Jump to content


Photo

[Balance] WrathCalcs


  • Please log in to reply
968 replies to this topic

#1 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 13 January 2011 - 07:19 PM

Just trying to make a unified place for talking about this. I'd put it off for a bit, but now with the PTR up, we're talking about an updating the sheet more, and the discussion is a bit scattered.

The a current version of the sheet generally lives here, but I often do small intermediate updates on this thread:
http://elitistjerks....on/#post2198943

And for the moment I'll upload the most current PTR working version (last updated from Tecton today).

Things going on right now:
--Fixing OpenOffice gem selection bug (done? Or does this still need one more fix, based on that reply in the TTT thread)
--Adding in stuff about multi-dotting (need to look at Erdluf's module)
--Have to see if there's a potential bug causing validation/formatting references to sometimes be lost.
--Long term, Armory import and maybe save/load profiles (presumably in Excel version only) would be features that help usability.

In general, questions/issues with the sheet can go here now. It's also a place to talk about new features we might want.

EDIT: Up-to-date sheet is on the main guide thread (link above). Smaller updates sometimes go here while we work on them.
  • Tearil likes this

#2 Moonwhisper

Moonwhisper

    Von Kaiser

  • Members
  • 46 posts

Posted 14 January 2011 - 04:42 PM

Things going on right now:
--Fixing OpenOffice gem selection bug (done? Or does this still need one more fix, based on that reply in the TTT thread)
...


Copied over from think tank thread for clarity reasons:

The gear selection page is ok till chest ... so the gems for the helm, neck, shoulder, cloak works ...but not downwards...
There still the Err502 error occurs
Perhaps some kind of table mismatch...it might have is origin in cell D26 or D27. The option menu is still available for cloaks but doesnt shows the correct options.

#3 Tecton

Tecton

    Soda Popinski

  •  Patrons
  • 3951 posts

Posted 14 January 2011 - 04:52 PM

Yeah, I noticed those when I got home to check it myself. I've got OpenOffice at home now, so as soon as it's properly working I'll upload another version (should get enough time this weekend to finish it off). Can I just say I hate OpenOffice?

As far as Armory import, I'm still intending to wait for them to enable the feeds from the new armory before implementing this. Saving and loading profiles should be easy enough to put together for Excel, I'll look at getting that in place.

#4 Moonwhisper

Moonwhisper

    Von Kaiser

  • Members
  • 46 posts

Posted 16 January 2011 - 06:14 PM

/comfort and big hug for help!!
If you need help with testing I am ready to assist.

#5 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 17 January 2011 - 06:51 AM

Quick PTR update. Not too much here, just wanted to put in a few minor fixes while I remembered, and posting it since I think Tecton is still working on a further fix for the gems issue.

Also, more notes on things to do sometime soon:
--Add in the rest of the random Conclave loot options (I already put in a few that I happen to have).
--Add in some estimate of Dark Intent DoT bonus.
--Adjust all ICD proc trinkets/enchants for nonzero proc times (has anyone checked on the actual proc rates of these things, and which ones can proc from DoT ticks?).
--Slightly more refined model for 4T11 bonus.

Attached Files



#6 Erdluf

Erdluf

    Great Tiger

  • Members
  • 968 posts

Posted 17 January 2011 - 02:08 PM

I'm attaching a sheet to show one way of handling temporary Haste procs (Bloodlust, NG, Hurricane, Trinkets).

This is not complete. A complete solution would also

- Feed results back to the rest of the sheet (particularly the Rotation tab).
- Pull trinket/Hurricane information from the Gear tab.
- Support stacking on-use trinkets with other procs (currently this models random on-cooldown procs).
- Support shorter fights (one Bloodlust/10 minutes is pessimistic).

As-is, if you fed numbers back to the Rotation tab, you'd have a circular dependency (NG uptime vs. Rotation duration). It is possible to break that dependency by removing the NG column, and instead making an NG table and a non-NG table, and recognizing that every (normal) rotation has 30s of NG.

Attached Files



#7 Tecton

Tecton

    Soda Popinski

  •  Patrons
  • 3951 posts

Posted 17 January 2011 - 02:36 PM

I'll be posting up a version today that I think should fix the OO issue, and just polishing off the profiles feature (which should let you save up to 9 gear/talent/raid buff combinations and load them up from the sidebar below the professions dropdown). I'm going to test this out on Wrathcalcs first then port to Treecalcs, if that's ok with you Hamlet?

[edit] Does anyone have a link to all the possible random loot that drops from Conclave/Al'akir?

Thanks Starfox.[/edit]

#8 Starfox

Starfox

    King Hippo

  • Members
  • 578 posts

Posted 17 January 2011 - 02:46 PM

I'll be posting up a version today that I think should fix the OO issue, and just polishing off the profiles feature (which should let you save up to 9 gear/talent/raid buff combinations and load them up from the sidebar below the professions dropdown). I'm going to test this out on Wrathcalcs first then port to Treecalcs, if that's ok with you Hamlet?

