Matrix Games Forums

Forums  Register  Login  Photo Gallery  Member List  Search  Calendars  FAQ 

My Profile  Inbox  Address Book  My Subscription  My Forums  Log Out

Good improvement, but two small gripes

 
View related threads: (in this forum | in all forums)

Logged in as: Guest
Users viewing this topic: none
  Printable Version
All Forums >> [New Releases from Matrix Games] >> Command: Modern Operations series >> Good improvement, but two small gripes Page: [1]
Login
Message << Older Topic   Newer Topic >>
Good improvement, but two small gripes - 11/16/2019 5:37:07 PM   
JPFisher55

 

Posts: 533
Joined: 11/22/2014
Status: offline
I have been playing a very large saved game left from CMANO on CMO. So far I like the changes and improvements. The latest update fixed two minor bugs. I have experimented with Tacview. I don't find the god'seye view a problem because I only use tacview to look at one airplane. Watching an airplane bomb is neat. I have yet to observe a dogfight on Tacview and look forward to it. I have the new message board, but generally prefer the old one which is available and now resizable. I do occasionally use the new board.

Overall, an excellent improvement to a great game.

I do have two small issues. I would like to get back the right click center. Holding the right click to pan is okay, but not as fast as the right click to recenter the map. The scrolling works well to zoom in and out.

The other small gripe comes with two questions. CMO definitely plays this large scenario faster than CMANO. However, I was hoping for even faster performance. I have an 8 core, I I9-9900 intel chip with 32GB RAM computer. So my questions are: Is CMO a 32 bit or 64 bit program? Is a 64 bit program that much faster than a 32 bit program?
Post #: 1
RE: Good improvement, but two small gripes - 11/16/2019 7:54:09 PM   
LMychajluk

 

Posts: 106
Joined: 10/8/2017
Status: offline
quote:

ORIGINAL: JPFisher55
...So my questions are: Is CMO a 32 bit or 64 bit program? Is a 64 bit program that much faster than a 32 bit program?


It's not 64-bit. You can see this in the Windows Task Manager. 64-bit isn't inherently faster, per se, but it does allow the program to use more memory (if available). That being said, I still wish it was 64-bit...

(in reply to JPFisher55)
Post #: 2
RE: Good improvement, but two small gripes - 11/16/2019 8:20:39 PM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
I am always surprised by the obsession with 32 vs. 64. I have had several discussions with the programmers that report to me and this is the paraphrased summation:

1) 64 bit OSs and programs built for them are more efficient in using RAM, for the right kinds of programs. One coder told me that 64 bit shows up as preferred, even in small programs, where there is a lot of moving info from storage and graphics back to the CPU. (there was actually conflicting opinions on that)
2) Because of the way indexing and pointers work between 32 and 64 bit Windows OSs, 64 bit programs use more memory to perform the same function. So below a certain point, 64 bit can actually be slower and use more memory than 32 bit.
3) The crossing point for efficiency is typically between 4Gb and 5Gb. The larger memory usage of the 64 bit program vs. the limits on memory usage cross over at that point.
4) if your program is using more than 4Gb of RAM, the RAM limit and memory efficiencies will push you to using 64 bit memory handling.

My tests have shown that CMNAO and CMO both use significantly less than 4Gb of RAM even in some large scenarios. Its been 20 years since I coded so this is somewhat hearsay from my coders. Take my interpretation with a grain of salt. I had originally asked because of a discussion on another game forum where the same complaints were raised about 32 vs 64 bit performance.

Again, this is my paraphrasing of they were telling me. I suspect that CMO would more benefit from more multi--core support, even though it does some today.

(in reply to LMychajluk)
Post #: 3
RE: Good improvement, but two small gripes - 11/16/2019 8:34:25 PM   
marksdoran

 

Posts: 35
Joined: 7/7/2017
Status: online
And more address space for the program can make things a lot faster *if* the working set in memory gets larger, particularly if it's more than 4GB. I haven't looked to see if that's the case (task manager and performance monitor can provide insight). Larger scenarios would presumably have larger working sets.

It was a surprise to me that there's no 64-bit executable with this new release. That said: I know from experience with game code bases that are, shall we say, steeped in long development history that making them build 64-bit clean can be non-trivial amounts of work. If the working sets comfortably fit in 4GB of address space then that clean-up/porting work might be hard to ask the development team to take on given the marginal return on doing it.

