Matrix Games Forums

ORIGINAL: dplummer

Gave it a shot and single ship at a time I could get to work in around 10-15 seconds, whole fleet gave up after about 4 minutes.

I am seeing lags for single land units of around 2-4 seconds each move.

I would like to have a saved game of this (zip the GAM file and attach it to a post in this thread). I've solved the problem that Zartacla was having with time delay when returning naval units to base. I would like to test those changes using your saved game to see if they solve your problem.

ORIGINAL: Zartacla

Completely broken now. Try to return the Italian fleet in the eastern med to base.

I've made some changes over the past couple of days which appear to have fixed this problem. Using your saved game, I was able to return naval units to base individually or as groups with the response time under 10 milliseconds (I have a debug feature that reports exactly how much time is spent calculating supply). I returned all the units for both sides without difficulty.

ORIGINAL: Zartacla

I back tracked and didn't collapse Vichy. The lag time dropped substantially, but is still noticeably much worse than pre patch. Additionally, the Bermuda triangle bug has become worse, affecting units in parts of China where it was not previously a problem.

I fixed the problem earlier this week with the supply bug referred to as "the Bermuda Triangle".

The reason version 1.1.5.0 takes longer to calculate supply than 1.1.3.0 is because it is now completing the search for overseas supply paths for cooperating major power. That bug (fixed in 1.1.5.0) had terminated the search prematurely, leaving units OOS that should have been in supply. The downside is that now that the search algorithm is working 'correctly', it takes too long.

So far I have cut the response time in half for moving land units. I'm still working on other efficiencies to reduce it more: by a factor of 10 should be enough.

Okay, I have fixed the response time when moving units on the map. Without going into the details, moving an HQ now typically takes less than a second (to recalculate supply). Moving other land units takes less than 10 milliseconds. There will be exceptions in some circumstances when control of hexes changes: new supply paths may be available for the moving side and/or old supply paths invalid for the non-phasing side. Indeed, there are even situations where the reverse may be true.

I haven't had the time to run a lot of tests on these modifications so I am going to wait another 24 hours before making a hot patch available. That will give both myself and the beta testers some time for additional testing.

I did run tests using the saved games from Zartacla and the numbers I quoted above apply to his saved game post-creation of Vichy France.

Sorry, meant that I took Z's saved file and tried it on my machine and that was what I experienced.

I backed mine to 1.1.3 once things went south and don't have any 1.1.5 saves.

ORIGINAL: Shannon V. OKeets

Okay, I have fixed the response time when moving units on the map. Without going into the details, moving an HQ now typically takes less than a second (to recalculate supply). Moving other land units takes less than 10 milliseconds. There will be exceptions in some circumstances when control of hexes changes: new supply paths may be available for the moving side and/or old supply paths invalid for the non-phasing side. Indeed, there are even situations where the reverse may be true.

I haven't had the time to run a lot of tests on these modifications so I am going to wait another 24 hours before making a hot patch available. That will give both myself and the beta testers some time for additional testing.

I did run tests using the saved games from Zartacla and the numbers I quoted above apply to his saved game post-creation of Vichy France.

It wasn't the creation of Vichy - it was the collapse of Vichy. I had assumed that was the poison pill that caused all of my moves to take so much longer both because of the timing and because the first time I tried to collapse Vichy, the game spit out a supply related error message then crashed. It was not reproducible - reverting to a save and collapsing Vichy again triggered the long calc time, but without the crash. The following impulse, I DOW'd Russia, and when I started moving units, the lag time was unbearable.

When I went back and chose not to collapse Vichy, the end of turn roll was different. The turn ended immediately, before I could DOW Russia. The next several turns were winter turns, with no break in the weather, so Russia was not attacked. Movement was fine, but I was still connecting that to the fact that I had not collapsed Vichy. When spring rolled around and I DOW'd Russia, the lag kicked back in. So it would seem that the problem was Russia all along, and Vichy was a red herring.

Guys, what would you say if the performance for HQ's is OK again, but the Bermuda Triangle bug moved north? In case the next release is like that, would you still prefer it to come out now? Or is the disappointment regarding the Triangle bug too much, and should it wait? The save/load workaround would still work. This is a theoretical question as work is still being done on this release. I presume the performance improvements would be getting much higher prio from you?

