Pat Proctor
03 Feb 05, 17:32
Gents,
One of the ProSIM developers asked a good question, the answer to which, I think all can benefit from:
Patrick
Minor question - what is the max number of units which can be played on ATF
before the AI fails?
Here's my answer:
The short answer is that you can always put a battalion of maneuver and a couple battalions of artillery on the board, in any situation, and be OK. Beyond that, the question gets complicated.
There IS NO HARD LIMIT on how many units you can put on the board. But, at some point, the engine bogs down so much that on low-end systems, the game is unplayable. Here are the factors that impact this and how.
1. Complex terrain = less units. The more terrain objects you add, the more they bog down the engine. Objects that obscure but do not block vision cost the most in CPU-cycles. 0 height terrain objects that effect movement only cost the least.
2. Mountainous terrain = less units. This actually HELPS the viewing process, but HURTS the path finding process. The net impact is that the more mountainous the terrain, the less units you can put on the board.
3. Friendly units cost more than enemy units. Friendly units cost more because the player might do ANYTHING with them, from the most complex company missions and company formation paths through the most constrictive terrain to no missions or formations at all. You have to "worse-case-it" and expect that the player will do the toughest stuff. You can be a little more liberal with adding friendly units if the player is in the defense, as he will probably not do much in the way of missions or formation movement.
4. More enemy missions = less units. The more missions you give to your enemy force, the fewer you can put on the board. The MOST costly missions are those that are triggered by the Faction AI, during game play. They can actually pause the simulation for a minute or more, if they are very complex.
5. More enemy formation movement = less units. The most costy formations and paths are those that are planned through constricted terrain AND triggered by the Faction AI. Whenever possible, give paths to hierarchies WITHOUT formations set, as this disables pathfinding.
6. Enemy in defense = more units. When the enemy is in defense, he has few or no missions and formations. This costs less CPU cycles.
7. Greater view range = less units. If you have units on the board that don't need to see anything, like civilians or ammo pallets, give them very low, or even 0 view range. This will cause them not to take up many cycles on viewing, the second most costly engine activity (after path planning).
8. More path finding = less units. Whenever possible, if the enemy must use missions that are started by the faction AI, use the "assign path" order first, select and use the ".none." for the formations in the property sheet. This will streamline the AI planning that happens in the mission initiation.
9. Piecemealing = more units. If you can, either by positioning of enemy defenses, or planning of enemy movement, prevent ALL of the enemy's units from being in contact with ALL of the player's units at once, it will prevent too much sighting and shooting all at once.
The MAXIMUM number of units you could have on the board:
1. The enemy in defense with no missions or formations assigned.
2. faction AI includes only simple Assign Path orders and only one phase.
3. All enemy artillery and ammo carriers have a 500m view range.
4. Completely open, flat terrain (ala Basrah attack from ATF).
= One friendly brigade with artillery battalion vs. two regiments with five artillery battalions.
(if the enemy did a full frontal attack, with all of his forces closing with the enemy at once, you might STILL get a little bit of lag at the highest contact point.
One of the ProSIM developers asked a good question, the answer to which, I think all can benefit from:
Patrick
Minor question - what is the max number of units which can be played on ATF
before the AI fails?
Here's my answer:
The short answer is that you can always put a battalion of maneuver and a couple battalions of artillery on the board, in any situation, and be OK. Beyond that, the question gets complicated.
There IS NO HARD LIMIT on how many units you can put on the board. But, at some point, the engine bogs down so much that on low-end systems, the game is unplayable. Here are the factors that impact this and how.
1. Complex terrain = less units. The more terrain objects you add, the more they bog down the engine. Objects that obscure but do not block vision cost the most in CPU-cycles. 0 height terrain objects that effect movement only cost the least.
2. Mountainous terrain = less units. This actually HELPS the viewing process, but HURTS the path finding process. The net impact is that the more mountainous the terrain, the less units you can put on the board.
3. Friendly units cost more than enemy units. Friendly units cost more because the player might do ANYTHING with them, from the most complex company missions and company formation paths through the most constrictive terrain to no missions or formations at all. You have to "worse-case-it" and expect that the player will do the toughest stuff. You can be a little more liberal with adding friendly units if the player is in the defense, as he will probably not do much in the way of missions or formation movement.
4. More enemy missions = less units. The more missions you give to your enemy force, the fewer you can put on the board. The MOST costly missions are those that are triggered by the Faction AI, during game play. They can actually pause the simulation for a minute or more, if they are very complex.
5. More enemy formation movement = less units. The most costy formations and paths are those that are planned through constricted terrain AND triggered by the Faction AI. Whenever possible, give paths to hierarchies WITHOUT formations set, as this disables pathfinding.
6. Enemy in defense = more units. When the enemy is in defense, he has few or no missions and formations. This costs less CPU cycles.
7. Greater view range = less units. If you have units on the board that don't need to see anything, like civilians or ammo pallets, give them very low, or even 0 view range. This will cause them not to take up many cycles on viewing, the second most costly engine activity (after path planning).
8. More path finding = less units. Whenever possible, if the enemy must use missions that are started by the faction AI, use the "assign path" order first, select and use the ".none." for the formations in the property sheet. This will streamline the AI planning that happens in the mission initiation.
9. Piecemealing = more units. If you can, either by positioning of enemy defenses, or planning of enemy movement, prevent ALL of the enemy's units from being in contact with ALL of the player's units at once, it will prevent too much sighting and shooting all at once.
The MAXIMUM number of units you could have on the board:
1. The enemy in defense with no missions or formations assigned.
2. faction AI includes only simple Assign Path orders and only one phase.
3. All enemy artillery and ammo carriers have a 500m view range.
4. Completely open, flat terrain (ala Basrah attack from ATF).
= One friendly brigade with artillery battalion vs. two regiments with five artillery battalions.
(if the enemy did a full frontal attack, with all of his forces closing with the enemy at once, you might STILL get a little bit of lag at the highest contact point.