(in reply to LMychajluk)
Post #: 4
RE: Good improvement, but two small gripes - 11/16/2019 8:48:53 PM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
I don't think it is near 4Gb, from what I can see. But I also don't use the new overlays. Not sure what the impact would be.

(in reply to marksdoran)
Post #: 5
RE: Good improvement, but two small gripes - 11/17/2019 10:34:55 AM   
apache85

 

Posts: 1632
Joined: 12/18/2014
From: Melbourne, Australia
Status: offline

quote:

ORIGINAL: JPFisher55

I have been playing a very large saved game left from CMANO on CMO. So far I like the changes and improvements. The latest update fixed two minor bugs. I have experimented with Tacview. I don't find the god'seye view a problem because I only use tacview to look at one airplane. Watching an airplane bomb is neat. I have yet to observe a dogfight on Tacview and look forward to it. I have the new message board, but generally prefer the old one which is available and now resizable. I do occasionally use the new board.

Overall, an excellent improvement to a great game.

I do have two small issues. I would like to get back the right click center. Holding the right click to pan is okay, but not as fast as the right click to recenter the map. The scrolling works well to zoom in and out.

The other small gripe comes with two questions. CMO definitely plays this large scenario faster than CMANO. However, I was hoping for even faster performance. I have an 8 core, I I9-9900 intel chip with 32GB RAM computer. So my questions are: Is CMO a 32 bit or 64 bit program? Is a 64 bit program that much faster than a 32 bit program?

Hey JPFisher55,

First I'm glad you're enjoying the improvements with C:MO. Performance has been the major focus in the development drive from C:MANO to C:MO. That said, I'm sorry to hear that the increases aren't meeting your expectations.

If you can post a save file we can look more closely at how to tune the performance of C:MO and make changes in a future patch.

Supplementary to that, if you are working with a very large save file it is likely that the scenario is not optimised. I am happy to help you improve the performance of your scenario, just DM me with a copy of the .scen file and we can take it from there.


_____________________________


(in reply to JPFisher55)
Post #: 6
RE: Good improvement, but two small gripes - 11/17/2019 4:03:24 PM   
JPFisher55

 

Posts: 533
Joined: 11/22/2014
Status: offline
I can't attach the saved game file to any communication method on the Matrix Games forum or private message system because the zip file is too large.

(in reply to apache85)
Post #: 7
RE: Good improvement, but two small gripes - 11/17/2019 4:35:08 PM   
magi

 

Posts: 1377
Joined: 2/1/2014
Status: online

quote:

ORIGINAL: JPFisher55

I have been playing a very large saved game left from CMANO on CMO. So far I like the changes and improvements. The latest update fixed two minor bugs. I have experimented with Tacview. I don't find the god'seye view a problem because I only use tacview to look at one airplane. Watching an airplane bomb is neat. I have yet to observe a dogfight on Tacview and look forward to it. I have the new message board, but generally prefer the old one which is available and now resizable. I do occasionally use the new board.

Overall, an excellent improvement to a great game.

I do have two small issues. I would like to get back the right click center. Holding the right click to pan is okay, but not as fast as the right click to recenter the map. The scrolling works well to zoom in and out.

The other small gripe comes with two questions. CMO definitely plays this large scenario faster than CMANO. However, I was hoping for even faster performance. I have an 8 core, I I9-9900 intel chip with 32GB RAM computer. So my questions are: Is CMO a 32 bit or 64 bit program? Is a 64 bit program that much faster than a 32 bit program?


"I would like to get back the right click center. Holding the right click to pan is okay, but not as fast as the right click to recenter the map."
hear hear.... i agree....

(in reply to JPFisher55)
Post #: 8
RE: Good improvement, but two small gripes - 11/17/2019 8:41:17 PM   
JPFisher55

 

Posts: 533
Joined: 11/22/2014
Status: offline
I forgot to mention my favorite improvement in CMO. It is the ability to left click on a group of units and then left click on one of them in a drop down box to select it. Now when I want to select an airbase surrounded by AAA units, I can just click near the base and then select it from a drop down box. This is a lot easier to clicking to cycle through units or zooming in to select the base.

(in reply to magi)
Post #: 9
RE: Good improvement, but two small gripes - 11/17/2019 9:19:14 PM   
mavfin

 

Posts: 274
Joined: 6/22/2010
Status: offline

quote:

ORIGINAL: magi

"I would like to get back the right click center. Holding the right click to pan is okay, but not as fast as the right click to recenter the map."
hear hear.... i agree....

Well, there's a thread on this. The change is by design, and it looks like several of us like the new choice, rather than the right-click center, which I never did like.

YMMV. Anyway, it's not a bug. It's a feature.

(in reply to magi)
Post #: 10
RE: Good improvement, but two small gripes - 11/18/2019 2:16:56 AM   
Eggstor

 

Posts: 285
Joined: 1/24/2016
Status: offline
To add to the 32-vs-64-bit question, a 64-bit program requires a 64-bit OS. Most Windows installations on systems with less than 4 GB of RAM tend to be 32-bit versions, and changing between 32-bit and 64-bit Windows involves a clean install.

(in reply to mavfin)
Post #: 11
RE: Good improvement, but two small gripes - 11/18/2019 6:21:59 PM   
JPFisher55

 

Posts: 533
Joined: 11/22/2014
Status: offline
I have a suggestion for the developers. Develope a 64 bit version of CMO and offer it for an additional fee. Warn users that the 64 bit version will only help those with high end computers. IMO, a sufficient market for a pure 64-bit version exists. I would gladly pay at least another $40 for it.

(in reply to Eggstor)
Post #: 12
RE: Good improvement, but two small gripes - 11/18/2019 7:16:56 PM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
I don't think you are going to get a great deal of improvement in CMO. Its not a huge memory hog and its not taxing i7 PCs right now. On the other hand, TacView can stress it out when running with CMO.

You might see some minor improvement, but not a huge amount. You'll only see a big improvement if they strip it down completely and rebuild it to take advantage of the advanced memory usage and hyperthreading. There are not many games that have done that. No wargames I know of have done the rebuild. A few have rebuilt portions to better take advantage of a couple extra cores to offload some tasks, but that's all I have seen.

Again, knot sure where the obsession for 64 bit comes from. I would rather see that time investment in more engine optimization and new simulation aspects.

(in reply to JPFisher55)
Post #: 13
RE: Good improvement, but two small gripes - 11/20/2019 11:54:13 PM   
apache85

 

Posts: 1632
Joined: 12/18/2014
From: Melbourne, Australia
Status: offline
I've had a look at the scenario and had some e-mail discussion with JPFisher55 about some steps to optimise performance on large scenarios, but thought it worth following up here with what is causing the slow performance.

The save file is 22.0Mb and utilises 5588 active units.

That's a monster. As a comparison I try to limit my scenarios to around 2,000 AU, to which I'll often include a couple of hundred 'clutter' units like civilian air and sea traffic as well as biologics and false contacts. This still leaves room to simulate multi-day, theatre level scenarios that include complex event framework and have added features (Sakura Blue is a good example--all of the Taiwanese OOB and a large portion of Japan's OOB duking it out over a week, with a scripted typhoon coming through and messing things up, weighing in at 2,427 AU on scenario load, and makes a savefile at around 1.4MB).

C:MO does run scenarios much faster than C:MANO, but since there is no limit to how large you can make scenarios, people will inevitably want to build bigger and more complex scenarios. That's a good thing, but when you add enough--and 'enough' as a number depends on your system as well as the steps you take to optimise scenarios--performance will of course be impacted.

Some of the tools that I use to optimise scenarios are:
o Single unit airbases
> Including using scripting to convert single unit airbases to multi-unit airbases and back again if required
o Using 'blind' sides for clutter units like civilian air/sea traffic, markers and biologicals
o Avoiding large, complex areas for triggers (Lua scripts on regular time triggers work really well here and allow more finegrained control)
o With C:MO, disabling AI calculations for units like runways, buildings, parking spaces etc using the new Lua extensions--this has potential to make huge performance gains in scenarios like the example above (3,187 units could have their AI disabled, leaving 2,401 with active AI calcs on each pulse)
o Cleanup scripts to eliminate units that are no longer useful or needed

So to sum up, there are tools available to enable building large and complex scenarios that run well, and C:MO runs scenarios much faster than C:MANO. However, short of implementing a ceiling on active units, users will always be able to build scenarios large and complex enough that performance will suffer.



