I play Age of Empires 2 Conquerors all the time online with 8 players at 1080P but I use a Q9550 and GTS250 .
If the game stutters only late game I would think all those AI and full population (do you have 200 pop?) are just hogging the CPU too much.

200 pop yes, but the stutters and little freezes already start creeping up when at 80 pop or so. i have tested this on some other hardware as well in recent times - pii 400 (strangely pretty similar performance from memory, so maybe an FSB bottleneck) and p200mmx (pretty bad right from the start and only 64mb ram made it even worse). the original p166/32mb ram minimal requirement is quite the joke.

the HD version is a different beast with running under d3d9 and all that, the original version was directdraw of course.

The HD version I am talking about is the original release that the people of Voobly.com added higher resolution to while playing online (still 2d sprites just higher resolutions).

Age of empires 2 requires quite a strong CPU. Period correct ones are insufficient to run the game with high population cap. Even a Tualatin 1400s has trouble in this game. I play on my socket 462 machine and it runs fine with Barton 3200. with 7 enemy AI and 200 population cap.

played the original on Duron 800 with 256mb and tnt2 card, no problems at 80 or so pop but can imagine with larger population and larger maps the cpu will end up as the bottleneck; it must be running a great many calculations for each entity, but shuttling all that info to and from memory would also potentially be another bottleneck. I'd imagine, in keeping with many RTS games, that if you ran it on a late P4 or even later you'd not notice the slowdown as much due to all round speedier systems.

I was always impressed with various RTS games being able to keep up with many units in action at once on single core modest Mhz CPUs, making it appear as if it was all happening at once in parallel

the game could well be reacting unusually strongly to FSB or memory speed - i don't know why else the pii 400 would have felt so similar in performance. and the performance issues are really little freezes instead of a more granular drop in performance, so along with those sound cutoffs, that's somewhat rare to see. unfortunate that there isn't a proper way to benchmark this.

i'd be curious to see this game running on a willamette, maybe this is one case where that thing could actually leverage its memory bandwidth.

I suspect the abruptness of the freezups in this game is a result of the AI script processing. Certain events may trigger a flood of work not just to interpret the scripts but also because the script may decide to (for example) launch an attack with 100 units, triggering a sudden load of searching for those units and pathfinding all of them at once. There may also be cascade effects of 1 AI action triggering more AI actions.
This kind of behavior could explain the sudden and inconsistent choppiness.

I haven't played in a few years but I recall I still had issues with 200pop on a P3 866/133, but maybe I didn't on a Barton 2100/200. Never played it on a Tualatin.

In Aoe2 I strongly suspect it is frequency above all else! Tested last night on 2 different systems: Barton 2500 - lowest Barton CPU - can handle a game with 200 population cap while Tualatin 1400s can not. I will do more tests tonight.

if you mean CPU frequency, i don't see how that comparison would especially support that notion - barton 2500 runs just 433mhz faster than tualatin 1400s but has 333mt/s FSB and ddr ram versus 133mhz FSB and i'm assuming just pc133 sdram on tualatin. granted athlon xp will probably have higher IPC than p3 too so that's another factor. make sure to have a savegame with huge armies for repeatable testing.

i dropped the FSB to 66mhz on my aforementioned setup and can say the stutters got quite a bit worse, but that's not a very meaningful test either. to test the FSB/memory influence the best possible way would be some i815 (or even BX if can get 133mhz stable, i suppose) and comparing 100/133mhz FSB versions of the same CPU speed. bit of a far-fetched setup, really, but there might be some options with multiplier shenanigans on the athlon side as well.

That is a reason AMD used PR rating: most Athlon XP if not all had higher performance than P4 with the same frequency with which they were compared against !

I tested over the weekend with 3 more CPU's: Athlon XP 2600 Thoroughbred FSB 266 and Athlon XP 2700 Thoroughbred FSB 333. Both are running just fine with 200 population cap. Last processor tested was Athlon XP 1500 Palomino: severe stutters that made Aoe2 unplayable. The threshold for play-ability must be somewhere in between!

interesting.

by the way, does anyone know what FPS this actually runs at? i have a feeling it's capped to 30 or something like that.

Tested last night with Palomino 2100 and the result is not good: not playable!

Could it just be lack of ram? I have AOE 2 gold installed on my son's socket A Thunderbird system and even with the 200 pop cap the game runs fine.

Athlon t-bird@1200mhz
Abit KT7a-raid
1.5gigs pc-133 ram
GeForce 4 MX460 64meg
X-FI fatality
250gig at 133 drive
XP Pro SP3

chrismeyer6 wrote on 2021-04-22, 12:50:

Could it just be lack of ram? I have AOE 2 gold installed on my son's socket A Thunderbird system and even with the 200 pop cap […]
Show full quote

Could it just be lack of ram? I have AOE 2 gold installed on my son's socket A Thunderbird system and even with the 200 pop cap the game runs fine.

Athlon t-bird@1200mhz
Abit KT7a-raid
1.5gigs pc-133 ram
GeForce 4 MX460 64meg
X-FI fatality
250gig at 133 drive
XP Pro SP3

How many players / AI opponents?
How big is the map? (Not sure if that's a significant factor)

It was me and 3 other AI players it was a larger map with a decent amount of water. I have roughly 170 population. The games isn't finished yet but one of the AI players has similar population. Next time me and my son play again I'll get more specific details.

Is the SBLive! running using WDM or VXD drivers?

All CPU have been tested on the following system:
Abit AN7; Corsair 2*1gb RAM, Leadtek geforce 7800gs, onboard audio - Sound storm; WD raptor 74gb.
Game settings: highest difficulty; 7 enemy AI in 4 different teams; huge map; population cap 200.
Latest CPU tested: Duron 1800. Result: unplayable!
Out of all CPU tested the lowest that can play with the above settings are: Barton 2500 -1833mhz and Thoroughbred 2600 - 2133mhz. I do not have further socket 462 CPU so this is all I can test on this platform!
Already done 1 test on socket 754:
Abit KV8 PRO; Corsair 2*1gb RAM, Leadtek geforce 7800gs, onboard audio - Realtek; WD 200gb.
CPU tested: ClawHammer 2800 - 1800mhz. Result: very good, fully playable. It seems that frequency alone is not everything; architecture also matters!

Done over the weekend more tests:
1. Pentium 4 2000 Willamette on Abit BE7-raid with 2*1gb ram: result unplayable!
2. Pentium 4 2800 Northwood on Abit BE7-raid with 2*1gb ram: result very good!
2. Athlon 64 3000 Newcastle on Abit AN8-SLI with 2*1gb ram: result very playable, no problems whatsoever!
So despite the lower frequency Athlon 64 can run this game just fine whilst the early Pentium 4 struggles.