SimulationCraft Model Development - Page 8

dedmonwakeen · 03/18/09, 2:03 PM

SimulationCraft r1882 available for download.

3.0.9 report: SampleOutput - simulationcraft - Google Code

3.1.0 report: SampleOutputPTR - simulationcraft - Google Code

All default profiles now use real gear-sets (with chardev links). We did our best to put BiS sets together. If you see any problems, be sure to let us know.

EDIT: The DPS-spread in the 3.0.9 report is rather tight...... (excluding DG shenanigans, of course!)

Starfox · 03/19/09, 1:34 AM

Saw the tier 8 bonuses/relics on mmo-champion.
Implemented in trunk right now are
Druid

Add Idol of the Corruptor (Need procchance/cd from PTR!)
Add Idol of the Crying Wind
Add Blanace Tier 8 2 piece bonus
Add Balance Tier 8 4 piece bonus (Need procchance/cd from PTR!)
Add Feral Tier 8 4 piece bonus
Starfire: Add option to take advantage on the new set bonus (instant=1)

I need actual numbers on the feral 2 piece bonus.

Priest

Add Shadow Tier 8 2 piece bonus

Wording of 4 piece bonus is borked, 15 haste rating? Need data on chance/cd and what it really is.

Rogue

Add Tier 8 bonuses

Shaman

Add Tier 8 bonuses for elemental and enhancement
Add Thunderfall Totem
Add Totem of the Dancing Flame

Need information on tier 8 4 piece elemental, time between ticks? Overall duration? I just assumed that it's a rolling DoT.

Warlock

Add Tier 8 bonuses

Need information on tier 8 2 piece bonus, is it additive to talents? Or multiplicative at the end?

• Zakalwe · 03/19/09, 8:37 AM

Originally Posted by Starfox

Warlock

Add Tier 8 bonuses

Need information on tier 8 2 piece bonus, is it additive to talents? Or multiplicative at the end?

I've changed your assumption from multiplicative to additive, because that's how glyphs work. Also because I'd rather underestimate than overestimate the value.

As an aside, I'm also assuming the 2 piece bonus will affect conflag damage, because the immolate glyph does now.

Starfox · 03/19/09, 10:49 AM

Elemental Focus Stone
Pyrite Infuser
Eye of the Broodmother

Are now in trunk

Corbetti · 03/24/09, 3:17 PM

Mac OS X binary now on downloads page.

Althor · 03/24/09, 8:17 PM

Priest Tier 8 4 piece is in the trunk. By default it assumes it procs instantly after landing a Mind Blast though there is support added to have it proc after an optional delay.

Althor · 03/26/09, 9:02 PM

Trunk is updated to build 9733.

Maje · 04/12/09, 3:25 PM

I've been going over Simcraft configuration files and it seems mage T8 is actually mage T7 ie. stats haven't changed at all and the mage is still using 4t7 on the other hand warlock model was updated.

First, I'm not judging or accusing anyone as the work is done in everyone's free time and is greatly appriciated, that said, there is no reason to show two classes on the same chart when obviously they aren't on the same footing, many people are looking at the charts and are basing their decisions on them. Not to mention zillions of others that take one look at them and go straight QQing to the blizzard forums.

A suggestion: it's fine that the mage gear/stats file wasn't updated, just make sure to add a comment to every model that doesn't use T8 gear that it doesn't or even just list the classes as T7 since that's what they are.

Starfox · 04/12/09, 4:30 PM

T8 files right now are mostly T7 files simulated with 3.1.0 mechanics.
Compiling the gearsets and updating them with every new item that is discovered on ptr is a bit tedious

Maje · 04/12/09, 4:44 PM

Originally Posted by Starfox

T8 files right now are mostly T7 files simulated with 3.1.0 mechanics.
Compiling the gearsets and updating them with every new item that is discovered on ptr is a bit tedious

I understand completely, yet my point remains, even if one model uses T8 gear set; placing it on the same chart with others is missleading and stands completely opposed to what the Simcraft is supposed to do (ie. simmulate rdps with classes compared).

dedmonwakeen · 04/12/09, 8:45 PM

Originally Posted by Maje

I've been going over Simcraft configuration files and it seems mage T8 is actually mage T7 ie. stats haven't changed at all and the mage is still using 4t7 on the other hand warlock model was updated.

