPS which is rapidly becoming the normal expectation of contemporary gamers.

 

Which leaves a complete rewrite of the code as the one option for 4.0 that would provide a solution with the downside (a major one to be sure) being it may take 2, 3 or even 4 to 5 years to get there.

 

There's a saying, Ssnake, that goes like this: It's GOOD to be the KING.

What it fails to say, though, is that along with the good comes the bad and sometimes, sometimes, it sucks to be the KING.

 

This may very well be one of those times for eSim.

 

.

but esim is arguably past the point of no return

 

meaning a situation where is when its better off to reach the ending that it is to go back to the beginning.

 

From what i gather 4,0 release is not too far off, possibly coming August. AT least with new 3d models and new content, before weather effects and land terrain come later around December. with all the effort put it in.  Wouldn't be easier to finish 4.0, release it , and continue to squad bugs to ensure smoother Fps, than to entirely scrap it and continue stay at 3.0.  it would be a shame to scrap all this effort. 

 

your argument may have been valid if 4.0 was only just annouced or still in early stages of development.

Edited by Kev2go
Link to comment
Share on other sites

  • Members

...and we would still have had to go all the way to confirm that certain changes would have the negative effect on performance that we now know in full detail. Also, while I try to be honest in assessing the situation, let me say again that it does not appear hopeless. There is still good reason to believe that the frame rates can be improved. All I wrote was that it's not an entirely easy thing to do. "Difficult" does not equal "impossible", though.

Link to comment
Share on other sites

  • 3 weeks later...
On 15/07/2016 at 7:39 AM, Ssnake said:

 

Yeah, I can relate to that. :o

 

 

Instinctively I would have said that a CPU upgrade would be more benficial. But then again, looking at the comparison between my notebook's i5 at 2.9 GHz and the desktop PC's i7 at 3.5GHz, the frame rate differences aren't that dramatic. If your other games will profit from the GTX1060, go for it. Assuming that we succeed in improving the overall efficiency of our render engine in the coming months, the performance gains would then be biggest for people with faster graphics cards as the CPU bottleneck is gradually being widened.

 

Like with all predictions about the future, this one comes with a bit of uncertainty as well. But the way I see it, we don't really have much of a choice. I don't see CPUs becoming significantly faster in their single thread performance. We've had a CPU clock stagnation for five years now, which is in pretty remarkable contrast to the past 50 years of microprocessor development. Growth seems to come mostly from getting done more instructions per cycle, and from parallelization; parallelization however has its limits, at least when we're looking at the two...12 CPU cores realm (note that I'm not alone in this. One of the Intel big wigs said in a 2014/15 interview that "parallelization was a failure" (in the sense that it turns out to be almost impossible for all practical matters to actually double or quadruple performance of code by rewriting it from single threaded mode to parallelization)).

 

Now, some processors are looming on the horizon that promise to bring 32 cores within the next one or two years. Maybe this more substantial step up will increase chances to boost performance. But right now turning legacy code into parallelized code in a non-destructive way is a very difficult job if you're expected to continue developing on the feature side of the application at the same time. But what good, for example, is a factor two or four when we increased the line of sight calculation requirements by (potentially) a factor of twelve when we introduced multi-party capability?

 

Anyways, in your situation I would go for the 1060 right now. Even if SB Pro PE 4.0 may not immediately profit from it, other games will. While your other games would ALSO benefit from a CPU upgrade, the overall gain will probably not be dramatic.

 

Does  the new graphics engine allow running more things in parallel? As you say there is nothing in sight for increase mhz but we get more and more cores in the machines. Some graphic engines like Cryengine now utilise CPUs very well. Also many things can be offloaded to GPUs these day; For instance the unity engine now allows much faster gpu rendering of similar objects/textures - something that SB has a lot of. 

 

Also I wonder if sbpro 3 still draw objects despite they are visible? I tried to load up a finish map. I'm guessing but I think that because the map is fairly dense in terms of houses, train lines, etc. that the fps went down to 15-20fps despite nothing was directly visible apart from the trees around me. 

Link to comment
Share on other sites

  • Members

Technically, a few things have been parallelized, and more things will in the future. But first of all, this transition is a slow-going process because we can't suspend the product development while we're working on it, and you need to be careful about the changes because you're inviting all kinds of bugs that are immensely harder to track down in parallel code. Second, unless we get a truly high number of cores the parallelization doesn't scale THAT impressively. Third, simply parallelizing everything isn't the answer. In Falcon 4, everything is parallelized (or so I've been told by people who should know), yet it doesn't actually boost performance because you have a stop for data synchonization at pretty much every step in the code.

