|
Dakota_Fanning
Hungary2359 Posts
EDIT:
The new Sc2gears 8.1 has already been released with EAPM support. This is how it looks like:
![[image loading]](https://lh4.googleusercontent.com/-Ki6vR6goRsc/ToQZlfbOBCI/AAAAAAAAgt0/8zuB_Qbs6Mo/s590/apm_types.png)
This thread should be closed since the EAPM is out there.
Details of the EAPM algorithm: APM Types
INTRODUCTION
Since Blizzard tried to transform APM into some kind of EAPM in their latest 1.4 Patch but failed to consider some key aspects as to what is considered effective, I decided to implement one myself in Sc2gears that is planned to be more useful.
APM stands for Actions Per Minute. APM measures the amount of commands that would be given by a player in 1 minute. Higher APM means more activity. EAPM stands for Effective APM. EAPM measures the amount of useful/effective commands that would be given by a player in 1 minute.
APM calculation is pretty simple: one just have to divide the number of actions with the number of minutes where the actions spread in time. For example 200 actions in 1 minute results in 200 APM. EAPM calculation is similar, but it excludes actions that are seemingly useless/needless. The algorithm uses different rules to classify actions effective or ineffective. The first 2 minutes (can be changed) of games are always excluded from APM/EAPM calculation. Moreover the durations for APM/EAPM calculations are determined by the last actions of the players. For example if a player leaves after 5 minutes, his/her APM and EAPM values will be calculated with a 3-minute long game.
EAPM RULES
Commands that are considered ineffective:
- Unit queue overflow: if more units are trained in a short time (~1.05 sec) than the selected buildings can queue; or in case of Zerg if more units are trained than the number of selected Larvas
- Too fast cancel (~0.84 sec): if a command (like train, research, upgrade, hatch) gets cancelled right away
- Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
- Too fast switch away from a selected unit or reselecting the same units without giving them any commands (~0.336 sec) (by too fast I mean there is not even time to check the state of the units and optionally react to it accordingly); double tapping a hotkey to center a group of units is NOT considered ineffective
- Repetition of some commands (without time restriction): hatch, morph, upgrade, research, build, cancel, merge archon, merge dark archon, lift, hotkey assign to same group
Commands that are NOT considered ineffective:
- Double tapping a hotkey to center the group
- Selecting units or buildings to check their state but not giving new order to it.
- Quickly adding waypoints which are not too close.
- I do not check the target point of move commands: if they are close in time, the first is ineffective (unneccessary); if they are not close in time, they are both effective even if they point to the same location (issuing another move will cause StarCraft II to recalculate path and therefore resulting in faster arriving in many cases).
Notes:
- Camera movement actions are not included in APM nor in EAPM calculations. For example it does not matter whether your Nexus is in the center or at the top of your screen when you click on it. The outcome of the actions and the game does not depend on where the objects are displayed on your screen, what matters is the position (x and y coordinates) where you click on the map.
OPINIONS
Please express your opinions if you agree or disagree with these rules and why; what would you change in these rules and what other rules you would define to create a useful EAPM definition.
I will update this post accordingly.
|
I wholeheartedly agree with the way EAPM will be calculated in the next update of SC2Gears. This is how it is meant to be. Thank you so much for always bringing us this amount and quality of utility. This will put an end to the apm debate nonsense too.
*Plus I always prefer real-time apm calculations instead of "blizzard time."*
|
Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
i see a problem with that... when someone is able to micro his units really fast, some of his actions will be considered ineffective. right?
|
Dakota_Fanning
Hungary2359 Posts
On September 27 2011 21:26 beg wrote:Show nested quote +Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign i see a problem with that... when someone is able to micro his units really fast, some of his actions will be considered ineffective. right?
If you are that fast that you can select a group of units and issue an order and select another group of units and issue an order and select a 3rd group of units and issue an order... all this very fast (let's say within 1 second), this will all be considered effective.
