What if...ZOS ran a script that halved all player damage, heal, and shield tooltip values...

KenaPKK · August 2016

...and all pve mob health values, and then removed Battle Spirit effects on damage, heals, and shields?

Exactly how many fewer server side calculations would we have to process? Would PvP be more approachable since numbers versus players would align with numbers against AI?

Or am I just half asleep and delirious?

Discuss.

jeedrzej · August 2016

I dont think this is a source of problems

Kartalin · August 2016

jeedrzej wrote: »

I dont think this is a source of problems

But is it more or less serious than the deer of Cyrodiil?

jeedrzej · August 2016

Kartalin wrote: »

But is it more or less serious than the deer of Cyrodiil?

Removing deer was rather... safe. Changing gamesystem is... well. More dengerous. And can be source of bugs.

Draxys · August 2016

jeedrzej wrote: »

Kartalin wrote: »

But is it more or less serious than the deer of Cyrodiil?

Removing deer was rather... safe. Changing gamesystem is... well. More dengerous. And can be source of bugs.

Devs seem to be perfectly fine with tons of bugs as it is, might as well try it at this point.

Performance in cyrodiil has only been worse once, in my opinion. That was the fps bug of 1.2 or 1.3. Yea we had big lag battles after that but today is just as bad.

jeedrzej · August 2016

Draxys wrote: »

Devs seem to be perfectly fine with tons of bugs as it is, might as well try it at this point.

Well... I agree. There are a lot of bugs in game but... it is not a reason to add more.

Draxys · August 2016

jeedrzej wrote: »

Draxys wrote: »

Devs seem to be perfectly fine with tons of bugs as it is, might as well try it at this point.

Well... I agree. There are a lot of bugs in game but... it is not a reason to add more.

Not a reason to add more, no, but considering that fact, I'm not sure it would hurt to look into the possibility. Maybe try something like it on the pts campaign,

jeedrzej · August 2016

Draxys wrote: »

Not a reason to add more, no, but considering that fact, I'm not sure it would hurt to look into the possibility. Maybe try something like it on the pts campaign,

Yup. Its good idea.

But... Isn't it waste of time if we aren't sure it would help?

Maybe better find more reliable way.

Although, ZOS is testing a lot of solution. Maybe they will test this also.

Asmael · August 2016

It would technically make a pretty small difference, since you're not reducing the amount of calculations but only the size of the numbers manipulated, which, to a computer, is not as relevant as it is to a human.

That said, #RemoveBattleSpirit and rework the numbers. We can keep the 20% damage reduction (and maybe the extra health), but the reduction to shields and healing mess up %based values waaaaay too much.

KenaPKK · August 2016

Asmael wrote: »

It would technically make a pretty small difference, since you're not reducing the amount of calculations but only the size of the numbers manipulated, which, to a computer, is not as relevant as it is to a human.

That said, #RemoveBattleSpirit and rework the numbers. We can keep the 20% damage reduction (and maybe the extra health), but the reduction to shields and healing mess up %based values waaaaay too much.

Every single damage value calculated while in Cyrodiil has an "if against a player, divide by 2, if against an AI, proceed" step, and every heal and shield undergoes the divide by two calculation by default.

Also outright removing the heal and shield reductions would throw Magplar and magicka sorc way it of whack, although I do agree that % based values don't work well under the current Battle Spirit. That said, there are nigh unkillable 60k health Dragonblood DKs and Blazing Shield Templars running around in Cyrodiil. Changes to % based values would have to be made surgically to achieve balance.

Sugaroverdose · August 2016

KenaPKK wrote: »

...and all pve mob health values, and then removed Battle Spirit effects on damage, heals, and shields?

Exactly how many fewer server side calculations would we have to process? Would PvP be more approachable since numbers versus players would align with numbers against AI?

Or am I just half asleep and delirious?

Discuss.

actually, i did think the same way for the whole time, there's no reason to have different damage in pvp and pve, also it would give more clear view of class and stamina/magicka balance

Kas · August 2016

from a computational point of few, cutting some values in half (can be in the end on a number of HP, so it's actually just integer division by 2, i.e. ">> 1") is really cheap. comparing this to deciding which targets (even in a 2D environment but there even seems consideration of a z axis) are hit by a circular AoE is a bit like comparing the money that can be saved from trying to use as little toilet paper when you always take your car to get home from work to take a dump.

thus, battle spirit would only effect performance if it's done "wrong". the dear could have actually been more of an issue because they would have to be considered for the "which targets hit by AoE?" question. Next to AoE there's client server communication, checks to prevent manupulations (i.e. cheats) and so on. there are tons, and tons of computationally hard things to get right. division by to in the right places is rather easy.

sure, it feels like LOTS of divisions throughout cyrodiil and lots of such cheap things add up. however even a couple of billion computations within the hour are virtually nothing. modern processors can perform more than one instruction per cycle and "Ghz" actually means billion cycles per second. algorithms whose cost grows more than linear in the number of affected targets tend to reach much more computations very quickly.

tl;dr: I would not expect a significant impact on performance if the battle spirit is done right in its current state.

Sugaroverdose · August 2016

Kas wrote: »

from a computational point of few, cutting some values in half (can be in the end on a number of HP, so it's actually just integer division by 2, i.e. ">> 1") is really cheap. comparing this to deciding which targets (even in a 2D environment but there even seems consideration of a z axis) are hit by a circular AoE is a bit like comparing the money that can be saved from trying to use as little toilet paper when you always take your car to get home from work to take a dump.

thus, battle spirit would only effect performance if it's done "wrong". the dear could have actually been more of an issue because they would have to be considered for the "which targets hit by AoE?" question. Next to AoE there's client server communication, checks to prevent manupulations (i.e. cheats) and so on. there are tons, and tons of computationally hard things to get right. division by to in the right places is rather easy.

sure, it feels like LOTS of divisions throughout cyrodiil and lots of such cheap things add up. however even a couple of billion computations within the hour are virtually nothing. modern processors can perform more than one instruction per cycle and "Ghz" actually means billion cycles per second. algorithms whose cost grows more than linear in the number of affected targets tend to reach much more computations very quickly.

tl;dr: I would not expect a significant impact on performance if the battle spirit is done right in its current state.

That's why we see race conditions on server-side(macroslise, which actually says that player input got calculated asynchronously or in multiple threads, both ways are just too complicated to sync, can't really understand what was on dev mind when he did that), and server-side freezes(any lag), because powerful hardware forgives too much and in result we have non-optimal code.

Any calculation isn't free anyway, even changing value type. And when numbers of such calculations grows to thousands, you'll see difference.

jhharvest · August 2016

I'm pretty sure if there is a server performance bottleneck it's not from simple math operations. It's probably geometry calculations like: line of sight, pathfinding, range checks (depending on how optimised / approximate it is*), location and speed calculations (for anti-botting) etc. and possibly AI operations.