 

I have every confidence (and so should you) that our programmers know what they are doing. If they say that it isn't easy, chances are that it's a rather complicated topic. Like with most other topics worth knowing anything about, the more you know the more you realize what you don't know. So, while we are indeed working on this, I would not expect massive results from it in the not-too-distant future. If we were to develop Steel Beasts from scratch we would probably do a lot of things differently. But for the moment at least we're stuck with what we have.

Link to comment
Share on other sites

Well I finally tested 4.0 and I'm relieved to say that it is completely playable on my rig (specs in sig below). My AMD CPU is comparable to an intel Core i3, but even at ~10fps things were bearable (my previous lows were ~16fps). And it only gets like that at the peak of my benchmark scenario where hell is breaking loose. Enabling rain bumps of framerates into the 20s under the same condition. Otherwise, under "normal" circumstances, it's like I got a lot of eye-candy at a minor cost in performance. So, consider me a happy camper, and thank you for the "alarmist" approach in the performance estimates here.

Link to comment
Share on other sites

This is starting to get sad.

The performance of SB was always pretty underwhelming during the last years, at least compared to technical standards these days and to what the sim offers in the end, in my opinion.

At this point however, when I read from a dev that the engine is basically completely outdated, (ab)used beyond capacity and capability, then I wonder why you still try to 'fix' things, while admiting that it just isn't enough,

rather than do the only logical thing: upgrade the engine/get a new one.

 

The engine surpassed its life-time years ago and while 4.0 offers nice new features - what good are they if the consequences are such abysmal performance issues that render the sim unplayable for a lot of people?

Link to comment
Share on other sites

  • Members

Well, we can't "just switch the engine". If it were that simple we would have done it years ago (actually, we kinda did it with 3.0). If we did that, we'd have to suspend all other development for years - but we have customers who cannot wait that long for their training requirements to be addressed. We're not developing games from scratch. We must find a balance between a lot of factors, a high frame rate being important, but hardly the only point of consideration.

The low frame rate was something that emerged in April. And even then it did not appear as dramatic to me and the programmers because either the hardware that we had covered up the deficits for lower end machines, or test runs were done while debugging, which always kills frames. We didn't really have a chance to turn around the wheel much after the issues became apparent. Our priority was to make it stable and reliable while retaining the chance for a summer release (mission accomplished, I'd say), the next step is to address the performance issues while finishing the work on the new terrain. I'm not expecting utter miracles, to be honest, this is a process that will take quite a while to address comprehensively. But I still think that subsequent updates in the coming six to twelve months will help boosting frame rates to more acceptable levels. I mean, it's not as if we are in denial about this, all right? I said so, publicly, on the record, and before releasing the software - in the first post of this thread that is dedicated to nothing but Hardware Recommendations. It's not an attempt to bury the topic under layers of feelgood phrasing.

 

We know what needs to be done, and we will do what's necessary.

Link to comment
Share on other sites

On 7/15/2016 at 0:42 PM, Skybird03 said:

I understand that. Thanks for taking the time nevertheless!

 

I'm in no urgent hurry to get a new gfx board and monitor, I just see that the cockpit experience in Assetto Corsa would benefit from a wider perspective (the mirrors, and looking at apex in corners...), thats all. My system is stable and reliable since many years and gets everything done I throw at it. Plus I want to stick with my "locked up" Windows 7 (no longer updated due to Microsoft's fantastic GWX bullying campaign which indeed made me upgrading my operation system: I now do all and everything under Linux Mint , and just launch some games under Windows); I have no plan to move to Windows 10 any time soon. If ever. If I would: then just as a game launching platform, nothing else. - The new times of "digitalness" are here. Something tells me they are not as good and well-meaning as expected.

 

If you ever consider to push SBP to higher DirX, please leave people the choice: do not go for Win10-only DirX12, but stick with DirX11. I am far from being alone with my anger and scepticism about Microsoft.

I haven't yet bought 4.0 since I doubt whether my computer will be up for it so my first question is what system specifications do you have? I have 4 gb of memory and a 3 mhz pentium processor and a gforce gt220 graphics card. And then another question: you mentioned that you're running it on a linux machine. This is something I have always wanted to do, So, how do you do it? And where did you subsequently get the win 7 version from? I take it you're running it through linux with some kind of emulator?

