I just added 32GB more RAM today and got the same score: 1297.6. I had 16GB of RAM before, so it seems the game possibly doesn’t look beyond or care about anything greater than 16GB of RAM. I made no other hardware changes.
Who knows, maybe the cap is 8GB? Either way, maybe it was already been known/talked about how much RAM the game uses, but just mentioning in case it hasn’t or if it helps anyone with decision-making on upgrades.
Note: The Steam Store recommends 8GB of RAM for the game.
Oh, I just realized maybe the game’s bottleneck could be elsewhere (CPU, or what not)… so even if it enjoyed more RAM, it maybe has a speed limit elsewhere on my system.
It uses around 10GB during the ranked benchmark test:
CPU usage is also fairly light, this is what the CPU usage of my PC looks like, my CPU is out-performed by pretty cheap modern Ryzen options as I use it for work so I need it to be silent:
The GPU is the bottleneck, but even my fanless GTX 1650 isn’t maxed out during the ranked benchmark test if I have vsync on, the limiter is the 60hz refresh rate of my 4k monitor:
This is consistent with my testing on my gaming PC, which has a much more powerful 5700 XT graphics card, and the fps of the game scales in line with typical benchmarks across many games.
So for anyone wanting to improve the game’s performance, look at the views in Task Manager shown above, and see where the bottleneck is on your system. It’s almost certainly GPU speed. Also, I’ve found that recording gameplay using NVENC and OBS does hurt the game’s performance noticeably, even though NVENC uses a separate part of the GPU.
Ooh, cool info, breeminator! Good point about looking at Task Manager. I’ll try that next time, just out of curiosity
Update: The game just seems to have an upper-threshold to resources it wants to use. I just ran the benchmark test and had no bottlenecks anywhere. The game just didn’t want to use more resources:
CPU hovered around 30-35% usage (the dropoff in chart is when the test ended)
You probably have vsync on (in the game’s graphic options), a 60hz monitor, and a fairly powerful graphics card? If you turn vsync off, it should max out your graphics card.
VSync was actually off, but yep, I have a 60Hz monitor. It’s been a decent monitor for 10 years strong now
That’s a great score! Nothing wrong with older PC. I had a GTX260 originally for like 5 years, then upgraded to GTX520(?) for the last 5 years… within the same basic surrounding machine for about 10 years. It managed for a long time! I finally had to just upgrade last year, though. Any time I’d play a 2014’ish or newer FPS game on that old machine, the fans would go full-bore and my computer would power down completely to protect itself or crash hard. AoE2:HD ran pretty well, though
I did a bit more investigation, using my gaming PC (Ryzen 5 3600 + 5700 XT).
Like your PC, GPU usage looked not to be the limiter, it was down around 30%.
For CPU, you need to change the view to look at the individual cores, rather than the overall usage. Right click on the CPU graph and select Change graph to, then Logical processors. On my PC, it still didn’t look like the CPU was maxed out, but it was certainly heavily using 4 cores. They were getting close to 100% some of the time.
So, I enabled Gaming Boost in my BIOS, which increases the CPU clock speed by 600MHz, and tried it again. With the graphics fully zoomed out in both cases (which gives a lower score than zoomed in), the ranked benchmark score was increased from 1290 to 1312 by boosting the CPU by 600MHz. However, what was really interesting was it totally changed how it used the CPU. Instead of heavy use of 4 cores, it now looked like solid 100% use of one core. The game seems to have some way to choose between different ways to use the cores, depending on the performance of one core.
So I’ve changed my mind about what I said before, it now looks like the game is CPU limited if you have a decent graphics card, and the power of each individual core is probably more important than the number of cores.
It wasn’t full load, but it was heavy enough some of the time to make me think it was what was limiting it, given that nothing else seemed to be anywhere near limiting it.
It looks like it has some logic where it prefers to use one core, and is optimised for that, but it seems to be able to use multiple cores if it “wants” to. It’s a bit puzzling, because you’d think if it decides it can perform better by using 4 cores at 3.6GHz, it would also be able to perform better by using 4 cores at 4.2GHz, but perhaps it has some concept of performance being “good enough” with one core, and the one core performance is good enough at 4.2GHz, but at 3.6GHz it feels it’s too slow in some sense and switches to multi-core mode. I’d guess the best ranked benchmark score would be from an Intel >5GHz CPU, and you’d want to boost the clock speed even higher with extreme cooling if really keen to get the best possible benchmark score.
Game only uses one core for gaming and might use another for sound and video but low usage, all the calculation fo the simulation is always running in one single core/thread, that is the weakness of the genie engine, if you saw more cores having load then it could have been another process running in the background.