[edit] Does anyone have a link to all the possible random loot that drops from Conclave/Al'akir? [/edit]


Item suffix - Wowpedia - Your wiki guide to the World of Warcraft
Hello.
Light the fuse.
For all my homies.
Do not run, we are your friends.
SimulationCraft Druid Guy

#9 Tecton

Tecton

    Soda Popinski

  •  Patrons
  • 3951 posts

Posted 17 January 2011 - 04:36 PM

Ok, I should have finished the final gem dropdowns for OO (I hope) and the profile system is in place. There are two profiles currently, the one Hamlet had in the sheet (looked like roughly BIS) and a quick one for myself I put in so that people could check the swaps.

It saves/loads all of the user settings in the sheet, including rotation/spec/glyphs as well as all your gear. It's nearly instant for me, but let me know if there are any performance issues for anyone else (OpenOffice/Mac users, sorry, this won't work for you).

I'll update next time with the random gear/etc.

Attached Files



#10 zaulsiin

zaulsiin

    Glass Joe

  • Members
  • 9 posts

Posted 17 January 2011 - 05:07 PM

Ok, I should have finished the final gem dropdowns for OO (I hope) [...]


Unfortunately the OpenOffice gem and reforging issues remain for me in this build as well (11-01-17 PTR). Same 502 error for the gems and "reforge into" selections. OpenOffice v3.1.2, build 9502. If there's something particular that you'd like me to look into that might help out, please let me know, but I'm afraid I have little experience with OpenOffice myself, so I'm rather unable to take the initiative for this without some sort of direction. It just happens to be what we use at work. (I'm a paragon of workplace responsibility, to be sure.)

All of the effort from you guys thus far is much appreciated, and I'm happy to do anything that might help, just let me know.

#11 Tecton

Tecton

    Soda Popinski

  •  Patrons
  • 3951 posts

Posted 17 January 2011 - 05:32 PM

Sorry, forgot to upload with OpenOffice mode turned on (as you may have guessed, I edit it in Excel), if you go to charsheetdata and choose "YES" under OpenOffice Mode, they should work for you. Just checked it and it appears to be working.

#12 Fonzey

Fonzey

    Von Kaiser

  • Members
  • 47 posts

Posted 17 January 2011 - 05:43 PM

OpenOffice appears fine now Tecton, thanks for that.

On the Subject of the Armory Import, a feature that would be much appreciated for the record - would it be less work allow importing from an external profiling website such as chardev 8 - Cataclysm Beta as they tend to have their own armoury importer built in, along with additional features such as a nice easy GUI for switching items/gems/reforging options.

It may save on some long term administration work of WrathCalcs if you piggyback an external tool!

#13 zaulsiin

zaulsiin

    Glass Joe

  • Members
  • 9 posts

Posted 17 January 2011 - 06:12 PM

Sorry, forgot to upload with OpenOffice mode turned on (as you may have guessed, I edit it in Excel), if you go to charsheetdata and choose "YES" under OpenOffice Mode, they should work for you. Just checked it and it appears to be working.


Ah, cheers, that appears to have done it. My own fault for just skimming through in a hurry, I suppose.
Thanks!

#14 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 17 January 2011 - 08:06 PM

I'm attaching a sheet to show one way of handling temporary Haste procs (Bloodlust, NG, Hurricane, Trinkets).

This is not complete. A complete solution would also

- Feed results back to the rest of the sheet (particularly the Rotation tab).
- Pull trinket/Hurricane information from the Gear tab.
- Support stacking on-use trinkets with other procs (currently this models random on-cooldown procs).
- Support shorter fights (one Bloodlust/10 minutes is pessimistic).

As-is, if you fed numbers back to the Rotation tab, you'd have a circular dependency (NG uptime vs. Rotation duration). It is possible to break that dependency by removing the NG column, and instead making an NG table and a non-NG table, and recognizing that every (normal) rotation has 30s of NG.


It's a nice way of setting up the basic calculations (although the binary scheme might get big if you have to keep adding things--already there's Berserking to put in). The problem I always had though is the structural issue--avoiding circularities. I don't quite get what you're suggesting in the last line. Thinking about it though, I could do this:

--Compute rotation length in the same way as now. Used average values for haste procs.
--Use this new table to compute the mean number of DoT ticks, and feed that back in solely for the purpose of DoT damage.
There's still one circularity left though, which is that rotation length depends on cast time ratio of DoT's, which depends on mean DoT duration, which depends on NG (in the new page), which depends on rotation length. At worst, some estimate of NG uptime can be used for this purpose without introducing large errors.

