Lix stress test

[Forestidia86]

I mentioned it already in this post but since it was something like a stress test for the game I post it with a bit more precise information:

System specs:

Old laptop (abbreviation: OL):
CPU: Pentium(R) Dual-Core CPU T4500 @ 2.30GHz
RAM: 4 GB
Graphics card: Mobile Intel(R) 4 Series Express Chipset Family
VRAM: Dedicated: 64 MB -  Total (including shared memory): 1695 MB
OS: Win 7 Home Premium 64 bit

Newer laptop (abbreviation: NL):
CPU: AMD A6-5200 APU Quad-Core @ 2.00GHz
RAM: 8 GB
Graphics card: AMD Radeon HD 8400
VRAM: Dedicated: 493 MB -  Total (including shared memory): 4077MB
OS: Win 8.1, 64 bit

The level "Any Way You Want" without doing anything:
OL: FPS: about 12, VRAM: max. 87 MB
NL: FPS: slightly below 60, VRAM: max. 87 MB

Stress test:

Map: Infinitus (8p) modified to have 999 lix, played within singleplayer

Test:
- let all lix spawn
- nuke

FPS at the beginning:
OL: about 5
NL: about 40-45

FPS before nuking:
OL: 2-4
NL: about 18

FPS while nuke:
OL: 2
NL: about 10
FPS eventually rise when much less lix are on the screen by the end.

max. VRAM-usage:
OL: about 218 MB (250 MB was from a bit different setting with more distributed lix)
NL: about 203 MB


 

[Simon]

Thanks for the detailed stats!

Even the good laptop suffers framerate drops at 0 lix out, hmm, but it's a very large map after all. 40 fps feel strange during fast-forward, and you might find that physics updates take variable real-time. But 40 fps should still make for a good networked game.

The VRAM usage (reported by Lix's ingame status panel) is a ballpark estimation, it might be off by 10 % or 30 %. I track all VRAM bitmap allocations, but the graphics card is free to handle the memory however it desires. Maybe it allocates more to guarantee better performance, maybe it allocates less. We don't need perfectly exact values; if we ever needed some, we should use hardware-monitoring software.

-- Simon

[Forestidia86]

Even the good laptop suffers framerate drops at 0 lix out

I let some lix out since at the very beginning the FPS-count seems to build itself until it reaches a real determinable value.
The newer laptop is a rather low-price model that is already not the newest anymore and probably not optimized for gaming.

[Forestidia86]

We don't need perfectly exact values; if we ever needed some, we should use hardware-monitoring software.

I actually have done this with the GPU of my newer laptop. Logfile and one screenshot of the diagrams attached. (I used fast forward to let the lix spawn.)
The highest value of VRAM-usage (can be seen in the rightmost columns of the logfile) seems to have been:
Memory Usage (Dedicated): 226 +
Memory Usage (Dynamic): 158

I'm not sure if these two values are to be added.

The VRAM usage without having the game Lix (or any other particular program in foreground) open is Memory Usage (Dedicated): 40 + Memory Usage (Dynamic): 15.

Edit: What seems to strain as well is the amount of lix in the level. How many lix are there in a multiplayer match of 8 in this map? Though SQron seemed to have crashed early.
(In the level "Any Way You Want" FPS drop to slightly below 45 on NL with 999 lix on the map.)

[Colorful Arty]

Oh, I thought this would be a thread about the stress Lix multiplayer causes.

[nin10doadict]

I was thinking the same thing, Arty. My sportsmanship is still a work in progress too, but at least I don't break things when I lose like I did when I was 7 years old.