What the quoted rule fragment implies is that you select a group of units and you issue them to move to a location, then 0.1 seconds later you issue them the same order. Now in this case if you would have given only 1 move command, the result would be the "same" therefore the 2nd move command was "ineffective/useless".
|
Interesting, I like the idea, but I had a few comments:
- Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
Does this affect waypointing? Also, what about micro situations where it is necessary to make a move command in one direction and then rapidly change direction, particularly when using faster units (e.g., mutas, lings or hellions to outmanoeuvre or bait slower units).
EDIT: I may have misunderstood - by repetition, do you mean the same command in more or less the same location?
- Too fast switch away from a selected unit or reselecting the same units without giving them any commands (~0.336 sec) (by too fast I mean there is not even time to check the state of the units and optionally react to it accordingly); double tapping a hotkey to center a group of units is NOT considered ineffective
I'm still not particularly good at tapping regularly throughout a game, but I thought (and I could be wrong) that the idea was that by rapidly cycling through the list of all buildings/units you could peripherally follow the status of each. And thanks to persistence of vision, you can still view what each group shows even if you are moving between groups faster than you can process a single tap. For instance, if you were to extremely rapidly switch between two groups, you could effectively see both groups at once, superimposed (the same way that a fan looks like a solid disc once it spins fast enough).
|
Dakota_Fanning
Hungary2359 Posts
On September 27 2011 21:33 whatthefat wrote:Interesting, I like the idea, but I had a few comments: Show nested quote +
- Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
Does this affect waypointing? Also, what about micro situations where it is necessary to make a move command in one direction and then rapidly change direction, particularly when using faster units (e.g., mutas, lings or hellions to outmanoeuvre or bait slower units). EDIT: I may have misunderstood - by repetition, do you mean the same command in more or less the same location? Show nested quote +- Too fast switch away from a selected unit or reselecting the same units without giving them any commands (~0.336 sec) (by too fast I mean there is not even time to check the state of the units and optionally react to it accordingly); double tapping a hotkey to center a group of units is NOT considered ineffective I'm still not particularly good at tapping regularly throughout a game, but I thought (and I could be wrong) that the idea was that by rapidly cycling through the list of all buildings/units you could peripherally follow the status of each. And thanks to persistence of vision, you can still view what each group shows even if you are moving between groups faster than you can process a single tap. For instance, if you were to extremely rapidly switch between two groups, you could effectively see both groups at once, superimposed (the same way that a fan looks like a solid disc once it spins fast enough).
"Too fast repetition of some commands" refers to repeating the same command. For example if you micro a group of stalkers like this:
Move, Attack move, Move, Attack move...
These actions will all be considered effective.
About the tapping: the time limit (I proposed 0.336 seconds) can be decreased to cover cases where you cycle through your units/groups fast .
|
On September 27 2011 21:39 Dakota_Fanning wrote:Show nested quote +On September 27 2011 21:33 whatthefat wrote:Interesting, I like the idea, but I had a few comments:
- Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
Does this affect waypointing? Also, what about micro situations where it is necessary to make a move command in one direction and then rapidly change direction, particularly when using faster units (e.g., mutas, lings or hellions to outmanoeuvre or bait slower units). EDIT: I may have misunderstood - by repetition, do you mean the same command in more or less the same location? - Too fast switch away from a selected unit or reselecting the same units without giving them any commands (~0.336 sec) (by too fast I mean there is not even time to check the state of the units and optionally react to it accordingly); double tapping a hotkey to center a group of units is NOT considered ineffective I'm still not particularly good at tapping regularly throughout a game, but I thought (and I could be wrong) that the idea was that by rapidly cycling through the list of all buildings/units you could peripherally follow the status of each. And thanks to persistence of vision, you can still view what each group shows even if you are moving between groups faster than you can process a single tap. For instance, if you were to extremely rapidly switch between two groups, you could effectively see both groups at once, superimposed (the same way that a fan looks like a solid disc once it spins fast enough). "Too fast repetition of some commands" refers to repeating the same command. For example if you micro a group of stalkers like this: Move, Attack move, Move, Attack move... These actions will all be considered effective.
