Benchmarking CPU's in Command Modern Operations

[thewood1]

Don't base how the game runs based on these benchmarks. But I would say that expecting a six year old laptop CPU to match what you are seeing in this thread is probably not right.

[DmitriyBlade]

+ GeForce 1050TI

I don't know why it shows only 8gb because I have 16. Maybe that is the reason why the game runs very bad?

Capture1.JPG

Might be that contact DDR with laptop mainboard insecure?
Had that issue while upgrading memory on my lap

[thewood1]

The half RAM thing was a common issue with the CMO benchmarking tool for a while. It went away when I bought a new laptop. I am 99% sure it a CMO issue and not your RAM.

[thewood1]

I had noticed CMO running fairly sluggish recently. I especially noted it since 1328.12 was dropped as a beta.

Here is my last run with 1328.10a...

Screenshot 2023-10-18 200425 (1).jpg (98.44 KiB) Viewed 2934 times

Note the 1.5 accel.

This is 1328.12 beta...

Screenshot 2023-11-22 203223.jpg (68.71 KiB) Viewed 2934 times

I see this sluggishness showing up in non-benchmark scenarios as well. I'll post this in Tech Support. Just wondering if anyone else is seeing this.

[schweggy]

Same here. To me the sim seemed to stutter off and on regardless of the scenario size. My results on the benchmark are similar to yours. And what's up with the lowest pulse time?

v1.06-Build 1328.10a

Screenshot 2023-11-03 140945.jpg (80.32 KiB) Viewed 2918 times

v1.06-Build 1328.12

Screenshot 2023-11-23 085936.jpg (82.28 KiB) Viewed 2918 times

[thewood1]

This is my 11 run. Devs responded in the tech support thread on this issue.

Screenshot 2023-11-22 212749.jpg (67.61 KiB) Viewed 2917 times

[Dimitris]

We have been fixing a lot of issues related to time acceleration lately (in addition to new features) and performance has suffered in some cases as a result. We've been steadily clawing back the losses through "engineering" improvements, but it's going to be a long road. Sorry!

[thewood1]

I understand that time compression isn't as compressed as more complex features are added. But the drop from 1.5x to .5x, along with a weird "lowest pulse" is a little disconcerting. Something strange happened between .11 and .12 that doesn't feel right in the benchmark or in game play. I have a 1000 unit scenario that stutters at any acceleration over 5x. it went from about 25x at double flame to about 5x. Its going to really hamper scenario development.

[Dimitris]

I understand that time compression isn't as compressed as more complex features are added. But the drop from 1.5x to .5x, along with a weird "lowest pulse" is a little disconcerting. Something strange happened between .11 and .12 that doesn't feel right in the benchmark or in game play. I have a 1000 unit scenario that stutters at any acceleration over 5x. it went from about 25x at double flame to about 5x. Its going to really hamper scenario development.

Any chance we can get a copy of that?

[thewood1]

Sure. I just updated it to db 503. Just switch to double flame and let it run through a couple game hours. There is significant missile activity that cause initial slow downs. But it calms down a lot after the first game hour. You'll note the surging and lagging of the execution regardless of activity.

[thewood1]

Also want to mention that at one point, CMO used all my cores in my various laptops. I notice in recent updates, it seems to only use half my P-cores and a small number of e-cores. I don't know if that changed with tiny/64 bit update or after a few more updates. I have a backup laptop with an older i9 that shows the same basic performance.

Laptop spec...19-13980HX, 4090 laptop. at speed, it draw 330W.

Here is a cap of performance management screen on my PC while the above scenario is running.

Screenshot 2023-11-24 184652.jpg (941.02 KiB) Viewed 2837 times

You can see its not using all the available cores. I understand there is a lot going on with Win11 and how it manages cores, but CMO is the only 3rd party app running and its definitely running differently than it did before. I have the PC on a high performance plan.

edit: Capped a better example of high pulse times, not a lot of activity, and only a few cores being pushed. It cycles through pulses of 40-80ms and then goes to 300-800ms in spurts. It makes it look very stuttery.

[Dimitris]

One thing that jumped out at me on that scenario is that the "Terrain type effects - Advanced" scenario feature was enabled. We strongly recommend disabling this if performance is a concern, as it really impacts simulation execution (there are other reasons not to prefer it, too; for example non-masted radars placed in the middle of forests will be blinded).

[thewood1]

Turned off the Advanced terrain effect and did a couple more run throughs at double-flame. Slightly better at about 8:1 actual compression, compared to 7:1 with that feature on. I called it laggy and that might be a misnomer. Its more like it surges for a few seconds (4-5) at a high acceleration rate then pulse times suddenly spike for anywhere from a few seconds to 20 seconds and slow everything down for that period.

[Dimitris]

We are adding some more speed tweaks for the next update release, so it will be interesting to compare when it's out.

[flex0r]

Turned off the Advanced terrain effect and did a couple more run throughs at double-flame. Slightly better at about 8:1 actual compression, compared to 7:1 with that feature on. I called it laggy and that might be a misnomer. Its more like it surges for a few seconds (4-5) at a high acceleration rate then pulse times suddenly spike for anywhere from a few seconds to 20 seconds and slow everything down for that period.

Where can this option be found? I want to try disable it and see if there is any improvement.

[blu3s]

In editor mode, load the scenario and go to Editor -> Scenario Features + Settings

[TempestII]

I've definitely noticed a deterioration in performance since downloading 1328.13. Benchmarking kind of shows this - my Average and Lowest pulse times are the best that I've seen, but the Highest is certainly the highest I've experienced. The Elapsed Time and Max Time Acceleration are about 30-40% of what it was at 1328.11. The game feels slower and more sluggish too, with 1000-1200 AU scens now running about the same speed as 2000-2500 AU scens did in previous builds.

20240121 - CMO 64 Bit RDR Benchmark.jpg (46.93 KiB) Viewed 2534 times

[Dimitris]

I've definitely noticed a deterioration in performance since downloading 1328.13. Benchmarking kind of shows this - my Average and Lowest pulse times are the best that I've seen, but the Highest is certainly the highest I've experienced. The Elapsed Time and Max Time Acceleration are about 30-40% of what it was at 1328.11. The game feels slower and more sluggish too, with 1000-1200 AU scens now running about the same speed as 2000-2500 AU scens did in previous builds.

Can we please get specific before-and-after examples (scen saves) of this performance slowdown on a new thread on Tech Support? (Referring to "real" scenarios, not the benchmark one)

As you may guess, it is supremely frustrating for us to read about performance issues without actually being able to apples-to-apples observe them and debug them.

Thanks.

[thewood1]

I think one was posted with limited response. I'll see if I can find it. But you can look at my exact before and after here...

https://www.matrixgames.com/forums/view ... 7#p5135737

I copied that exact post into tech support, but most of the responses were in this thread. I also created a thread with saves for my testing scenario in tech support and got no response on it. The responses came in this thread. As to the Benchmarking scenario, you won't get saves from the benchmarking process.

Claification: 1328.13 runs a lot better from a quick superficial viewing. But the bug in prosecution zones has de-motivated me from testing any further.

[thewood1]

Performance on 1328.12 has been report by a couple people in the Tech Support forum with minimal response from anyone on the dev side. This one stands out...

https://www.matrixgames.com/forums/view ... e#p5141109