• We are currently performing site maintenance, parts of civfanatics are currently offline, but will come back online in the coming days. For more updates please see here.

Post your AI benchmark score

With my new rig:-
i7-9700
RTX 2070
16Gb RAM
All tested at 3840 x 2160

AI original test 7.87
Ave frame time: 14.027

Gathering Storm AI test: 34.84 turn time
Ave frame time: 18.903
 
Well, I ran the GS AI test but then I remembered I have a few interface and similar mods installed. It might have had some influence on performance.
Still, if it can be of any help to anyone:

i5-8400
8GB DDR4@3200Mhz (single channel)
W10x64, dx11

average turn time: 43,20
 
CPU: Intel Core 2 Quad Q8400 @ 2.66 GHz
GPU: Nvidia GeForce GTX670 (EVGA)
RAM: 8GB DDR2
AI Result: 11.19 seconds average turn time
GS AI Result: 103.2 seconds average turn time
 
Was fortunate to get a new 1660 for next to nothing so I thought I'd update my benchmarks.

CPU:
i5-4690k (4.5GHz o/c)
GPU: GTX 1660 XC ULTRA 6GB (with additional o/c +64MHz to base & +500MHz to memory)
RAM: 16GB DDR3 2133MHz
API: DX12
RESOLUTION: 1080p
---------------------------------------------------------------------------------
Average Turn Time: 8.69 (Up a little from 7.26)
-----------------------------------------
Graphics Settings: Ultra/Max
Average fps: 86.51 (Significant improvement over 59.59)
99th Percentile: 58.00
---------------------------------------------------------------------------------
Gathering Storm Average Turn Time: 45.35 (Up some from 38.71)
-----------------------------------------
Graphics Settings: Ultra/Max
Gathering Storm Average fps: 64.40 (Significant improvement over 46.82)
99th Percentile: 45.04
---------------------------------------------------------------------------------

I imagine the slightly higher AI turn times were probably due to additional background processes that I had running at the time. The 1660 XC ULTRA however, showed significantly better fps than my old 1060 SSC.
 
Last edited:
Congratulations to your new graphics card! As you are highlighting the Graphics Benchmarks, I would like to ask, at which screen resolution the measured frame rates were achived?

(I suppose also "normal" map sizes - huge maps may challenge even 6GB and influence the frame rates)

.
 
Last edited:
Congratulations to your new graphics card! As you are highlighting the Graphics Benchmarks, I would like to ask, at which screen resolution the measured frame rates were achived?
Thank you! I wasn't really in the market for a new GPU, but it fell into my lap so to speak.

I'm running @ 1080p. There's no way you'd be able to get those fps with a 1660 on higher resolutions @ max settings. From the benchmarks I've seen at various sites you'd probably be looking at a 15~20% or so reduction in fps @ 1440p and another 15~20% @ 2160p. If I was in the market for a new monitor and moved up to 2K I would have gone with a 1660 ti, or with a 4K monitor, a 2060 minimum.
(I suppose also "normal" map sizes - huge maps may challenge even 6GB and influence the frame rates)
The in-game benchmarking tool is all one size, but yes, larger map sizes would tax the GPU more.

*Edit - Fwiw, I thought I'd run the benchmarks again, but at the same "high" settings I ran on the 1060 instead of "max" settings.
---------------------------------------------------------------------------------
Graphics Settings: High
Average fps: 119.13
---------------------------------------------------------------------------------
Graphics Settings: High
Gathering Storm Average fps: 76.63
---------------------------------------------------------------------------------
It's all kind of moot though, since my monitor is only 60Hz.
 