Anyway in my old PC, i had 1105 score with an fx 6300, using an entry level GPU rx560, using the same GPU but with ryzen 7 2700 i get 1275, using a better GPU like the 1660 super the score increases to 1300, so the CPU makes the biggest jump in performance.
There probably aren’t many people who have tested the Xbox vs Steam versions, so I thought I’d share my findings.
The Xbox version performs significantly better than the Steam version. The ranked benchmark score was consistently 15 points higher for the Xbox version, running it twice for each version. I also used a 3rd party FPS display tool to have a consistent FPS measurement across both versions, and the Xbox version ran at significantly higher FPS during the ranked benchmark test. In the early part of the benchmark, the Xbox version was running at 30-35fps, whereas the Steam version was running at 25-30fps.
In case anyone isn’t aware of the technical differences, the two versions are different, they aren’t simply the same code sold via two different stores. The Steam version is a classic Windows desktop app, whereas the Xbox version is a Universal Windows Platform app. If you read articles online about which you should buy, I have only ever seen it advocated that you should always buy the Steam version, but some of the complaints about UWP apps in older articles have now been addressed.
As far as I can see, the only benefit now of having the Steam version of AoE II DE is the ability to access the files for creating mods. For actually playing the game, the Xbox version seems distinctly better.
The PC where I was seeing it using multiple cores is a bit of a pain for me to get screenshots off, but I’m now seeing multiple core use on this PC, so thought I’d share a screenshot. The CPU usage when running the benchmark and when idle on the desktop is completely reproducible, so there’s no doubt at all that it’s the game making use of all the cores like the screenshot shows. Incidentally, it also uses significant CPU just showing the menu, again this is reproducible by repeatedly running the game up and shutting it down.
There is something weird in ur pc, mine using one core only, 7-8% represents one single thread in my cpu, it is the same playing or running the benchmark, the only value that changes is memory usage.
I recently got hold of a 3080 graphics card, so I thought I’d share the benchmark results. TLDR - it makes zero difference to the AoE II DE benchmark score, it’s CPU limited with even a pretty low end graphics card, even at 4k with enhanced graphics.
Ryzen 5 3600 CPU
1660 Super GPU
1080p no UHD 100% zoom 1265 105W
2160p UHD 75% zoom 1179 151W (75% zoom gives same map view as 1080p 100%)
3080 GPU scores unchanged, but power increases to 151W and 164W
So, I had previously thought that the score drop at 4k would be addressed by a more powerful graphics card, but the 3080, which gives over double the frame rate of the 1660 Super at 4k in some other games, had no effect on the ranked benchmark score at 4k. So don’t buy one for AoE II DE, it just costs more, consumes more power, and makes more noise!
I also trickled the 1660 Super down to my productivity PC, replacing the 1650 that was in it:
Intel i7 6700T CPU
1650 GPU
1440p no UHD 50% zoom 1166 82W (same map view as 1080p 100%)
1660 Super GPU score unchanged but power increases to 104W. I didn’t test this PC at 4k because 1166 is already too low a score, the frame rate visibly suffers even for busy single player campaign games. So for playing at 1440p, even the 1650 is not a limiter, it’s all about the CPU (for the ranked benchmark, at least).
they really need to allow host to create lobby with benchmark restriction, any game with 3v3 or 4v4 involving 200 pop cap lags the game to no end. everybody in those game needs to be at least above 1200.
Current patch is broken, i was in a game with no fps loss and 130 ping, yet i had red clock and the game was lagging so bad, it was nothing on my end, 4x4 are unplayable for some reason or at least more than before.
Also i upgraded my old GPU to a 5700xt and got 20 less score same res and settings, in fact the game gpu is under 30% usage,the frequency doesn’t even pass the 500mhz because the game is not even pushing it, it is all about cpu.
game is not about GPU. its about CPU. for CPUs from intel starting from gen 3 needs at least a 3.5-3.6ghz to play 2v2 without lag. anything over that might be even tough for that.
Today, I scored 1293.9 in my 1st test, 1290.1 in my 2nd test a few minutes later, and 1286.2 in my 3rd test a few minutes later after shutting off a bunch of mods (such as tracer fire, original AoE2 sounds, large projectiles, etc.). Back in June, I had 1297.6.
Not a big difference, but thought I’d mention.
Hehe, with my score dropping a few points each test today, I’m afraid to test a 4th time!
thats within margin of error including possible small background task your window is running. i’d say anything over 20 points would be considered a drop in performance. within maybe 10-15 i’d just ignore it.
