Benchmarking CPU's in Command Modern Operations

[TempestII]

I still find 1662 to be a quick build by default:

20250518 - CMO 1662 RDR Benchmark 1.jpg (46.27 KiB) Viewed 1664 times

There is a notably but explainable speed drop off when more features are ticked:

20250518 - CMO 1662 RDR Benchmark 2.jpg (104.15 KiB) Viewed 1664 times

[Nikel]

The last is the slowest build (in the last beta series) in my laptop in the benchmark scenario, and back above 1000 ms.

CMO Benchmark 1674.png (18.01 KiB) Viewed 1465 times

For comparison:

Build 1645: 987 ms

Build 1662: 974 ms

[Nikel]

Waiting for the next patch with the high CPU/GPU usage fixed

1676.png (17.76 KiB) Viewed 1068 times

[Nikel]

Build 1706 improved in the benchmark scenario, but still not as good as 1645 and 1662 were (below 1000 ms here).

B1706.png (18.12 KiB) Viewed 974 times

[Nikel]

B1728.png (18 KiB) Viewed 293 times

[Nikel]

Test, adding this line in Command.ini

HighSimSpeedTimeSync = True

ADDED: Added HighSimSpeedTimeSync setting in Command.ini (Optional; Allows throttling of flame and double-flame sim speeds to stay in sync with clock time while running in the GUI, making them equivalent to 30x and 150x respectively)

B1728H.png (20.66 KiB) Viewed 256 times

No idea what it means, slower (better) result, but the simulation is slower.

Compared with the standard test (ms increase with time), the ms remain very stable along the test.

[Dimitris]

Test, adding this line in Command.ini

HighSimSpeedTimeSync = True

ADDED: Added HighSimSpeedTimeSync setting in Command.ini (Optional; Allows throttling of flame and double-flame sim speeds to stay in sync with clock time while running in the GUI, making them equivalent to 30x and 150x respectively)

B1728H.png



No idea what it means, slower (better) result, but the simulation is slower.

Compared with the standard test (ms increase with time), the ms remain very stable along the test.

This goes hand-in-hand with this change, as described in the B1728.2 release notes:

TWEAK: Tweaked the logic of sim-loop execution in order to more precisely match elapsed sim-time and wall-time when running in realtime (1x). This fixes a reported problem where e.g. a 1-minute of sim time could take 1m6sec of wall-time while running in real-time mode.

This is almost exclusively a "for pros" feature, it does not concern commercial players.

[Nikel]

OK, thanks for the explanation.

[Knightpawn]

HighSimSpeedTimeSync = True

ADDED: Added HighSimSpeedTimeSync setting in Command.ini (Optional; Allows throttling of flame and double-flame sim speeds to stay in sync with clock time while running in the GUI, making them equivalent to 30x and 150x respectively)

No idea what it means, slower (better) result, but the simulation is slower.

Compared with the standard test (ms increase with time), the ms remain very stable along the test.

I included it in the ini file and essentially there is no acceleration at all when flame or double flame is pressed.