Last edited:
CPU: Intel Core i7 7700K @4.2GHz (5.0GHz Actual)
GPU: Nvidia GeForce GTX 1060 6GB (EVGA)
RAM: 16GB DDR4

Turn 1: 5.00339
Turn 2: 7.88343
Turn 3: 6.8902
Turn 4: 8.3651
Turn 5: 11.84082

upload_2019-4-20_21-49-11.png
 
CPU: i5-4690k (4.5GHz o/c)
RAM: 16GB DDR3 2133MHz
API: DX12
Resolution: 1920x1080
=========================================
GPU: GTX 1060 SSC 6GB
Graphics Settings: High
GS Launch Update for Vanilla+R&F Average fps: 59.59
Gathering Storm Average fps: 46.82
-----------------------------------------
GPU: GTX 1660 XC ULTRA 6GB (with additional o/c +64MHz to base & +500MHz to memory)
Graphics Settings: High
GS Launch Update for Vanilla+R&F Average fps: 119.13
Gathering Storm Average fps: 76.63
---
Graphics Settings: Ultra/Max
GS Launch Update for Vanilla+R&F Average fps: 86.51
Gathering Storm Average fps: 64.40
I'm running @ 1080p. [...] From the benchmarks I've seen at various sites you'd probably be looking at a 15~20% or so reduction in fps @ 1440p and another 15~20% @ 2160p. If I was in the market for a new monitor and moved up to 2K I would have gone with a 1660 ti, or with a 4K monitor, a 2060 minimum.
When moving from 1080p/FHD to 2160p/UHD I would have expected a bigger reduction than 30~40%.

(Unfortunately I have still some time to buy my civ6workstation ... after the release of "Don't you dare to name it!"
OTOH this is great from the financial point of view! :D)
 
hi,gs benchmark has problem
I can not finish the ai test benchmark with gs dlc,just showing please wait
e3 1230 v3
8G
anyone like me?
how can i do?

I also have a xeon e3 1230 v3, 16GB RAM, and GTX 960 card. Light benchmark [with Quicken and Chrome running while testing] was 10.31. My file from last week appears to have vanished, but this is very similar to what I got with just Civ running. On the GS [heavy] benchmark, turn time was about 65 seconds. I've also noted that generally my CPU runs between 20-40% while playing but my GPU mostly runs in the 98-100% utilization level. I'll recheck again later just to ensure, but it appears that the other programs running are having minimum effect on turn times. Even as I type this, with Civ just sitting in the background doing nothing, I'm getting I'm getting 25-30% utilization of the CPU an d98-100% of the GPU.

Update: I reran the test with everything closed, and turn time decreased to 9.93. So difference with several open windows and just Civ was only about 0.4 seconds. Both runs were on DX11 as I found it was faster for me than DX12. Running a second test on DX12 resulted in turn times of 11.18 seconds - so I'll keep playing on DX11 for now.
 
Last edited:
Current Build:
CPU: Ryzen 9 3900x @ 4.2 ghz
RAM: 32gb 3600 CL14

DX12 AI scores
AI bench: 6.72
AI bench GS: 34.83

An interesting point is that I was using 12 threads more than 50% with 2 more around 20-30% and occasional spikes on an additional thread. After that the threads really die off.
 
  • Like
Reactions: cvb
An interesting point is that I was using 12 threads more than 50% with 2 more around 20-30% and occasional spikes on an additional thread. After that the threads really die off.
there is a thread for each civ and a few additional threads including a core thread. When the test is running they will happily burn your CPU and when there is nothing to process they quieten. Not sure when yours really died off.
 
I thought I'd re-run the benchmarks since the September 2019 patch notes specified some optimizations and performance improvements.

CPU:
i5-4690k (4.5GHz o/c)
GPU: GTX 1660 XC ULTRA 6GB (with additional o/c +64MHz to base & +500MHz to memory)
RAM: 16GB DDR3 2133MHz
API: DX12
RESOLUTION: 1080p
---------------------------------------------------------------------------------
Average Turn Time: 7.91 (Improvement from previous - 8.69)
-----------------------------------------
Graphics Settings: Ultra/Max
Average fps: 82.23 (Decline from previous - 86.51)
99th Percentile: 58.57 (Improvement from previous - 58.00)
---------------------------------------------------------------------------------
Gathering Storm Average Turn Time: 43.33 (Improvement from previous - 45.35)
-----------------------------------------
Graphics Settings: Ultra/Max
Gathering Storm Average fps: 63.81 (Decline from previous - 64.40)
99th Percentile: 46.60 (Improvement from previous - 45.04)
---------------------------------------------------------------------------------

First impression looks like turn times have improved slightly, while fps remain about the same on average (either slightly better or slightly worse). I also didn't notice any significant reduction in RAM usage though it seems that was supposed to be one of the performance optimizations.
 
