Benchmark this JAR - yes, a Vangard demo!

Intel ® Core i3-3110M CPU @2.4GHz
Intel HD Graphics 4000 Graphics Card
1600x900 resolution

At 1.0 zoom
Logic: ~18ms
Render: ~250ms

At 10.0 zoom
Logic: ~18ms
Render: ~125ms

At 75.0 zoom
Logic: ~15ms
Render: ~100ms

turns out I have the worst computer out of all JGO ::slight_smile:

i7 4700MQ 2.4 GHz
Intel HD Graphics 4600
1080p

`zoom 1:
logic 0-2 ms
render 63 ms

zoom 10:
logic 0-2 ms
render 28 ms

zoom 75:
logic 1-2 ms
render 35 ms
`
Logic times display as ‘0’ most of the time, so no problems there. Task manager shows all 8 (virtual) cores at about 1/4 to 1/3 usage, with one of them at 3/4 (probably hosting the render thread).

I couldn’t get the process to pick the GT 755M for the GPU unfortunately.

Hmmmmm… Looks like I had better “optimise” the render pipeline. As in: replace it completely with LibGDX.

AS a lighter lift, I might try convert it to JavaFX first, that would be quite a small change for me (and no learning curve). It would be interesting to see if the graphics performance changes under JavaFX.

Just wondering, is your laptop’s memory single-channel? My AMD iGPU is managing 6x better performance…

=============================

I actually own one of those too, although mine is tricked out with a GT430 graphics powerhouse. I never tried to run Vangard on that because I pretty much knew it would be hopeless.

CPU-Z is reporting that it is single channel. That’s annoying.

I’d like to think you are hitting a hardware bottleneck there, it’s not my shoddy code :slight_smile:

My iGPU got boosted significantly when I upgraded to dual-channel RAM. Technically I have a second dGPU in the laptop but I am very certain it does nothing under all circumstances.

One problem I have with the game is that even x10 speed might not be enough of a fast forward… But it looks like on many systems I can simply double or triple the logic frames for those systems that achieve 12ms or under. That will mean that on faster machines (most desktops, i7 laptops) speeds of x20 or x30 or even x40 would be possible.

In a feat of brilliant insight, I found a way to improve the logic millis on my laptop from 21ms to 13ms.

I plugged the power cord in.

I just ran this JAR on my new PC: it’s an i7-6700 with Intel HD Graphics 530.

Logic millis = 5
Graphics millis (zoomed in): 20
Graphics millis (zoomed out): 56

I set this system variable:

J2D_D3D_NO_HWCHECK=true

Now I get these much improved times on my i7-6700 system:

Logic Millis: 0-1ms
Render Millis (zoomed in): 8-14
Render Millis (zoomed out): 15-20