ORIGINAL: AxelNL

Guys, what would you say if the performance for HQ's is OK again, but the Bermuda Triangle bug moved north? In case the next release is like that, would you still prefer it to come out now? Or is the disappointment regarding the Triangle bug too much, and should it wait? The save/load workaround would still work. This is a theoretical question as work is still being done on this release. I presume the performance improvements would be getting much higher prio from you?

warspite1

How far north? Can you move it to the Arctic? That would work [;)]

ORIGINAL: warspite1

ORIGINAL: AxelNL

Guys, what would you say if the performance for HQ's is OK again, but the Bermuda Triangle bug moved north? In case the next release is like that, would you still prefer it to come out now? Or is the disappointment regarding the Triangle bug too much, and should it wait? The save/load workaround would still work. This is a theoretical question as work is still being done on this release. I presume the performance improvements would be getting much higher prio from you?

warspite1

How far north? Can you move it to the Arctic? That would work [;)]

Parts of North China, but also parts of South America. The latter is not of interest, but the first is not nice. Manchuria also still shows effects of this, but that was earlier as well. Again: Save/load works and Steve is still working on this.
I think we have a combination of bugs here, Steve has fixed one, but there are still others left. the Bermuda Triangle bug might have 9 lives, and I don't know how many there are left.

the lag time makes the game unplayable after the stat of Barbarossa, so I'm in favor of releasing a hotfix for that three days ago. As it stands now, it's taken far too long to put out a fix for that. Other bugs are a separate issue.

If the triangle bug moves, you're still going to hear about it from me. If it moves to somewhere that is a more active area of the map, you'll hear even more from me. If it moves to a less active part of the map, then I think it's a no brainer.

For example, if moving it north means that half of Russia is affected, then leave it be until you've killed it completely. If moving north means Sian/Lanchow is affected, then there's probably not much difference for us than the current issue.

It did not move - it became more dominant due to the fixes to improve the supply calculations (these look fine, BTW). Steve has killed another of its 9 lives today, and we are expecting the next beta soon.

ORIGINAL: Zartacla

the lag time makes the game unplayable after the stat of Barbarossa, so I'm in favor of releasing a hotfix for that three days ago. As it stands now, it's taken far too long to put out a fix for that. Other bugs are a separate issue.

If the triangle bug moves, you're still going to hear about it from me. If it moves to somewhere that is a more active area of the map, you'll hear even more from me. If it moves to a less active part of the map, then I think it's a no brainer.

For example, if moving it north means that half of Russia is affected, then leave it be until you've killed it completely. If moving north means Sian/Lanchow is affected, then there's probably not much difference for us than the current issue.

I have the response time for moving units better than it has ever been.

There are still some bugs for recalculating supply that I want to fix (the beta testers' advice is to fix those before releasing a new version).

I'll either do Hot Patch today (Sunday) or Matrix will do a new Update tomorrow (Monday) - if they can find room in their schedule.

As for the myth of the Bermuda Triangle, it is just that, a myth. An overseas supply bug was due to not propagating improvements throughout a previously explored path.

The details are: after finding a path to a port that required 4 Basic Path Hexes, an improved path requiring 3 Basic Path Hexes was not carried forward to the port itself. The search algorithm then concluded that overseas supply was impossible since 1 BPH was needed to go overseas. If the shorter path was found first, then all was well. The failure only occurred when the search first found a round-about path to the ports. I fixed this with two corrections: (1) improved paths are now propagated correctly, and (2) the search algorithm finds the shorter path first.

Does this concern the bug where you cannot leave rails overseas in rain? Or the Checkiang rail line? And the unit duplication in the F8 window? And the cases where HQ trace to themselve?


I am unclear how paths can improve now that you find shorter paths first What speaks against Dijkstra's? Most routing protocols use it and you should have only positive weight cycles.

ORIGINAL: Dabrion

Does this concern the bug where you cannot leave rails overseas in rain? Or the Checkiang rail line? And the unit duplication in the F8 window? And the cases where HQ trace to themselve?


I am unclear how paths can improve now that you find shorter paths first What speaks against Dijkstra's? Most routing protocols use it and you should have only positive weight cycles.

That is pure paranoia on my part. Belt and suspenders.

I will bump this to the top.

do you consumers still have this issue

using the latest public beta - I assume?

It was better for a while, now it's pretty bad again (although not as bad as it was when I first posted this). Move times inside Russia after the war starts can be pretty bad. RTB times for fleets are randomly bad - from some sea zones, there's no lag time at all, but from others it can be pretty rough. For example, axis subs in Scapa Floe take a while for me to RTB. Italian ships in the Eastern Med also.

RTB is bad at my side as well as described, the bigger the stack the longer it takes. But for something happening only at the end of the turn it is less important than during land moves. The picking up time is the first on my list to be tackled (again).

It doesn't bother me. However, I'm used to this during beta testing. I would say that it isn't important to fix, since it is dangerous to do so (the last time we ran into an awful lot of regression bugs, which resulted in the very long time this clearing of supply bugs has taken, and still is taking). So lets wait until the final supply patch is made and than ask everyone here if this should be improved now, or never.

Is the unplayable lag time still a live issue? I have been experiencing this for a while, mainly around land movement and production, got really bad after Russia in the war. Just updated to latest patch and at first glance it doesnt seem any better. I know my laptop isnt the fastest, so could it be memory or processor?