Right, but what if I have a group of lings outside a Protoss natural, and I click MOVE to mineral line, then very rapidly click MOVE back, so as to bait the Protoss defense. I'm clicking MOVE to 2 completely different locations - does that count as two actions or one? A similar situation can occur with mutas - I move them into a Terran base, then very rapidly move back - either to avoid a Thor I just spotted, or to try to bait a stim from marines. I would argue that these are both useful commands. Spamming to me is when MOVE is issued to the exact same LOCATION twice very rapidly in a row.
Similarly with waypointing, I may very quickly input a path consisting of several points, e.g., for waypointing a drop, or for setting up a patrol.
About the tapping: the time limit (I proposed 0.336 seconds) can be decreased to cover cases where you cycle through your units/groups fast .
I agree, I'm just not sure what is an appropriate lower limit. It seems like some pros cycle through extremely quickly. I'm not sure whether they are purely spamming, or whether they are doing what I suggested (i.e., using persistence of vision to view more than one group at once).
|
|
|
Dakota_Fanning
Hungary2359 Posts
On September 27 2011 21:51 whatthefat wrote:Show nested quote +On September 27 2011 21:39 Dakota_Fanning wrote:On September 27 2011 21:33 whatthefat wrote:Interesting, I like the idea, but I had a few comments:
- Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
Does this affect waypointing? Also, what about micro situations where it is necessary to make a move command in one direction and then rapidly change direction, particularly when using faster units (e.g., mutas, lings or hellions to outmanoeuvre or bait slower units). EDIT: I may have misunderstood - by repetition, do you mean the same command in more or less the same location? - Too fast switch away from a selected unit or reselecting the same units without giving them any commands (~0.336 sec) (by too fast I mean there is not even time to check the state of the units and optionally react to it accordingly); double tapping a hotkey to center a group of units is NOT considered ineffective I'm still not particularly good at tapping regularly throughout a game, but I thought (and I could be wrong) that the idea was that by rapidly cycling through the list of all buildings/units you could peripherally follow the status of each. And thanks to persistence of vision, you can still view what each group shows even if you are moving between groups faster than you can process a single tap. For instance, if you were to extremely rapidly switch between two groups, you could effectively see both groups at once, superimposed (the same way that a fan looks like a solid disc once it spins fast enough). "Too fast repetition of some commands" refers to repeating the same command. For example if you micro a group of stalkers like this: Move, Attack move, Move, Attack move... These actions will all be considered effective. Right, but what if I have a group of lings outside a Protoss natural, and I click MOVE to mineral line, then very rapidly click MOVE back, so as to bait the Protoss defense. I'm clicking MOVE to 2 completely different locations - does that count as two actions or one? A similar situation can occur with mutas - I move them into a Terran base, then very rapidly move back - either to avoid a Thor I just spotted, or to try to bait a stim from marines. I would argue that these are both useful commands. Spamming to me is when MOVE is issued to the exact same LOCATION twice very rapidly in a row. Similarly with waypointing, I may very quickly input a path consisting of several points, e.g., for waypointing a drop, or for setting up a patrol. Show nested quote +
About the tapping: the time limit (I proposed 0.336 seconds) can be decreased to cover cases where you cycle through your units/groups fast .
I agree, I'm just not sure what is an appropriate lower limit. It seems like some pros cycle through extremely quickly. I'm not sure whether they are purely spamming, or whether they are doing what I suggested (i.e., using persistence of vision to view more than one group at once).
This comes down again to the time limit that will be used eventually in the implementation.
If you move to the expand and 1 ms later you move back, then obviously your lings didn't even moved forward, so it was useless. If you wait 1 sec before you move back, then yes, it was meaningful and the action was effective. For this limit I proposed 0.42 seconds (which can also be changed if justified).
Quickly adding waypoints which are not too close is effective.
|
Thank you for your good work. This is such a useful tool!
|
I have been using your Sc2Gears to save all my replays, and I wanted to issue a huge thank you for this awesome tool  .
