Tacview: Should I ?

[thewood1]

The above is probably the best point. If you already have it or want it for a flight simulator, it can be fun to play around with it on CMO. Otherwise, I struggle to see the long-term use.

[nukkxx5058]

I just opted for Kashmir. Just bought it. Will try it asap.
Thanks for your imputs guys !

[R_TEAM]

have it bought just after CMO .... from the website self .... its a nice thing to visualize the going-on in simply 3D GFX - if you put the effort in to DL the CMO terrain files, it even look more better ...
It is a musst have ...... no ..... but this is practicaly going for "all" GFX advancing in hardcore strategic/war-sim games - have played PanzerStrike on the C64 ... The full Harpoon pack on Amiga ... so the fun dont realy come from the GFX - but would you like play an wargame today with C64 GFX ?? .... i am not ....

Regards

[nukkxx5058]

ORIGINAL: R_TEAM

have it bought just after CMO .... from the website self .... its a nice thing to visualize the going-on in simply 3D GFX - if you put the effort in to DL the CMO terrain files, it even look more better ...
It is a musst have ...... no ..... but this is practicaly going for "all" GFX advancing in hardcore strategic/war-sim games - have played PanzerStrike on the C64 ... The full Harpoon pack on Amiga ... so the fun dont realy come from the GFX - but would you like play an wargame today with C64 GFX ?? .... i am not ....

Regards

I'm just afraid about the performance drag. This is why I decided to to but it for now (and the price of course. If it was 20 bucks I wouldn't hesitate)

You mentioned the CMO terrain files : Is it just for the 3D or does it has an impact on 2D satellite map too ? Like having it on local drive instead of the cloud to make loading times faster ?

[R_TEAM]

i dont get an performance slowdown - but have an 12 core xeon cpu - 16GB RAM - and an good GPU with 8GB VRAM ... so no problems on this side
The terrain files are "only" for TacView - to show the terrain in the same textures as the sentinel2 view in CMO .... it is not for, or help CMO in any way
(and actual it takes ~10min to load up the files properly over the default Tacview terrain textures part for part - so an good addition, but far from perfect ...)

Regards

[R_TEAM]

Oh - and an addition ....
CMO and TacView run better together, if you split the cpu cores self around the two ...
I have 12 Cores - TacView get 4 and CMO 8 ... (i know - overkill - but you get the idea)
Both run then even smother as before ...

Regards

[thewood1]

Outside splitting cores, if you run Op Brass Drum at full time compression, measure the game time compared to real time, then do the same with TacView running, the TacView run will be 10-20% slower. I have run this test on multiple PCs and the TacView run is measurably slower. Thats on a 20 core i9 and a then on an i7. 2080 on the first one and 1660 on the second.

Thats not saying that at lower time compressions, you might not consider it an issue, but there's no free lunch. The cycles have to come from somewhere.

[R_TEAM]

mhhh ....
i don't really think so, the CMO is designed to handle an high amount of cores (parralel threads) ... so i think the problem come here from the same i see ...
If you "not" distribute the cores self (windows is here not an good example of doing this "smart" ) , then both programms fight over the same cores and slow down the game.
would be interesting to see the same test with distribute a couple of cores (dont think TacView can handle self more than 4 threads...) to TacView and the other to CMO exclusively...
if then still a slow down, then it comes from the interprocess communication between CMO > TacView

Regards

[thewood1]

So run my test. Its fairly easy to do. Lets see what it does on your machine.

[boogabooga]

ORIGINAL: thewood1

Outside splitting cores, if you run Op Brass Drum at full time compression, measure the game time compared to real time, then do the same with TacView running, the TacView run will be 10-20% slower. I have run this test on multiple PCs and the TacView run is measurably slower. Thats on a 20 core i9 and a then on an i7. 2080 on the first one and 1660 on the second.

Thats not saying that at lower time compressions, you might not consider it an issue, but there's no free lunch. The cycles have to come from somewhere.

What exactly is the point of using Tacview in full time compression anyway? If anything, it's best used with 1:1 or very light time compression so that you can actually see what is going on. At 1:1, performance impact is not really noticeable, as far as I can tell.

[thewood1]

It was a test to show the impact of TacView. It was easier to test. My point is that there is an impact on using TacView. I thought that was obvious. Its especially noticeable with any time compression on a mid-range PC.

And thats not considering the size of the scenario. A large scenario where you are pushing processing limits also makes it more noticeable. As I have said before, you might not notice it with a good PC and a mid-sized scenario. But it is there.

[thewood1]

Just ran a benchmark using a run of a modded Op Brass Drum on the i9 with the 2080. This first pic is a 10 minute real time run with fastest time compression and no TacView. The pic needs a little explaining. I have CMO running with the stopwatch at 10 minute. CMO is now paused so I can calculate the time compression. Its about 86 game minutes to 1 real minute. The CPU load is between 18% and 20%. I also monitored the 10 seperate cores. They ran between 10% and 20% by eye.

[thewood1]

This pic is the same scenario and timing but with TacView running. The time compression goes from the 86 to 77. Over 15% drop in time compression. Not the higher CPU usage on the left of the charts of the Intel CPU monitoring. It goes from around 20% without TacView to around 30% CPU usage. About a 30% change in performance.

So even for a relatively top range machine, you have a pretty big performance hit. I tried these tests with a 1:5 time compression. At that compression, the TacView hit was barely noticeable. The PC had plenty of capacity to absorb the extra cycles needed. But there were extra cycles. With no TacView, CPU usage was 4-5%. With TacView, it was around 12%. Again, not noticeable on a good machine. But on a mid-range laptop or lower end desktop, TacView will most likely have a noticeable impact on even low time compression performance.

[thewood1]

btw, all of that above is on a mid-ranged 800 unit scenario. Get into anything above 3000 and I bet the performance hit gets noticeable faster.