First, I'm not judging or accusing anyone as the work is done in everyone's free time and is greatly appriciated, that said, there is no reason to show two classes on the same chart when obviously they aren't on the same footing, many people are looking at the charts and are basing their decisions on them. Not to mention zillions of others that take one look at them and go straight QQing to the blizzard forums.

A suggestion: it's fine that the mage gear/stats file wasn't updated, just make sure to add a comment to every model that doesn't use T8 gear that it doesn't or even just list the classes as T7 since that's what they are.

The naming oddness is all my fault: The "T8" naming came out before T8 gear was really known..... The only difference between T7 and T8 is 3.0.9 mechanics vs 3.1.0 mechanics and new glyphs.

The T8 profiles for ALL classes should be T7 BiS gear with 3.1 mechanics. In retrospec, I should have used T7_310 instead of T8.

Now that T8 gear is reasonably well known, we should probably develop three sets of profiles:

T7 => Evaluate BiS for Live
T7_310 => Evaluate BiS for when 3.1.0 goes Live and people want to change gear/specs (start of gear progression)
T8 => Evaluate Tier8 BiS to see where classes will end up eventually (end of gear progression)

Maje · 04/12/09, 9:05 PM

Yep, I completely agree, I don't mind one model being more developed then other etc. but in this case some models had their stats updated to T8 level while others haven't and they are still shown on the same chart and that what bothers me. I would bet it would bother a lot more people if they knew that's is how the PTR output is rendered.

But as you say just editing the names to show what the classes are using should be enough.

ramenchef · 04/12/09, 9:13 PM

Which armor penetration model is the latest simulationcraft using? The tooltip numbers, the 81% of the tooltip bugged numbers, or the 3.0.9 formula?

dedmonwakeen · 04/12/09, 10:45 PM

Originally Posted by Maje

Yep, I completely agree, I don't mind one model being more developed then other etc. but in this case some models had their stats updated to T8 level while others haven't and they are still shown on the same chart and that what bothers me. I would bet it would bother a lot more people if they knew that's is how the PTR output is rendered.

But as you say just editing the names to show what the classes are using should be enough.

I'll check with the devs of each module, but "T8" is supposed to mean "T7 with 3.1 mechanics" for all classes...... If this is not the case, then I need to know which models are out of sync. For example, the Warlock T8 profiles still invoke the Warlock_T7_Gear.simcraft config file.....

If the devs each feel there is sufficient information regarding T8 level gear for their class then we will make the switch this week...... for ALL classes.

Originally Posted by ramenchef

Which armor penetration model is the latest simulationcraft using? The tooltip numbers, the 81% of the tooltip bugged numbers, or the 3.0.9 formula?

This is what we currently use. (Druids/Shamans tweak the "attack_haste" field after init.)

static rating_t& rating_80()
{
  static rating_t r;

  if( r.spell_haste == 0 )
  {
    r.spell_haste       = 3279;
    r.spell_hit         = 2623;
    r.spell_crit        = 4591;
    r.attack_haste      = 3279;
    r.attack_hit        = 3279;
    r.attack_crit       = 4591;
    r.expertise         = 3279;
    r.armor_penetration = 1232;
  }

• Zakalwe · 04/13/09, 7:06 AM

It's actually all my fault - after implementing the 3.1 warlock changes I wanted to check in profiles that made use of the new specs/rotations, so I took the existing Warlock_T7_XX_XX_XX.simcraft file, tweaked them, and called them Warlock_T8_XX_XX_XX.simcraft. This wasn't discussed or thought through at all, but the other devs followed my lead.

I do see how it can be confusing, but just to make it completely clear: ALL [class]_T8_XX_XX_XX.simcraft profiles use T7-level gear. There's no gear-based "unfairness" in any of the charts you've seen.