I didn"t dive in it, but is it possible to have "standard old apm" in SC2Gears ? I want to compare my apm evolution from Warcraft 3 back in the days to right now, and I would like to compare those.
Thanks by advance, keep up your awesome work.
|
I like your idea and agree with pretty much everything. I just wish that it showed it in-game. Your way is our best bet though unless blizzard decides to change it to a normal minute and let people spam.
People in LR threads are already making fun of the new APM measurements from Blizzard.
|
When microing zerglings I tend to click really fast to one place, then if I have to re-micro them I'll click to another place. Between originally sending them and then resending them might be two or three seconds, but they'll only count as one action, not two with your method, I think, because of the continuous clicking.
|
Dakota_Fanning
Hungary2359 Posts
On September 27 2011 22:10 ArhK wrote:I have been using your Sc2Gears to save all my replays, and I wanted to issue a huge thank you for this awesome tool . I didn"t dive in it, but is it possible to have "standard old apm" in SC2Gears ? I want to compare my apm evolution from Warcraft 3 back in the days to right now, and I would like to compare those. Thanks by advance, keep up your awesome work.
In Sc2gears you can choose whether you want times and APM values to display using game-time or real-time (default) measurement. Go to the Analyzer tab of the Misc settings dialog.
If game-time is enabled, APM should be the old in-game APM, if real-time is enabled, it shows your real APM.
When EAPM will be included: calculating with game-time should result a value close to what Blizzards show currently in game; with real time it will show your real EAPM.
|
Dakota_Fanning
Hungary2359 Posts
On September 27 2011 22:11 Demonace34 wrote:
I like your idea and agree with pretty much everything. I just wish that it showed it in-game. Your way is our best bet though unless blizzard decides to change it to a normal minute and let people spam.
People in LR threads are already making fun of the new APM measurements from Blizzard.
As posted just above, you can tell Sc2gears to use either game-time or real-time for all time and APM values and calculations. Analyzer tab in the Misc settings dialog.
|
Dakota_Fanning
Hungary2359 Posts
On September 27 2011 22:12 Grumbels wrote:
When microing zerglings I tend to click really fast to one place, then if I have to re-micro them I'll click to another place. Between originally sending them and then resending them might be two or three seconds, but they'll only count as one action, not two with your method, I think, because of the continuous clicking.
It depends on how fast you do it. If you click 1 ms away, it will be considered ineffective, but it IS ineffective in fact. If you click like 0.5 seconds away, this algorithm will not consider them ineffective and will be counted as 2 separate actions toward EAPM.
|
Was this how BW programs calculated EAPM? For some reason just seems reminiscent and since it's BW I am partial so I will agree with your standards.
|
On September 27 2011 22:19 Dakota_Fanning wrote:Show nested quote +On September 27 2011 22:10 ArhK wrote:I have been using your Sc2Gears to save all my replays, and I wanted to issue a huge thank you for this awesome tool . I didn"t dive in it, but is it possible to have "standard old apm" in SC2Gears ? I want to compare my apm evolution from Warcraft 3 back in the days to right now, and I would like to compare those. Thanks by advance, keep up your awesome work. In Sc2gears you can choose whether you want times and APM values to display using game-time or real-time (default) measurement. Go to the Analyzer tab of the Misc settings dialog. If game-time is enabled, APM should be the old in-game APM, if real-time is enabled, it shows your real APM. When EAPM will be included: calculating with game-time should result a value close to what Blizzards show currently in game; with real time it will show your real EAPM.
Ok, thanks for the tips.
