Fern
Member
Hello GMC,
This is (kind of) a followup post to the GM:S 1.0 vs 2.0 general/graphic performance experiment thread. Since the previous topic got a lot of positive response and feedback, we are doing a second thread to elaborate on the speed of the various collision systems throughout the Game Maker versions. These include GMS 1's normal collision system, fast(beta) collision system, and GMS 2's only collision system.
PC Specs:
Constants:
Conclusion:
On every single non-colliding test, GMS 2.0 did many multiple times better than it's GMS 1.0, but in the 50,000 check colliding test, GMS 1.0 outperformed GMS 2.0. Since many sources point towards 2.0s collision system being better suited for instances where most objects are not colliding, and with the results of this test, we can pretty accurately determine that that is the case. It's does seem to outperform the GMS 1.0 in the 100,000 check test, though, so that may either be from skewed data or just the nature of how the system handles that many objects. Since we have so few data point at the moment, and the data looks slightly skewed, I can't give you guys exact number; however, what I can do is say that for games where you have only a few collisions at a time, GMS 2.0 is great (which is most games). This performance boost ranges from 30x the amount of performance of the 1.0 system upwards. The performance of GMS 2.0's system while colliding depends on the amount of objects it's colliding with, so it varies widely. We can also draw the conclusion that GM:S 1.0's Fast Collision system is basically useless in every capacity (Always the lowest score), and was probably just to get data for 2.0's system instead of being shippable in any presentable game.
Best Wishes,
@Seabass (The Human) & @Shadow Gamer
Rendering & General Compute Benchmark:
https://forum.yoyogames.com/index.php?threads/gms-2-0-vs-gms-1-0-benchmark.11844/
This is (kind of) a followup post to the GM:S 1.0 vs 2.0 general/graphic performance experiment thread. Since the previous topic got a lot of positive response and feedback, we are doing a second thread to elaborate on the speed of the various collision systems throughout the Game Maker versions. These include GMS 1's normal collision system, fast(beta) collision system, and GMS 2's only collision system.
PC Specs:
AMD FX 9590
EVGA GeForce GTX 970
8GB of RAM (1600Mhz)
Crucial M500 SSD
EVGA GeForce GTX 970
8GB of RAM (1600Mhz)
Crucial M500 SSD
Constants:
Room Speed was 30 FPS
Background programs remained constant
PC Specifications (See above)
Instances/Versions of GMS 1.x and GMS 2.x used for each test
Background programs remained constant
PC Specifications (See above)
Instances/Versions of GMS 1.x and GMS 2.x used for each test
Procedure:
Instantiate 1 object with a 64x64 sprite
-Instantiate 1 object with a 64x64 sprite, has (X) amount of place_meeting() functions being called against the previous object
-Wait 5 frames (to offset initial frame drop)
-Begin averaging the FPS for 12 seconds (Real-Time, not based on frames)
Code Collision Performance Test (Objects Not Colliding):
Instantiate 1 object with a 64x64 sprite
-Instantiate 1 object with a 64x64 sprite, has (X) amount of place_meeting() functions being called against the previous object
-Wait 5 frames (to offset initial frame drop)
-Begin averaging the FPS for 12 seconds (Real-Time, not based on frames)
Code Collision Performance Test (Objects Not Colliding):
50,000 Collisions:
GMS 1.x (Normal Collisions): 71 FPS
GMS 1.x (Fast Collisions): 60 FPS
GMS 2.x: 2572 FPS
100,000 Collisions:
GMS 1.x (Normal Collisions): 43 FPS
GMS 1.x (Fast Collisions): 31 FPS
GMS 2.x: 3062 FPS
GMS 1.x (Normal Collisions): 71 FPS
GMS 1.x (Fast Collisions): 60 FPS
GMS 2.x: 2572 FPS
100,000 Collisions:
GMS 1.x (Normal Collisions): 43 FPS
GMS 1.x (Fast Collisions): 31 FPS
GMS 2.x: 3062 FPS
Code Collision Performance Test (Objects Colliding):
50,000 Collisions:
GMS 1.x (Normal Collisions): 85 FPS
GMS 1.x (Fast Collisions): 78 FPS
GMS 2.x: 66 FPS
100,000 Collisions:
GMS 1.x (Normal Collisions): 46 FPS
GMS 1.x (Fast Collisions): 42 FPS
GMS 2.x: 53 FPS
GMS 1.x (Normal Collisions): 85 FPS
GMS 1.x (Fast Collisions): 78 FPS
GMS 2.x: 66 FPS
100,000 Collisions:
GMS 1.x (Normal Collisions): 46 FPS
GMS 1.x (Fast Collisions): 42 FPS
GMS 2.x: 53 FPS
Conclusion:
On every single non-colliding test, GMS 2.0 did many multiple times better than it's GMS 1.0, but in the 50,000 check colliding test, GMS 1.0 outperformed GMS 2.0. Since many sources point towards 2.0s collision system being better suited for instances where most objects are not colliding, and with the results of this test, we can pretty accurately determine that that is the case. It's does seem to outperform the GMS 1.0 in the 100,000 check test, though, so that may either be from skewed data or just the nature of how the system handles that many objects. Since we have so few data point at the moment, and the data looks slightly skewed, I can't give you guys exact number; however, what I can do is say that for games where you have only a few collisions at a time, GMS 2.0 is great (which is most games). This performance boost ranges from 30x the amount of performance of the 1.0 system upwards. The performance of GMS 2.0's system while colliding depends on the amount of objects it's colliding with, so it varies widely. We can also draw the conclusion that GM:S 1.0's Fast Collision system is basically useless in every capacity (Always the lowest score), and was probably just to get data for 2.0's system instead of being shippable in any presentable game.
Best Wishes,
@Seabass (The Human) & @Shadow Gamer
Rendering & General Compute Benchmark:
https://forum.yoyogames.com/index.php?threads/gms-2-0-vs-gms-1-0-benchmark.11844/
Last edited: