Bad Weather = Bad FPS? I ran some tests to find out!

Hi all, I'm not sure if this has been posted before, but I recently wanted to know how much bad weather and/or time-of-day would impact the performance on my system, so I ran some informal tests and took notes.

I made a video about it and posted it on YT, but self-promotion isn't permitted here, so I just post my final findings here instead in case anyone is curious:

Note: I play AMS2 on a 65" Hisense U8G 4K TV which maxes out at 120Hz, so I normally cap my FPS at 117 via NVIDIA Control Panel. That said, I tend to avoid thunderstorms because not only is it taxing on my system, it's also a real PITA to drive in (natch).

These are my current settings:



My PC Specs:

• CPU: AMD Ryzen 7 7700X (ECO mode, PBO enabled, CO -15 all cores)
• Video Card: NVIDIA Founders Edition GeForce RTX 4080 16GB (100% stock, no overclocking, latest drivers at time of publishing)
• Liquid CPU Cooler: Lian Li GALAHAD AIO 360 RGB UNI FAN SL120 EDITION
• Motherboard: Asus ROG STRIX X670E-A GAMING WIFI ATX AM5 (BIOS 3205)
• Memory: G.Skill Trident Z5 NEO RGB Series 32GB (2x16GB) 288-Pin SDRAM DDR5 6000 CL30-38-38-96 1.35V Dual Channel (EXPO II profile)
• Storage: Samsung 990 Pro 2TB m.2 2280 PCIe 4.0x4 NVME SSD
• Case: Lian Li O11 Dynamic EVO ATX Mid Tower (White)
• Power Supply: MSI A1000G PCIE5 1000 Watt 80+ Gold Certified Fully Modular ATX 3.0
• Case Fans: 10 x Lian Li Uni Fan SL-Infinity 61.3 CFM 120mm
• OS: Windows 11 (latest build and patches

 

Thank you for taking the time to run these tests and for sharing the info. I've been wondering if upgrading my hardware could let me muscle my way out of the frame drops I get on my own system. I also run on a 4K120 panel (driver capped at 116Hz, in the event a game uses NV Reflex) so this information is right up my alley. I don't really drive in Thunderstorm conditions as I cannot keep up with the AI, but I am still having a tough time with dry day/night transitions despite having a semi-beefy mid-range system (12900k/RTX 3080).

I wonder if the Madness Engine is as friendly towards the modern AMD 3D Cache CPUs as some other titles (ie ACC).

 

Does anyone have any information on how CPU-bound the game might be? I use triple screens (2560x1440), have a RTX 4080 Super, but get barely above 60fps. Which I should be happy with, but then if it is wet or there are lots of cars it slows...

so perhaps my Ryzen 3900x is the problem?

 

Hard to say for certain, but I'm running a Ryzen 7600x with a RTX 4070 super at 3440x1440 with everything maxed out and daylight racing is 150+ FPS.

60 seems extremely low though, even for CPU limiting that doesn't seem right, the RTX should be doing the vast majority of the heavy lifting.

.... I've just said all that, but just realised you're running triple screens. That does put a significantly higher load on the GPU, so it wouldn't surprise me if you're getting to the point where the CPU is running out of capacity as well. Hopefully others with triple screen setups can share their experience

 

While it might not be the primary issue with AMS 2 with your current monitor setup, that CPU is old enough to hold back a 4080 in game that are known to be CPU hogs (Spider-Man/Miles Morales, Hogwarts, etc). Running triples always complicates things a bit because you add so many more pixels to the rendering load versus a single screen (which usually leaves you GPU-limited).

This benchmark is one of the most applicable I could find, featuring a Ryzen 3000 series CPU matched against the newer 3D Cache 5800x3D in ACC (not AMS2, but in the same ballpark). 4090 is the GPU tested here.

Finally, a different benchmark showing the same game with the newest CPUs on the market with the 4090 as well.


ACC actually scales very well on newer CPU designs (as each additional AI car adds some CPU load). Again, this isn't AMS 2, but gives you some insight into how your 3900 Ryzen may be holding back your 4080 in CPU-limited scenarios.

 

This is amazing. Thank you so much for replying like this, it is really appreciated. Of course it's not just for me, this is 'research' on the topic, but I am very grateful nonetheless. I was not expecting the CPU to cause so much of a difference. I am seriously thinking about upgrading, in that case.

 

The easiest upgrade in your situation would be the Ryzen 5800X3D as it is compatible with your current motherboard and memory. Sadly, that CPU went out of production more than a year ago and is only available through the secondhand market. The easy recommendation is the 9800X3D, but it usually sells for close to $500 in the States and would require you to upgrade to an AM5 socket motherboard and DDR5 memory. That being said, such an swap would leave you on a strong upgrade path once your 4080 begins to show its age. If PC gamers get lucky in a few years, NVIDIA or AMD will launch GPUs at the 5090 performance tier that could be acquired for less than $1000 (not holding my breath, though). Loads of 9800X3D benchmarks out in the wild with both a 4090 and 5090 GPU that can give you a strong idea of what future upgrades could provide on the performance front.

With that being said, no video game is perfect and AMS 2's Madness Engine is not particularly known for CPU scaling. The most realistic scenario you'd see moving from a 3900X to a modern X3D Ryzen CPU would be a middling boost to average fps and massive gains to the 1% low frames. Your frametime graph would be much smoother and the baseline driving experience will feel better with less random fps spikes. The CPU upgrade will not improve things when it comes to the AMS 2 particle system and rain/spray. Each individual particle of spray is typically made from drawing and blending multiple alpha effects together and most modern game engines just aren't able to deal out that load to CPUs/GPUs in a cleanly parallelized fashion (where hardware with more and faster cores cannot be leveraged). You should see big gains when it comes to AI car count and possibly AI pit crews (though enabling AI pit crews seems to be problematic for the Madness Engine beyond simply demanding more hardware resources - the game simply stutters when too many pitcrews are being drawn).

At the end of the day, your monitor setup requires a 7680x1440 rendering resolution. This comes out to more than 11 million pixels per frame rendered. 4K monitors only display a bit more than 8 million pixels per frame. When looking at data, understand that results you see for 4K will need to have ~30-40% more overhead factored in to be comparable for your particular setup. Additionally, you could pray at the altar of Reiza and hope that AMS 2 gets a major rendering upgrade down the line that allows for upscaling technologies like FSR or DLSS to ease the burden on your GPU. Some games also offer a customizable resolution scaling slider that forces the game to internally render at lower resolutions before upscaling, but there is no real evidence that such a solution would ever come to AMS 2.

Hopefully some of this rambling helps. In the short term, try lowering Grass Detail, Enhanced Mirrors, Particle Density, and Particle Detail to the lowest setting. Also, only enable pit crews for your own car to avoid frame spikes or dips as you circle around near the pit straight on circuits. The rest of the graphics settings (save for car count) mostly hammer the GPU and can be kept as high as you like on a 4080 (so long as you have VRAM to spare).

 

[QUOTE="so perhaps my Ryzen 3900x is the problem?[/QUOTE]

CrapsJarrard already gave an amazing response above. But to give some perspective on my end, in my test video, my Ryzen 7 7700X hovers around 40% in any weather/time scenario -- and I'm running it in ECO mode. If you haven't already, I recommend running MSI afterburner and watching the CPU % while gaming to see if that's the bottleneck in your case.

I won't link to it, but my video is titled "Does BAD WEATHER in AUTOMOBILISTA 2 = BAD FPS? We test and FIND OUT! RTX 4080, RYZEN 7700X, AMS2, 4K". (Mods, if this is not permitted, please let me know and I'll remove it.)