CPU performance tests

Can be deleted.

I am a bit late to the game. But maybe someone will find it useful.

I am using a very old pc, i5 2500k, 12GB ddr3 (dual channel) and gtx 1060 6gb. Right now my ram is set to 1600 9-9-9 xmp profile. I tested stock cpu vs 4ghz vs 4.5ghz. I am now in the process of testing how memory latency vs frequency affects the game. I used the replay someone here posted, the one ending with teleport to kill SMD and nuke destroying paragon. I attached the replay incase someone else wants to use it. First number is lobby score (best of 5), second is time to finish the replay attached using the cheat code wld_RunWithTheWind. I start counting the moment i see ingame map, i stop counting the moment i see "replay ended" message. I record the replay using OBS and use it to calculate how much it took without having to be sitting and waiting for replay to end. I also have FPSMonitor overlay with statistics running for some useful stats. Here are my results:

I5 2500k@default 285 17:35s
i5 2500k@4Ghz 246 15:48s
i5 [email protected] 220 14:36s

1620037747247-8097180-rename-to-dot-fafreplay.zip

It's just time to build a new computer, i still mostly play RTS games

Is a 3D CPU from AMD the better choice or just go full raw power and core speed?

I think i will keep my 2080ti 1 or 2 years longer, still enough for 1440p

The 3d CPUs of AMD are extremely good at this game and can manage up to 10.000 units without slowing down the game. I can recommend the 7800x3d, or the 5800x3d. This game is primarily CPU limited; it runs fine on much weaker GPUs than the 2080ti

A work of art is never finished, merely abandoned

Thank you, that's what i was not sure about. In the past all you needed was IPC and fast RAM but with chips like the i9 lineup its like 300+ watts...

I think i'll give AMD a try, much lower TDP and this 3D cache sounds amazing

@jip said in CPU performance tests:

The 3d CPUs of AMD are extremely good at this game and can manage up to 10.000 units without slowing down the game. I can recommend the 7800x3d, or the 5800x3d. This game is primarily CPU limited; it runs fine on much weaker GPUs than the 2080ti

I was assuming the X3D parts would do well in FA, but hadn't seen much mention of them.

Do you have any idea how much of a boost an X3D part gives over it's non X3D version or the difference between a 5600x and 5800X3D would be?

I have no direct comparison, you can use any existing material that compares the two. As one example that resulted from a quick search (do turn off the sound):

You can relatively safely assume that the impact on the FPS in these games is identical in Supreme Commander. And specifically for Supreme Commander, it does not only apply to the framerate but also to the simulation. Which matters most when playing alone (against AI or a campaign, etc), as in a multiplayer game it runs as fast as the slowest CPU. But you can be certain that you're not the slowest CPU when you're rolling a 5600x3d.

Based on basic tests a 5600x3d, the 7800x3d and the 7950x3d can manage about 10k units in a 'regular scenario' without the game slowing down. Which is pretty impressive 🙂

A work of art is never finished, merely abandoned

@jip Much appreciated Jip.

I had watched a Hardware Unboxed video covering the X3D parts recently and they looked very impressive, was just curious if anyone had done FAF tests with them (in case something else would be a bottleneck). Looks like they're definitely the way to go for gaming.

I got a 10850K a few years ago at a good price, before the X3D chips were a thing, but kind of regret not going with the AM4 platform.

I have experience with the 7800x3d and currently have a 7950x3d. They work great 👍

A work of art is never finished, merely abandoned