_____________________________


(in reply to thewood1)
Post #: 14
RE: Good improvement, but two small gripes - 11/21/2019 12:00:17 AM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
5558 units! Do people really think they shouldn't see some slow performance with a massive scenario like that? I am doubtful 32 bit or 64 bit would make much of a difference.

The devs should put a hard upper limit on scenario size to stop these kinds of threads.

(in reply to apache85)
Post #: 15
RE: Good improvement, but two small gripes - 11/21/2019 12:31:50 AM   
76mm


Posts: 3799
Joined: 5/2/2004
From: Washington, DC
Status: online
quote:

ORIGINAL: thewood1
The devs should put a hard upper limit on scenario size to stop these kinds of threads.

Or maybe we should let people play big scenarios and simply put up with the horrible trauma of an occasional thread like this.

< Message edited by 76mm -- 11/21/2019 12:32:11 AM >

(in reply to thewood1)
Post #: 16
RE: Good improvement, but two small gripes - 11/21/2019 12:36:21 AM   
JPFisher55

 

Posts: 533
Joined: 11/22/2014
Status: offline
Dear Apache85;

Will you please post instructions on how to disable AI calculations for those units using LUA for those who really don't understand LUA. IMO, a lot of players like these large scenarios, so making them run faster will benefit quite a few players.

(in reply to apache85)
Post #: 17
RE: Good improvement, but two small gripes - 11/21/2019 12:53:23 AM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
Its not traumatic they play out of scope scenarios, its the fact devs have to step in and waste time babysitting them through their own self-inflicted trauma.

< Message edited by thewood1 -- 11/21/2019 12:54:04 AM >

(in reply to JPFisher55)
Post #: 18
RE: Good improvement, but two small gripes - 11/21/2019 12:55:09 AM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
"So to sum up, there are tools available to enable building large and complex scenarios that run well, and C:MO runs scenarios much faster than C:MANO. However, short of implementing a ceiling on active units, users will always be able to build scenarios large and complex enough that performance will suffer."

And that right there is it.

(in reply to thewood1)
Post #: 19
RE: Good improvement, but two small gripes - 11/21/2019 1:23:34 AM   
76mm


Posts: 3799
Joined: 5/2/2004
From: Washington, DC
Status: online
quote:

ORIGINAL: thewood1
Its not traumatic they play out of scope scenarios, its the fact devs have to step in and waste time babysitting them through their own self-inflicted trauma.

Technically it is not out of scope, simply very large. Somehow I doubt we'll be seeing many complaints here about poor performance with scenarios featuring 5000+ units. Certainly vastly fewer of these complaints than the complaints there would be if there was some arbitrary limit on scenario size.

(in reply to thewood1)
Post #: 20
RE: Good improvement, but two small gripes - 11/21/2019 3:19:35 AM   
KnightHawk75

 

Posts: 198
Joined: 11/15/2018
Status: offline

quote:

ORIGINAL: 76mm

Technically it is not out of scope, simply very large. Somehow I doubt we'll be seeing many complaints here about poor performance with scenarios featuring 5000+ units. Certainly vastly fewer of these complaints than the complaints there would be if there was some arbitrary limit on scenario size.


^ This exactly.

(in reply to 76mm)
Post #: 21
RE: Good improvement, but two small gripes - 11/21/2019 9:56:08 AM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
Take a look at some of the complaints about various issues over the years. You'll see a number of them are for scenarios over the 2000 unit mark. And by out of scope, I mean the infrastructure of the game. No single person in real life or in the simulation is expected to manage 5000+ units. The game wasn't built for that and turns it into a completely different game. Its completely out of scope expecting an operational-level sim to manage 5000+ units.

(in reply to KnightHawk75)
Post #: 22
RE: Good improvement, but two small gripes - 11/22/2019 10:08:11 PM   
apache85

 

Posts: 1632
Joined: 12/18/2014
From: Melbourne, Australia
Status: offline
Using the disable AI feature I mentioned above, I was able to JPFisher55's scenario to run considerably faster. In my testing after making the changes to static unit AI, I noticed a significant increase in simulation speed, from around 3x up to around 5x. A scenario this size is never going to run blazingly fast, but this is a good step forward with no tradeoffs.

