Hunter DPS Analyzer - Page 40

JSorrentino · 02/24/11, 1:58 PM

Whytefyst,

I think I found a bug in the fd site. Something is not properly computing my stats.

Site shows 2 variables (AGI and Crit) as being wrong.

Copied from site.
Agility: 4488 (Should be 4922)
Stamina: 5116
Attack Power: 190
Hit Rating: 961
Crit Rating: 1690 (Should be 1511)
Haste Rating: 1549
Mastery Rating: 752
Avg Item Level: 359

Gear all matches up in game / site for those stats.

Not sure what's going on, unless I am missing something. Gear was properly loaded, seems to be a calculation issue.

JS

EDIT: Average iLVL showing as 359 on site, should be 361 as well.

Rivkah · 02/24/11, 2:31 PM

The crit rating is just because the worgen bonus is being added in as 179 crit rating rather than as 1% crit (which amounts to roughly the same thing but confuses people). I do need to change that at some point but I've been swamped.

If those are from the gear stats and not the final stats, those stats are only what you gain from your gear, I don't think it includes racial agility and 5% agil bonus. I checked your armory and the gear you have equipped calculates out as 3506 agility, plus you have 387 agil in enchants, 535 agil in gems and 60 from socket bonus. That amounts to 4488 before any passive bonuses and 4712 with the 5% mail bonus. Once you include in your racial agility, the total is 4922 which matches your armory. If you disable averaging of abilities to turn off the agility proc bonuses, your final agility on the site comes out to 4922, which matches your armory.

With regards to avg ilevel, if you look at armory it says you have 359 ilevel equipped, 361 includes gear you may have in the bank or bags but are not wearing.

JSorrentino · 02/24/11, 2:54 PM

OK, thanks for clarifying. Sorry to make you do the work ;-)

alienangel · 02/24/11, 3:12 PM

Originally Posted by Whitefyst

...
- If your pet is a Ferocity pet, its crit chance is 9% higher at 47% resulting in a 88% uptime.
...

The second issue is whether you want to incorporate such a estimate into the simulation and by what method. This is where things get tricker.
- Do you just use the calculated uptime multiplied by your average DPS without it to determine your DPS with it. That is only a good option if your DPS and your crit rates are are constant, which they are not.

Since the numbers in the example seem to be based off my linked profile, the 38% crit is for my cunning pet (no other 4% physical debuff in our 10 man), so the 74% uptime estimate matches my logs quite closely.

Also IIRC this is how we modelled 2pT10 uptime and FI uptime bonuses in the spreadsheet, and I don't see why it's not acceptable as an estimate. On average the benefit it yields is the product of the average % uptime and the magnitude of the buff, unless it's a buff which has 2nd or higher order effects, which this does not. Working it into the simulation would be a lot of extra work for a very minor gain in accuracy for an effect that is only ever +/- 3%, and since FD uses similar uptime calculations to work out the average stat gain from a lot of trinkets that proc things like AP/Agi/Crit too, FD isn't tracking our crit rate changing from things like AGI/crit procs during the rotation simulation anyway, so trying to include CotH in the shot simulation wouldn't really change much - the sim in FD is mostly a GCD usage/autoshot timing simulation, not a full simulation of random events like Simcraft.

Originally Posted by Whitefyst

Hence, if all people want is a CtH estimate, that is great, but I would suggest reconsidering having CtH uptime worked into the simulation. Personally, I think that just assuming 100% uptime (which is not far from accurate from ideal situation modeled) is the safer route. That way it is applied at all times evenly instead of being susceptible to its uptime timing with events in your shot selections.

I disagree with this, we can easily calculate a more accurate uptime than 100% with an O(1) operation, since even with very good [current] gear it's not 100%. There is no reason to assume a clearly incorrect figure, especially when it directly affects the value of crit rating (for instance if for some reason in 4.1 we decide to reforge all our crit rating into mastery, leaving CotH uptime at a fake 100% would understate the actual loss in DPS from loss of crit). Additionally it brings to bear a non-trivial difference between Cunning and Ferocity pets, since as you point out Ferocity pets yield higher uptime.

