Matrix Games Forums

My travel setup:

"Saw up to 46GB memory use which seems crazy"

There is something broken somewhere if you are getting that much RAM usage due to CMO. I think mine never went about 2-3 GB of CMO usage and 8GB overall.

Something a little different - Lots of RAM and lots of CPU cores (but poor single thread performance):
Saw RAM consumption peak at ~ 22GB during this.

[1567.1]
Dual Socket (2x processors) Intel Xeon E5-2640V4. Total: 20 cores, 40 threads.
96GB DDR4 2133MHz RAM
Highest:16533 ms
Lowest: 412 ms
Average: 3504.3 ms
Elapsed time: 10 Min 34 sec

13900K
P@5.7 E@4.5, DDR-5 7600 32-46-46-58-2T
HT ON

HT OFF

I am generally not very impressed with the performance of this game.
Does anyone have an idea why it sees only 16GB RAM when my system is 32GB

"I am generally not very impressed with the performance of this game."

And exactly what did you expect? I am curious what you are comparing it to as far as performance goes.

thewood1 wrote: Thu Mar 13, 2025 10:59 pm "I am generally not very impressed with the performance of this game."

And exactly what did you expect? I am curious what you are comparing it to as far as performance goes.

1. Fair question. I guess for a 5y old game run on a new CPU as mine, I would expect minimal slowdowns even in full acceleration. But maybe I am just wrong.
2. Is there an explanation for the benchmarking tool seeing only one of my 2 DDR5 16GB RAM components? My system saws 32GB RAM (2x16)

1) I think you are wrong. Did you even look at the scenario and what its doing? Its a pretty aggressive statement made in a vacuum.

Knightpawn wrote: Fri Mar 14, 2025 8:09 am 2. Is there an explanation for the benchmarking tool seeing only one of my 2 DDR5 16GB RAM components? My system saws 32GB RAM (2x16)

The relevant code makes certain assumptions on how the memory is reported to the OS, and it is possible that your config is a case that the code hasn't accounted for. This is a minor UI bug, considering this is purely for "reporting" purposes as part of posting benchmark results.

Note that this does NOT prevent Command from using more than 16GB if necessary, since RAM allocation is handled by the operating system. (Command, like most applications, makes a memory allocation request to the OS, and the OS will do its best to fulfill it, either by allocating raw RAM or resorting to virtual memory if necessary.)

Thank you for the response. Appreciated.

A benchmark first.

Steam Deck OLED


[1567.6]
Steam Deck OLED
16GB RAM / Shared Video Memory
Highest:57406 ms!
Lowest: 502 ms
Average: 7328.9 ms
Elapsed time: 21 mins 19 sec

Additional Information: Desktop Mode. Plugged in. Proton GE 9.13



Edit: Also tested in "game mode" with Proton GE 9.26. Similar performance. All CPU cores seem to peek at ~70% usage and temps of ~75 degrees C.

Test on the new 1638 Open Beta on Steam:

Ryzo90 wrote: Mon Mar 03, 2025 1:23 pm Something a little different - Lots of RAM and lots of CPU cores (but poor single thread performance):
Saw RAM consumption peak at ~ 22GB during this.

[1567.1]
Dual Socket (2x processors) Intel Xeon E5-2640V4. Total: 20 cores, 40 threads.
96GB DDR4 2133MHz RAM
Highest:16533 ms
Lowest: 412 ms
Average: 3504.3 ms
Elapsed time: 10 Min 34 sec

Performance in 1.08 Beta Build 1645 seems to have increased significantly:

[1.08 Beta 1645]
Highest: 10929 ms
Lowest: 314 ms
Average: 2279.2 ms
Elapsed Time: 7 min 7 sec

Test on 1645 beta

I concur, 1645 is notably faster than 1638 when Benchmarked.
Edit - just checked the spreadsheet, wow 1022.3 is currently the fastest time! I didn't expect that.

There are 2 posts below 1000 ms in the previous page, in what was probably a slower build.

Slower than previous built (greater variance in pulse times). This is one of three tests with similar results

Knightpawn wrote: Fri May 16, 2025 8:23 pm Slower than previous built (greater variance in pulse times). This is one of three tests with similar results

I am getting a few ms lower (faster) result, 974 vs 987 ms in 1645.

Was it using the Intel GPU? It happened to me

What annoys me is the higher use of CPU and GPU when the game is idle or playing or testing a not complicated scenario, and is worse now in the GPU compared to 1645.