Link to comment
Share on other sites

Ok I have an I7(4th Gen) system, EVGA GTX 970 Graphics card, 20 gigs of Ram, windows 10. I get playing in a game  everything runs ok for a good while. The problem is I start switching between diff sites. Then I get the black screen of death. I can hear the vehicle still running but the screen is black. I use a 32" Panasonic tv as my monitor. Any ideas what could be wrong. Could it be my graphics card getting too hot, or a drop in fps?

 

Link to comment
Share on other sites

10 hours ago, Rosmarus said:

I don't know if my experience is of any use to anyone but I seem to gain most FPS when tuning ground clutter down. My rig is i7-4790K @ 4 GHz, GTX 970, 16 gb of RAM and Windows 10. I run at 1080p.

Well it's of a bit use to me because your processor is not much bigger than that of mine, so the processor can't be the issue. You have a lot more ram though......damn! I guess there's no way around having to buy a new "rig"....sumbitch! 80531324398

Link to comment
Share on other sites

On 8/11/2016 at 9:35 PM, Scrapper_511 said:

The Release Notes recommends at least 8Gb RAM. I'm not sure what OS you have running with 4Gigs, but mind you, 4.0 requires a 64-bit OS, of Win7 or up.

 

I'm using Window's 10 64 bit. My main concern with 4 gig of ram is loading big maps. Maybe, I'll be able to get by with 4 gig until they release the hi-def terrain?

 

Oh, I forgot to mention! My video card is 1 gig and total virtual memory is 8 gig.

Edited by Stormin Norman
Forgot to mention
Link to comment
Share on other sites

6 hours ago, Stormin Norman said:

 

I'm using Window's 10 64 bit. My main concern with 4 gig of ram is loading big maps. Maybe, I'll be able to get by with 4 gig until they release the hi-def terrain?

 

Oh, I forgot to mention! My video card is 1 gig and total virtual memory is 8 gig.

Adding another 4GB of RAM is probably cheaper than the upgrade to 4.0 and will make you PC snappier for other applications as well. 

Link to comment
Share on other sites

Looking at some of the screenshots that users post might indicate that some of them might not have the HDR bloom effect turned on, since the default is off. I didn't really pay attention to it before, but now it makes more of a difference particularly with glowing tracers combined with some of the fire effects. I think it also slows the framerates down more generally than it did before. My system was top of the line about 6 years ago, I tended to run with everything maxed out but a resolution of 1024 x 768, which usually did well in most circumstances. But here the ICM artillery effects and the bloom effect (not per se the particle effects, but just in general- the sky, the ambient environment) seem to start to slow down scenarios a bit more than before. Still, if there are a few users out there who haven't turned on the bloom effect, they ought to have a look, I hadn't thought to do this until now. The effects are better for certain.

 

 

 

 

 

 

 

 

SS_22_04_09.jpg

Edited by Captain_Colossus
Link to comment
Share on other sites

Sbpro 4 running at 3440x1440 and almost all settings maxed out I get ~35% use of my gpu. My CPU has one core maxed out and a secondary core ~30-40% use. 

 

If you can get a CPU to overclock to 5ghz it may be your best bet for improving fps. 

Edited by inexus
Link to comment
Share on other sites

  • 2 weeks later...

You know, I know it's a lot of pain porting to a new more modern engine but the sooner that gets done the sooner we'll have a better SB Pro PE that's well optimized and looks gorgeous at the same time.

 

Time will pass anyway. If this was done say, 4 years ago ...

 

I think the issue is the limited personnel that Esim has. There aren't enough people to do the porting to a new engine and support current military customers. But the longer this is deferred the more painful it will be for Esim. I'm not saying it's easy, I'm saying it NEEDS to be done if SB Pro PE is to be selling well and alive in the next 10 years.

 

Personally, I'm not bothered much by SB Pro PE graphics. When it came out it blew everything else out of the water.

It's more a sim than a game. But Esim is compelled to improve on things and with it the graphics. The strain of an old engine is going to show sooner or later if you keep on improving the graphics. But I guess that's life so a new engine may not be optional, however painful. The alternative is a badly run game which won't appeal to new gamers.

 

Edited by []_--__[]KITT
Link to comment
Share on other sites

  • 2 weeks later...

As far as I know.  A i5 at 3.4 Ghz, is the same thing as a i7 at 3.4 Ghz.  The only difference is that the i7 3.4 Ghz can use hyper-threading, but if the game/sim in question dosent use hyper-threading you are basicaly using the same CPU.  Iam not talking about the new gen CPU´s mind you, I have no idea how the Haswel CPU´s are working in this regard.  Also you must balance your rig, a better GPU (faster) will not benefit from a slow CPU thus the bottle-neck effect.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...