Sorry for taking up all this space over this, don't mean to be ranty. I'm *not* asking Zeherah to code up a whole new sim of CotH procs - I'm just pointing out that FD seems to be assuming 100% uptime when I'm fairly sure that is not her intention (the old spreadsheet for instance didn't assume this, even though it and FD were used when we had 70+% crit rates and near 100% uptime anyway) - a good and likely quite simple fix would just be to the plug in the stochastic uptime estimate instead, and multiply herd_dmg_bonus by this fraction.

Rivkah · 02/24/11, 4:01 PM

I'm actually kind of surprised the existing culling the herd formula is as simplistic as it is on the site so I'm glad you brought it up. I thought I had implemented something a bit more advanced but looking at it, it could use some work. I will take a closer look at what sceone posted when I have a chance and see if I can improve it. Just becoming hard to find time to make all these adjustments but hopefully I can get to it this weekend. I don't plan to implement it as a part of the simulation, since that would be a lot of work for minimal gain, but I can certainly improve on what's there now.

Whitefyst · 02/25/11, 12:43 PM

As we dig deeper with the AI nerf's impact on MM, I believe the following generally summarizes when we want to use AI hardcast as the focus dump and when we want to use AS as the focus dump:

CA Phase: AI hardcast is the focus dump always - the 60% crit bonus makes it too good to pass up especially since hasted during a lot of this phase.

Standard, Termination, and KS Phase:
- AS as the focus dump when not under large dynamic haste effects (RF or BL)
- AI when under large dynamic haste effects (the faster cast times make AI higher DPS and closes the gap some on Sic'Em procs)

As such, would it be possible to add an option to use AI hardcast when under the effects of RF and BL.

Rivkah · 02/26/11, 7:35 PM

I've updated the Culling the Herd uptime code to be more accurate based on sceone's suggestions. Thanks all for your help with it!

With regards to adding additional options for rotation simulation, that's next on my list. I'm in the middle of cleaning up all the simulation code right now (it was a veritable rats nest) so I need to finish that and test it before I can add new features to it, but it should be a little easier for me to work with now. In the process I found a few small bugs potentially affecting ISS and LnL uptime that I fixed.

I will also note that for very complex rotation simulations, Simulationcraft is obviously going to be far better for testing, since it's more customizable and far more detailed in the simulation. It's probably far better to experiment there when trying to figure out new options for optimal rotations. That being said, modifications to rotation support on my site to allow it to better emulate recommended practices are definitely worth doing, since I want the dps estimates to come out as close to practice as possible, so I'll see what I can make work without making it too more complicated.

I'm thinking adding an option to specify the maximum speed of an aimed shot for hardcasting would be the best way to handle the dynamic haste effects situation. If set, it would suppress casting of arcane shot when aimed is below that speed to allow focus to be saved for aimed. Does this seem logical?

I'm thinking I would also add in 3 options for chimera shot:
1) default priority
2) delay chimera until.. 2.9 seconds or less are left on the sting during CA
3) disable sting/chimera during CA

The 2.9 seconds is because I want to make sure there's enough time to finish the current cast and still get off a chimera before the sting wears off- does that seem like it should be appropriate?

The other question I guess is whether to disable the save focus during CA for option #3. I guess I should, but in #2 obviously I can't or we'll end up not being able to fire chimera before the sting wears off.

Edit: I also fixed the bug with rapid recuperation giving too much focus that Whitefyst mentioned in the MM thread. This should result in some dps loss for MM specs.

Rivkah · 02/26/11, 10:42 PM

I've completed my cleanup of the simulation code and made it live. This shouldn't change dps calculated at all, but it should make modifications easier in the future (hopefully it'll run a bit faster too but I haven't done extensive testing on speed). It's possible a bug slipped through in the changes- so if you notice anything weirder than usual let me know.