That would fix the only current problem with temporary haste, which is that DoT tick numbers come out wrong. If we wanted to start worrying about syncing haste effects, that's a separate problem in the first place--we'd need some way to model the overlap to begin with.

e: okay I think I get what you're saying about the two tables. Might be needed anyway, since IS and MF have different NG uptimes.

#15 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 17 January 2011 - 11:23 PM

Few updates:
--Dark Intent added, with user-selected uptime.
--Added a simple Sunfire/IS spam model. Just steady-state, N targets, 100% DoT uptime. So a bit optimistic and ignores ramp-up, but gives some basic results: roughly 80% DPS on a single target, a little over 100% DPS on two targets (again, provided you maintain perfect DoT uptime and don't clip IS), and substantially better on more targets. This is without any spells besides IS and Sunfire at the moment. Worth noting that BotG is actually a pretty bad DPS increase even here.
--Improved 4T11 bonus to account for Moonfury and Eclipse. Suddenly the 4T11 bonus looks extremely good, and that's not even taking into account any added DoT crits or the possibility of Starsurge crits. It does assume you don't waste charges on Moonfire.

I didn't put in added DoT crits for 4T11 because DoT refreshing becomes a bit weird. If you have a leftover DoT when Eclipse starts, it gets the benefit. If not, then IS gets the benefit if you refresh immediately, but MF won't get any benefit, because you're going to wait to cast it until the bonus falls off. And in the steady state, we tend to not have DoT's leftover when Eclipse starts, because they usually fall off right before. Maybe I should just start assuming a "refresh on each Eclipse" setup as the default behavior, rather than a "refresh always" setup. With 4T11, I'm pretty sure refreshing each DoT once per Eclipse is best anyway--IS immediately and MF after 3 casts. The only issue with this model is that it relies on perfectly stationary casting.

Attached Files



#16 Guest_aceofsween_*

Guest_aceofsween_*
  • Guests

Posted 18 January 2011 - 12:55 AM

Coming across a small error with regards to DoTs and haste. It appears that the DoT tick breakpoints are off.

I have 1928 Haste as of right now. Normally, I get 11 ticks in game, which WrathCalcs agrees with. However, when NG is up, I am getting 12 ticks only, although WrathCalcs estimates 13 ticks. It is definitely getting the tick interval correct (1.44 and 1.66 respectively) but in this case, the game does not allow for the 13th tick (would be a ~18.72 second duration total).

#17 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 18 January 2011 - 01:37 AM

Probably from averaging in Troll Berserking. That's part of the general temporary-haste problem.

e: although, 1928 should get 13 ticks anyway, hmm. There might be some rounding going on in-game.

#18 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 18 January 2011 - 02:25 AM

After playing with this a bit with some creative reforging, it seems I get the 13th tick at 1930 haste but not at 1926. Assuming your 1928 test is right, the real breakpoint is 1929 or 1930. So there is definitely some kind of rounding going on in the in-game computations. We'd known this somewhat because I think someone mentioned pre-4.0 that a Rejuv breakpoint was a little off, but never really looked into it.

For one idea: 1929 is the first haste level at which the tick rate of IS, rounded to the nearest millisecond, is 1.439, and 1.439 is the first millisecond at which you get a 13th tick. Maybe this is a hypothesis worth checking for other spells.

#19 Tecton

Tecton

    Soda Popinski

  •  Patrons
  • 3951 posts

Posted 18 January 2011 - 09:16 AM

Looks like some of the conditional formatting references got broken with that last save, going through and updating now. Updating the profile function for the new fields, too.

[edit] Yeah, a ton of references have disappeared, so regemming is broken. Will get it fixed today, just posting so people know I'm aware of it. [/edit]

#20 Hamlet

Hamlet

    Mike Tyson

  • • Guide Author
  • 11581 posts

Posted 18 January 2011 - 09:39 AM

In the last one I uploaded? Hmm, I didn't even notice. Like I'd mentioned, that happened once before and I don't know what's causing it. At time I just backed up a save. You can look at the notes in my post above or compare to the previous day's version--I don't recall changing anything on "your" pages (front page, gear list, charsheetdata). In prior updates I'd added some items manually (and even a meta gem), but not between these last two versions.

I know it's kind of a pain to put them all back in--although maybe going through and putting them all in once will fix whatever the problem is.

Here's the only detail I can think of:
The worst culprits were always cells I94 and I106 on the front page. The last reference in their conditional format (to G94/G106) would break to #REF sometimes, but I never figured out a repro. Once just I fixed them manually with $'s in them (as you see now), and I don't think the broke after that. But now looking at the 0117 one I uploaded before, a different reference in the I94/I106 conditional formats broke. Something about those two cells. This more drastic version where all the gem-related things broke I only saw once before.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users