Build 551 Still too slow..

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

Build 551 Still too slow..

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 >> Tech Support >> Build 551 Still too slow..

Page: [1]

Message

<< Older Topic Newer Topic >>

Build 551 Still too slow.. - 6/26/2014 10:03:39 PM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

I'm still seeing excessive pulse times, and major slow downs on even small scenarios like the one I just uploaded. Rolling back to B538... again.

Attachment (1)

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

Post #: 1

RE: Build 551 Still too slow.. - 6/27/2014 1:16:48 AM

chesmart

Posts: 908
Joined: 1/14/2007
From: Malta
Status: offline

Me i am using 536 because its faster. 551 totally stops in my case

(in reply to NakedWeasel)

Post #: 2

RE: Build 551 Still too slow.. - 6/27/2014 1:18:28 AM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

Ok guys we'll take a look. If you can grab some saves so we can better investigate what might be causing it.

Mike

< Message edited by mikmyk -- 6/27/2014 2:20:03 AM >

_____________________________

(in reply to chesmart)

Post #: 3

RE: Build 551 Still too slow.. - 6/27/2014 1:32:06 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

I appreciate all that you do for this game, and this community, Mike. Way to support your product and your customers.

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to mikmykWS)

Post #: 4

RE: Build 551 Still too slow.. - 6/27/2014 2:05:23 AM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

No prob. Learned a lot from our harpoon days which we hope is reflected today.

Mike

_____________________________

(in reply to NakedWeasel)

Post #: 5

RE: Build 551 Still too slow.. - 6/29/2014 1:48:45 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

Well, After extensive playtesting, I must say that I am suitably impressed with B552's performance. All that said, I am still having major slowdowns with my Eagle Vs Bear scenario, the most current version posted above. I've also created a "clean Version, with reduced units, no-navigation zones, events, etc, and still seeing major slowdowns that make it almost choke and die. Can someone tell me what might be causing the above scenario to hang, because I really want to play that version of it, and I cant really get halfway through it without having to close Command and restart it a dozen times. Super frustrating!

Attachment (1)

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to mikmykWS)

Post #: 6

RE: Build 551 Still too slow.. - 6/29/2014 1:50:24 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

I'm also including the "Clean" version, which also bogs down.

Attachment (1)

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to NakedWeasel)

Post #: 7

RE: Build 551 Still too slow.. - 6/29/2014 2:39:50 AM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

If you can grab a save of when this happens this would be most helpful.

BTW I do think some of it is the navigator and the russian aircraft trying to navigate around Kola. I got the best results out of Duelists by getting rid of the sharp edges.

Thanks!

Mike

< Message edited by mikmyk -- 6/29/2014 3:41:14 AM >

_____________________________

(in reply to NakedWeasel)

Post #: 8

RE: Build 551 Still too slow.. - 6/29/2014 3:09:17 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

In the clean version, I've completely removed the no-nav zones. The slowdown occurs about 2-3 mins after the beginning. I'll go ahead and run it, so I can get a save.

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to mikmykWS)

Post #: 9

RE: Build 551 Still too slow.. - 6/29/2014 3:34:27 AM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

One of our internal beta testers may have caught something on it as well.

Yeah I can stress enough how valuable it is to grab a save when you see stuff.

Thanks!

Mike

_____________________________

(in reply to NakedWeasel)

Post #: 10

RE: Build 551 Still too slow.. - 6/29/2014 3:46:06 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

OK, not sure what happened there, did not post. Anyway, clean version slowdown, begins within minutes of game start.

Unpaused it after making this post and the latency settled down. Ran it at 30X compression, and continued to see double digit pulses for a couple minutes, and then it started spiking again. This time up to about 5500+MS. Also, even when/if it settles back down to double digits, game stutters. (ticking movement, and stuttery sound FX)
This will not stop, and if I run any other scenarios, their performance also takes a hit. I have to stop the program and restart it, to get back to decent performance on any scenarios after that.

Reducing time to compression down to 1:1, and even setting for pulsed time does not help. The game is using about 18% of available CPU power at this point, but at slowdown peak, can jump up to 75+% CPU usage. CMANO consistently uses about 750-800 MBs of available memory. I run the game with no other programs active, and kill all but the most required processes in the background.

Attachment (1)

< Message edited by NakedWeasel -- 6/29/2014 5:00:54 AM >

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to mikmykWS)

Post #: 11

RE: Build 551 Still too slow.. - 6/29/2014 4:05:38 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

With everything but the basic structure of the scenario deleted, this is still 100% reproducible. The only things that could be causing this problem must be the unit pathing, or the jamming, or something in the B552 DB. I cant play test it with the earlier DBs, because when I load it up, no units for either side load into the game. Basically I get an empty canvas.

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to NakedWeasel)

Post #: 12

RE: Build 551 Still too slow.. - 6/29/2014 5:03:41 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

Now, then I ran the "standard" version of Eagle vs Bear, with the nav zones, KGB ships, and satellites, etc. I turned off all the jamming. I think the lack of jamming certainly helps, but, my fighters tried to intercept the first satellite that cam near, and that produced a spike that started at 25000MS, and the next pulse went to 55000+MS, and the third pulse went up to 95000MS!!! it did settle back down to triple digits after that, but the damage was done, and the "ticking/pulsing" gameplay continued thereafter. Disgusted, killed the program, and restarted the "standard version" again. This time I eliminated the satellites. It ran fine for about the first 2-3 minutes, and then the spike began, this time not going much higher than 1000-3000MS and would then settle down to double digits for a few seconds, then spike to 300-500MS. The ticking was going on, but less pronounced. I still wasn't happy about the game performance at this point. I noticed the CSG's pathing was a dotted line at this point, running the border of the Scandinavian No-Nav zone I had set up, ending in the Patrol zone it was supposed to reach.

So in conclusion, I believe it's a combination of at least three factors: 1, Navigation is a problem. 2, Jamming is a problem. 3, Satellites are a problem.

Attachment (1)

< Message edited by NakedWeasel -- 6/29/2014 6:04:49 AM >

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to NakedWeasel)

Post #: 13

RE: Build 551 Still too slow.. - 6/29/2014 5:37:46 AM

Maromak

Posts: 1030
Joined: 12/26/2007
From: Australia
Status: offline

quote:

ORIGINAL: NakedWeasel

So in conclusion, I believe it's a combination of at least three factors: 1, Navigation is a problem. 2, Jamming is a problem. 3, Satellites are a problem.

I concur with your second point. I was playing Red Flag 2017 and the speed took a huge hit when a large number of MALDs were fired. Once the MALDs were destroyed etc, the speed returned to normal.

< Message edited by Maromak -- 6/29/2014 6:38:17 AM >

_____________________________

Certa Cito

(in reply to NakedWeasel)

Post #: 14

RE: Build 551 Still too slow.. - 6/29/2014 6:29:51 AM

NakedWeasel

Posts: 500
Joined: 1/14/2014
Status: offline

I am now convinced that the biggest impairment in this scenario is unit pathing and navigation, by far. Playing the Russian side, and watching them try to intercept the CSG when they detect it is causing the processor all kinds of grief.

_____________________________

Though surrounded by a great number of enemies
View them as a single foe
And so fight on!

(in reply to Maromak)

Post #: 15

RE: Build 551 Still too slow.. - 6/29/2014 10:24:30 AM

Dimitris

Posts: 13282
Joined: 7/31/2005
Status: offline

Yes, pathfinding in general is a CPU killer and above the arctic circle doubly so (because of the necessary ice checks).

This is why we regularly recommend to scen authors that they use pathfinding only when necessary, and otherwise pre-plot courses for units that don't need it.

