78e3bc11e6
initial update and render with concurrent DS
2023-05-20 22:15:15 +02:00
1d3132cf3c
use oneTBB library
2023-04-29 14:54:09 +02:00
52537715ef
chunks: make chunk state atomic
2023-04-23 16:28:36 +02:00
3daf994ac3
opengl: disable vsync
...
This way I can better watch how changes affect FPS/frame times, rather than have it capped at
60FPS/16.6ms
Does not introduce visible tearing.
The engine runs at about 170 FPS on my Tesla M40, and so does minecraft. It never goes past that, I think this is a
limitation of using prime offloading. Runs at 300FPS on average on the Radeon HD6850 to which I
normally offload the Tesla to
2023-04-10 00:45:25 +02:00
47543c9101
intervalmap: use updated upper_bound when checking removal of next node
2023-04-10 00:30:28 +02:00
1a4412c5b1
renderer: initial texture support via ArrayTextures
...
totally stolen textures from minecraft
2023-04-10 00:30:28 +02:00
2b1991ff2b
separate rendering in a dedicated file
2023-04-10 00:30:24 +02:00
b4f3bdad06
experimental 3d generation
...
a cave-like effect
2023-03-29 21:39:53 +02:00
40fec8a649
actual random generation
2023-03-29 21:39:53 +02:00
179bd20aba
update .gitignore
2023-03-27 21:10:45 +02:00
a039ced6cf
gamma correction
2023-03-27 21:10:45 +02:00
44b027c3e2
Blinn-Phong lighting
2023-03-27 21:10:45 +02:00
6bd3bdecb8
shader: allow to send vec3 uniforms
2023-03-27 21:10:45 +02:00
b4dd3be584
chunkmesher: push vertices normals to GPU
2023-03-27 21:10:38 +02:00
da6608c66a
chunkmanager: only calculate update sphere indices at startup
2023-03-25 16:32:01 +01:00
8584d2e974
general code cleanup
2023-03-25 16:28:12 +01:00
19cd81b9fb
chunkmesher/manager: move sending data to gpu in dedicated function
2023-03-23 16:19:25 +01:00
19b3c68015
camera: nearer near clipping plane
2023-03-23 14:26:12 +01:00
e609f4858b
initial blockpicking
2023-03-23 14:25:34 +01:00
e18a2cca05
intervalmap: solve bugs with insertion, printing and retrival
2023-03-21 22:15:46 +01:00
8040868055
chunkmanager: unload chunks after a time treshold
2023-03-13 13:33:32 +01:00
b7077af00f
chunkmanager: delete far chunks from memory
2023-03-13 12:56:46 +01:00
ba35689c6c
chunkmesher: also cleanup colors at the start
2023-03-13 12:56:10 +01:00
bc2225e0f3
chunk: first delete buffer, then vao
2023-03-13 12:55:20 +01:00
f3d89a2e5a
chunk release resources on destruction
2023-03-12 22:14:28 +01:00
ddd047e82f
interval map: clean map on destruction
2023-03-12 22:14:05 +01:00
ab762eb9bc
use unique_ptr when dumping intervalmap to array
2023-03-12 19:28:23 +01:00
f7eb04a5f1
use unique_ptr when dumping intervalmap to array
2023-03-12 19:24:09 +01:00
383fb60686
proper frustum culling
2023-03-12 19:24:05 +01:00
e69c58abd3
gracefully shut down secondary threads
2023-03-12 12:43:41 +01:00
716056b8c5
experimental multithreaded generation and meshing
2023-03-03 21:33:11 +01:00
2ec728897e
main: print currently used GPU at the start
2023-02-23 19:07:35 +01:00
b1e420f564
camera: allow to strafe up and down
2023-02-23 19:07:17 +01:00
cf0bb3f111
update .gitignore
2023-02-23 19:06:42 +01:00
b600ef5214
memory: delete shader when terminating
2023-02-23 17:02:20 +01:00
0baa2eb649
memory: use proper delete[]
2023-02-22 21:54:15 +01:00
c547d11cd5
chunks: drop per-chunk chunkmesh
...
Use a single mesher. This, on top of the previous commits, shaves off another 20MB of total RAM usage, using a total of about 120MB for a standard world (16 chunk size, 8 render distance), of which 22 in heap
2023-02-12 14:36:59 +01:00
ec2feb919f
chunkmanager: chunkmesh on the stack
2023-02-12 12:20:03 +01:00
9f82a17bd2
chunkmesh: do not create shader for every object
2023-02-12 12:18:30 +01:00
2e0ed7b057
clear mesh data after sending to gpu
2023-02-11 22:26:34 +01:00
3c54de5f21
update .gitignore
2023-02-11 21:52:31 +01:00
4aef800210
ship glfw
2023-02-11 21:49:35 +01:00
69c44e3609
enable faceculling
...
Doesn't give a massive improve in performance right now, but a nice one to have and was already made from the jme3 porting
2022-12-01 23:45:05 +01:00
773b7999d4
chunkmgr: do not update same chunk multiple times
...
Iterating on radius from 0 up to RENDER_DISTANCE made the algorithm check the same "sphere" multiple times
2022-12-01 23:45:05 +01:00
1ccbf8b157
rudimentary frustum culling
...
TODO: maybe manually extract planes and check against those
2022-12-01 23:45:05 +01:00
8116791dcf
use hilbert curve instead of flatten array
...
Using valgrind heap memory tracker, this does not seem to bring any advantage on heap memory usage. To be tested more in depth
2022-12-01 23:44:47 +01:00
b724640384
chunkgen: take advantage of intervalmap structure
...
brings down heap memory usage to about 67% of that using arrays (valgrind, tested on standard terrain with renderdistance at 8)
2022-12-01 23:44:03 +01:00
8eb15f4725
chunk: migration to intervalmaps
2022-12-01 23:44:03 +01:00
5dd924abf7
implement intervalmaps
2022-12-01 23:44:03 +01:00
2418c481bd
camera: wider and further view
2022-12-01 23:44:03 +01:00
EmaMaker