CPU: i5-7200U
GPU: GeForce 940MX
RAM: 8GB DDR3 2133MHz
API: DX11

Turn 1: 7.57990
Turn 2: 6.76635
Turn 3: 9.98812
Turn 4: 13.21637
Turn 5: 11.64344
 
EDIT: Changed format and some content of post after double checking and expanding results

Using a Windows Surface Book (1st gen) with Enhanced performance. Disabled all mods before test. Graphics settings had no significant impact on test results.

CPU: i7-6600U
GPU: NVIDIA GeForce GTX 965M
RAM: 16GB

AI Benchmark:

base average time per turn (DX11 / DX12):
11.83s / 9.15s
Gathering Storm average time per turn (DX11 / DX12): 75.94s / 74.89s

--

Graphics Benchmark (minimum impact settings):

base average time per frame (DX11 / DX12): 21.592ms / 25.650ms, (46.31fps / 38.98fps)
Gathering Storm average time per frame (DX11 / DX12): 81.229 / 73.738ms, (12.31fps / 13.56fps)

--

Graphics Benchmark (ultra impact settings):
base average time per frame (DX11 / DX12): 57.385ms / 184.317ms, (17.43fps / 5.43fps)
Gathering Storm average time per frame (DX11 / DX12): 253.340ms / 454.538ms, (3.95fps / 2.20fps)

--

DX12 doesn't seem to impact the turn time much, maybe a slight improvement, but it really breaks my gpu. I think the graphics card was designed for DX11. I noticed this in game play when the DX12 option first became available, it made the game unplayable. I always use DX11 now and usually with low impact settings on both graphics and performance settings. I guess compared to other results I've seen, my graphics card is the limiting factor which isn't surprising for a laptop.

Note on the surface book: Occasionally, the connection between the "notepad" and base of the computer is not properly detected, in that case the game will switch to the Intel HD Graphics 520 chip built into the tablet portion rather than the dedicated graphics card in the base (and might remain as the default gpu until switched back). This obviously has a significant impact on game performance which confused me in early days. It leads to things like the game turn passes normally but things like scrolling across the map take forever. I always check my graphics settings in-game to make sure it's set to the NVIDIA card before starting my game.
 
Last edited:
AI Benchmark 7.68

CPU: AMD Ryzen 5 3600
GPU: Radeon RX 5700 XT
RAM: 32 GB
API: DX12
RESOLUTION: 1440
 
Changing the graphics settings should have definitely had an impact on your fps. Going from max to just high increased my fps ≈ 30.
I only tried changing graphics settings for the non-GS Graphics Benchmark. The first time I set both Performance and Graphics Impact to the max setting (Custom in my case which glossing over looked like the same settings as maximum). I then set it to the lowest (miminum) impact settings and got the same average frame time in both cases. I took it to mean that the benchmark is designed to test pre-set settings but maybe I forgot to save the changes to graphics settings or something. I'll double check when I get a chance later tonight.
 
Changing the graphics settings should have definitely had an impact on your fps. Going from max to just high increased my fps ≈ 30.
okay you're right, thanks for pointing this out. I've updated my previous answer:
EDIT: Changed format and some content of post after double checking and expanding results

Any idea what sized map is used in the GS benchmark?
 
for those posting the time by turn, where do I find that?

Just got a new machine so thought I'd repost my times:

CPU: i7-9700
GPU: GTX 1660ti
RAM: 16GB DDR4 2666MHz
API: DX12
RESOLUTION: 1080p

Average turn time - old benchmark: 7.32
Graphics:
Average fps:
168.41
99th percentile: 127.58

Average turn time - GS benchmark: 33.71
Graphics:
Average fps:
123.95
99th percentile: 94.04

Nice improvement from my xeon e3 1230 v.3 and gtx 960, which had average turn times of 8.88 and 54.72 respectively.
 
Back
Top Bottom