There is a guide to using the feature here.

_____________________________


(in reply to thewood1)
Post #: 23
RE: Good improvement, but two small gripes - 11/22/2019 10:22:45 PM   
thewood1

 

Posts: 4410
Joined: 11/27/2005
Status: online
What are the implications for killing the AI?

(in reply to apache85)
Post #: 24
RE: Good improvement, but two small gripes - 11/22/2019 10:33:09 PM   
apache85

 

Posts: 1632
Joined: 12/18/2014
From: Melbourne, Australia
Status: offline
The currently implemented features affect DeterminePrimaryTarget and EvaluateTargets. Disabling them will mean that the relevant units (this is applied at unit level, not side or mission or group or anything else) don't look for targets and don't decide which target should be the primary.

In the case of disabling this for a hangar, the implications are zero apart from the performance improvement. In the case of disabling this for, e.g. a submarine on an ASuW patrol, it means that the submarine will not automatically look for and prioritise targets. Targets existing before this function is called are retained, and manual target allocation works as usual.

The actual intention of making this possible is to allow users to develop their own customised targeting priority through Lua. The performance enhancements are just a nifty side effect that can help out in scenarios that are heavy on static, weaponless units (the game already takes steps to reduce the calcs done by these units but as you can imagine there is only so much you can cull back before it will start to cause undesired behaviour when someone does something like add a missile launcher to a hangar).

_____________________________


(in reply to thewood1)
Post #: 25
RE: Good improvement, but two small gripes - 11/23/2019 12:37:37 AM   
KnightHawk75

 

Posts: 198
Joined: 11/15/2018
Status: offline
quote:

ORIGINAL: apache85

The currently implemented features affect DeterminePrimaryTarget and EvaluateTargets. Disabling them will mean that the relevant units (this is applied at unit level, not side or mission or group or anything else) don't look for targets and don't decide which target should be the primary.

In the case of disabling this for a hangar, the implications are zero apart from the performance improvement. In the case of disabling this for, e.g. a submarine on an ASuW patrol, it means that the submarine will not automatically look for and prioritise targets. Targets existing before this function is called are retained, and manual target allocation works as usual.

The actual intention of making this possible is to allow users to develop their own customised targeting priority through Lua. The performance enhancements are just a nifty side effect that can help out in scenarios that are heavy on static, weaponless units (the game already takes steps to reduce the calcs done by these units but as you can imagine there is only so much you can cull back before it will start to cause undesired behaviour when someone does something like add a missile launcher to a hangar).


Very helpful. Thank you.

" when someone does something like add a missile launcher to a hangar" ..and I've been known to do that :)

< Message edited by KnightHawk75 -- 11/23/2019 6:29:01 AM >

(in reply to apache85)
Post #: 26
RE: Good improvement, but two small gripes - 11/23/2019 4:04:27 AM   
LMychajluk

 

Posts: 106
Joined: 10/8/2017
Status: offline
@apache85 - Out of curiosity, was it ever considered for CMANO to support Nvidia cards, not so much for video acceleration, but for the Cuda Math libraries? It seems like there would be a huge number of small calculations that would need to happen within CMO each second (or cycle), and offloading some of those calculations to a secondary specialized processor (if available) for mathematical functions would make sense, especially when getting into the larger scenarios. I would also think most people that 'game' would have some iteration of a GTX or RTX card in their PCs, and I think most of the GTX-series or newer cards support the Cuda Libs, so I would wager that a significant portion of players could benefit from this (if, in fact, CMO can be boosted in this manner).

Just as an reference, running 'Don of a New Era' during a Tomahawk attack is using ~85-90% of all 8 cores on my 8th gen i7 at 1x speed, but the GTX GPU in my PC isn't even registering any usage. It would be nice to put it the GTX to work!

(in reply to KnightHawk75)
Post #: 27
Page:   [1]
All Forums >> [New Releases from Matrix Games] >> Command: Modern Operations series >> Good improvement, but two small gripes Page: [1]
Jump to:





New Messages No New Messages
Hot Topic w/ New Messages Hot Topic w/o New Messages
Locked w/ New Messages Locked w/o New Messages
 Post New Thread
 Reply to Message
 Post New Poll
 Submit Vote
 Delete My Own Post
 Delete My Own Thread
 Rate Posts


Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI

0.152