Seleyana · 04/16/09, 3:49 AM

Just a short question, and don't worry, it will (probably) be the only on such a low level:
Is a SimulationCraft documentation available anywhere? I just stumbled over this program a few weeks ago and I am not very familiar with the notation of C++ (I've been forced to Fortran)... or is the answer just: learn C++ and read the code!

dedmonwakeen · 04/16/09, 4:09 AM

Originally Posted by Seleyana

Just a short question, and don't worry, it will (probably) be the only on such a low level:
Is a SimulationCraft documentation available anywhere? I just stumbled over this program a few weeks ago and I am not very familiar with the notation of C++ (I've been forced to Fortran)... or is the answer just: learn C++ and read the code!

If you go to the main site, you find a small amount of user-level docs on the "Wiki" tab.
It is -mostly- up to date, but to be honest, your best bet is to look at the example .simcraft files to see ways to drive the sim.

If you are looking for developer documentation I'm afraid you have to dig through the code. The core thing to realize that it is an event-driven simulation. Combat begins by each player being kicked off with a "player_ready" event. This makes a player walk through his action list and it will schedule for execution the first action that is "ready". When an action completes, it schedules another "player_ready" event, and so it continues....

Fizzl · 04/23/09, 3:32 PM

Using

tier8_2pc=1
tier8_4pc=1

In a mage 20/51/00 scale factors made it recommend stamina and expertise, I guess they don't work yet :O

____

The weighting behind spirit seemed a bit high so I turned on infinate mana (in case it was factoring in regen) and removed student of the mind from my usual build giving me Talent Calculator - World of Warcraft
I was expecting 1 spirit to then be worth 55% of 1 crit but it was more like 65%.
In this case .71 spirit / 1.1 crit = .645
Seems like spirit isn't quite right yet or am I doing it wrong?

Starfox · 04/23/09, 4:12 PM

Originally Posted by Fizzl

In a mage 20/51/00 scale factors made it recommend stamina and expertise, I guess they don't work yet :O

I get

Scale Factors:
  Mage_T7_20_51_00           strength=0.00  agility=0.00  stamina=0.00  intellect=0.39  spirit=0.82  spell_power=1.57  attack_power=0.00  expertise_rating=0.00  armor_penetrating_rating=0.00  hit_rating=2.62  crit_rating=1.18  haste_rating=0.78

with your profile

Fizzl · 04/23/09, 4:13 PM

Ran it without the T8 bits and it stopped offering me stamina and expertise.

dedmonwakeen · 04/23/09, 7:16 PM

Originally Posted by Fizzl

Ran it without the T8 bits and it stopped offering me stamina and expertise.

The default setup of 10k iterations is "barely acceptable" for scale factor generation...... but I'll look into the T8 bonuses just to make sure.

Kalku · 04/23/09, 8:48 PM

Mage T8 profiles use T7 gear stats.

Edit: I know this is known (at least to you guys), but until they are updated maybe don't put the T8 mage profiles in the release, since they are not T8 profiles.

dedmonwakeen · 04/24/09, 9:04 AM

Originally Posted by Kalku

Mage T8 profiles use T7 gear stats.

Edit: I know this is known (at least to you guys), but until they are updated maybe don't put the T8 mage profiles in the release, since they are not T8 profiles.

That's a good idea. I keep thinking the Mage T8 profiles are "about to get done" so I don't take action on the fact that they are "not done yet".

Kalku · 05/06/09, 1:03 AM

target_health and max_time do not seem to work.

I set max_time=15, with 10000 iterations, and got a 300 second sim.
I set max_time=100, with 10000 iterations, and got a 300 second sim.
I set target_health=100, 10000 iterations, got an 18 second sim.
I set target_health=100, 100000 iterations, got an 101 second sim.

• Zakalwe · 05/06/09, 10:04 AM

Originally Posted by Kalku

target_health and max_time do not seem to work.

I set max_time=15, with 10000 iterations, and got a 300 second sim.
I set max_time=100, with 10000 iterations, and got a 300 second sim.
I set target_health=100, 10000 iterations, got an 18 second sim.
I set target_health=100, 100000 iterations, got an 101 second sim.

My guess would be that max_time is being set to the default 300 somewhere else in your config file layout - perhaps by including Global_T8.simcraft after your max_time line? This would prevent your max_time settings from having any effect.

Your target_health settings would then be applied *together* with the default max_time setting, and setting both of those config switches for the same simulation doesn't appear to be properly supported. Nor am I sure how that could be supported, since you'd be giving the simulator two conflicting instructions.