Regarding your last statement, does this mean once you will have implemented EAPM we won't be able to see "standard good old apm" through SC2Gears ?
|
On September 27 2011 22:01 Dakota_Fanning wrote:Show nested quote +On September 27 2011 21:51 whatthefat wrote:On September 27 2011 21:39 Dakota_Fanning wrote:On September 27 2011 21:33 whatthefat wrote:Interesting, I like the idea, but I had a few comments:
- Too fast repetition of some commands (~0.42 sec): move, attack move, set rally, stop, hold, hotkey assign
Does this affect waypointing? Also, what about micro situations where it is necessary to make a move command in one direction and then rapidly change direction, particularly when using faster units (e.g., mutas, lings or hellions to outmanoeuvre or bait slower units). EDIT: I may have misunderstood - by repetition, do you mean the same command in more or less the same location? - Too fast switch away from a selected unit or reselecting the same units without giving them any commands (~0.336 sec) (by too fast I mean there is not even time to check the state of the units and optionally react to it accordingly); double tapping a hotkey to center a group of units is NOT considered ineffective I'm still not particularly good at tapping regularly throughout a game, but I thought (and I could be wrong) that the idea was that by rapidly cycling through the list of all buildings/units you could peripherally follow the status of each. And thanks to persistence of vision, you can still view what each group shows even if you are moving between groups faster than you can process a single tap. For instance, if you were to extremely rapidly switch between two groups, you could effectively see both groups at once, superimposed (the same way that a fan looks like a solid disc once it spins fast enough). "Too fast repetition of some commands" refers to repeating the same command. For example if you micro a group of stalkers like this: Move, Attack move, Move, Attack move... These actions will all be considered effective. Right, but what if I have a group of lings outside a Protoss natural, and I click MOVE to mineral line, then very rapidly click MOVE back, so as to bait the Protoss defense. I'm clicking MOVE to 2 completely different locations - does that count as two actions or one? A similar situation can occur with mutas - I move them into a Terran base, then very rapidly move back - either to avoid a Thor I just spotted, or to try to bait a stim from marines. I would argue that these are both useful commands. Spamming to me is when MOVE is issued to the exact same LOCATION twice very rapidly in a row. Similarly with waypointing, I may very quickly input a path consisting of several points, e.g., for waypointing a drop, or for setting up a patrol.
About the tapping: the time limit (I proposed 0.336 seconds) can be decreased to cover cases where you cycle through your units/groups fast .
I agree, I'm just not sure what is an appropriate lower limit. It seems like some pros cycle through extremely quickly. I'm not sure whether they are purely spamming, or whether they are doing what I suggested (i.e., using persistence of vision to view more than one group at once). This comes down again to the time limit that will be used eventually in the implementation. If you move to the expand and 1 ms later you move back, then obviously your lings didn't even moved forward, so it was useless. If you wait 1 sec before you move back, then yes, it was meaningful and the action was effective. For this limit I proposed 0.42 seconds (which can also be changed if justified). Quickly adding waypoints which are not too close is effective.
Sure, there has to be some lower limit at which it becomes ineffective. But I think that limit is much shorter than 0.42 if location is not taken into account. If it is easy to implement, including location would make life a lot easier for you in determining whether an action is spam or not.
As for waypoints, they are set on the minimap for drops, meaning they can be very rapidly entered (much less than 0.4s per click) even though they cover a large distance.
|
On September 27 2011 22:27 ArhK wrote:Show nested quote +On September 27 2011 22:19 Dakota_Fanning wrote:On September 27 2011 22:10 ArhK wrote:I have been using your Sc2Gears to save all my replays, and I wanted to issue a huge thank you for this awesome tool . I didn"t dive in it, but is it possible to have "standard old apm" in SC2Gears ? I want to compare my apm evolution from Warcraft 3 back in the days to right now, and I would like to compare those. Thanks by advance, keep up your awesome work. In Sc2gears you can choose whether you want times and APM values to display using game-time or real-time (default) measurement. Go to the Analyzer tab of the Misc settings dialog. If game-time is enabled, APM should be the old in-game APM, if real-time is enabled, it shows your real APM. When EAPM will be included: calculating with game-time should result a value close to what Blizzards show currently in game; with real time it will show your real EAPM. Ok, thanks for the tips. Regarding your last statement, does this mean once you will have implemented EAPM we won't be able to see "standard good old apm" through SC2Gears ?
I believe it just comes up as a new selection, under that whole drop down menu thing.
|
|
|
|
|
|
EPT