< Message edited by Sunburn -- 6/29/2014 11:24:42 AM >

_____________________________

Facebook
Twitter

(in reply to NakedWeasel)

Post #: 16

RE: Build 551 Still too slow.. - 6/29/2014 12:17:41 PM

thewood1

Posts: 6529
Joined: 11/27/2005
Status: offline

I'll bring my point back that I don't see this in scenarios on 536 and earlier. I see the spikes on all scenarios, its just that it isn't very noticeable if your regular pules are less than 10ms. The spikes are anywhere from 5-20 times the average pulse time and happen every 5 or 6 regular pules. The game is doing something every 5-6 pulses that is causing the spike.

(in reply to Dimitris)

Post #: 17

RE: Build 551 Still too slow.. - 6/29/2014 2:07:49 PM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

Weasel some things that should help the performance of your scenario.

1)Your KGB merchant fleet has a patrol zone that crosses many miles into the ice field and isn't constrained. So the navigator is going absolutely bonkers replotting. Move the western points eastward out of the ice.

2)Same KGB fleet doesn't have a prosecution zone defined. They're on a sea control mission with no prosecution zone defined. This means that any surface and subsurface contacts will be chased anywhere on the map. Given the us CVBG is behind island its absolutely certain that an initial detection will be made by the sats. The 50 or so fleet will then all begin trying to plot a path through Iceland.

3)US CVGB: Some of your CAP patrol zones are initially defined inside the Iceland no fly zone. Navigator goes bonkers.

4) Russian Missions: Strike mission are great but the US CVBG is behind island. This forces them to have to plot around it.

Hope this helps.

Mike

_____________________________

(in reply to thewood1)

Post #: 18

RE: Build 551 Still too slow.. - 6/29/2014 2:11:03 PM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

quote:

ORIGINAL: thewood1

I'll bring my point back that I don't see this in scenarios on 536 and earlier. I see the spikes on all scenarios, its just that it isn't very noticeable if your regular pules are less than 10ms. The spikes are anywhere from 5-20 times the average pulse time and happen every 5 or 6 regular pules. The game is doing something every 5-6 pulses that is causing the spike.

Your point may be part of the issue in this case but we've got to eliminate all the other things that could be causing the slowdown to get to these. We're going to address the navigator issues which can be fixed on the user side. We'll then have a little less cloudy scenario to look for all those other things you think is going on

Thanks!

Mike

< Message edited by mikmyk -- 6/29/2014 3:11:26 PM >

_____________________________

(in reply to thewood1)

Post #: 19

RE: Build 551 Still too slow.. - 6/29/2014 3:11:54 PM

thewood1

Posts: 6529
Joined: 11/27/2005
Status: offline

One part of a solution on the navigator side might be to let a mission designer set a form up point for a mission. In the Eagle vs Bear example, let a form up or organization point be set above the peninsula. That gives the designer the ability to clear any major obstacles before the mission navigator kicks in.

(in reply to mikmykWS)

Post #: 20

RE: Build 551 Still too slow.. - 6/29/2014 8:17:31 PM

Dimitris

Posts: 13282
Joined: 7/31/2005
Status: offline

The every-5th-pulse spike is known and is part of the anti-crash measures.

_____________________________

Facebook
Twitter

(in reply to thewood1)

Post #: 21

RE: Build 551 Still too slow.. - 6/29/2014 9:36:42 PM

thewood1

Posts: 6529
Joined: 11/27/2005
Status: offline

So I assume that means we have to live with it for now. I just want to know so I can decide how I want to upgrade.

(in reply to Dimitris)

Post #: 22

RE: Build 551 Still too slow.. - 6/29/2014 9:49:49 PM

mikmykWS

Posts: 11524
Joined: 3/22/2005
Status: offline

Yes. The consensus we've gotten from our beta folks and our posters here is that its ok to go forward. As always we'll continue to work on speed issues etc.

Thanks!

Mike

_____________________________

(in reply to thewood1)

Post #: 23