Rivkah · 02/27/11, 1:16 PM

I've made some modifications to the autoshot loss calculation for aimed shot based on recommendations in the MM thread which should result in a slight loss of MM dps for hardcast aimed.

Rivkah · 02/27/11, 4:01 PM

I've added 2 new options to the MM settings under the shot priority- one to control when chimera is used during careful aim, and one to control whether to cast aimed shot manually based on speed.

When the aimed speed filter is enabled, arcane shot will not be cast during any time that aimed is below the speed maximum. This should allow for shot priorities that include arcane shot to appropriately switch between them based on haste effects. Note that by default the speed filter is set to never, which means it isn't used. To enable it, you have to switch it to either use the filter always or only when not in careful aim.

The chimera shot careful aim limitations are a bit flawed, in that I can only safely allow suppression of focus saving for the case when you disable sting/chimera completely. In that case the focus saving will re-enable 5 seconds before careful aim is set to expire (which seemed like a reasonable amount of time to start prepping for a normal rotation). If you are just delaying chimera till the sting is close to expiring, focus saving can't be suppressed or else you'd likely end up dropping the sting.

Hopefully these options will allow better reflection of the choices we face when playing MM. Additional suggestions are welcome but I'm trying to keep from making the options too overwhelming, so I'll likely only implement things which allow better emulation of best practices.

alienangel · 02/27/11, 5:35 PM

One thing that tripped me up for a while (in case someone else doesn't look at the debug output to notice it): the new "Maximum speed to cast Aimed Shot" thing only seems to be active if you have Aimed Shot behaviour to "Manual Cast and Instant" instead of the old "Manual Cast during CA and Instant". I didn't initially change the Aimed Shot behaviour and still got a DPS boost (probably from the CS changes) and thought it was simulating our new rotation, then noticed in the debug that it wasn't hardcasting even during RF/Hero.

Basically the options seem to mean "Manual Cast always during CA, and only below the specified Max Speed if one is specified outside CA, and Instant everywhere" and "Manual Cast only during CA and Instant everywhere".

Rivkah · 02/27/11, 6:11 PM

To clarify, the max speed option should only be applied if you're in a situation where manual casting is already enabled. So if you have it set to just instant cast, the max speed will be ignored. If you have it set to only manual cast during careful aim, it'll apply the speed filter during careful aim, but ignore it for the rest of the fight and revert to instant only casting. If you tell it to manual cast throughout the fight, then the max speed filter will be applied throughout the fight (except during careful aim if you told it not to). I realize it's a bit confusing but I was trying to figure out a way to make it work without drastically changing the first setting.

dvorjak · 02/27/11, 7:46 PM

With my current gear, and these settings:

Zeherah's Hunter DPS Analyzer

An AS focus dump always shows up as a significant dps loss. Even if I change the "Maximum speed to cast Aimed Shot (suppressing arcane shot):" to 1.6...which is my cast time when I have Bloodlust, ISS and 10%attack speed up. It shows as a 900dps loss.

Whitefyst · 02/28/11, 1:03 AM

Originally Posted by dvorjak

With my current gear, and these settings:

Zeherah's Hunter DPS Analyzer

An AS focus dump always shows up as a significant dps loss. Even if I change the "Maximum speed to cast Aimed Shot (suppressing arcane shot):" to 1.6...which is my cast time when I have Bloodlust, ISS and 10%attack speed up. It shows as a 900dps loss.

Did you remember to change glyphs? I changed them to Steady, Arcane, and Chimera.

Did you change Maximum speed to cast Aimed Shot (suppressing arcane shot): to "except during careful aim" since if you don't it will no longer cast hardcast AI during the CA phase after RF expires since you do BL late. I also suggest upping the 1.6 a little.

I did your character, and only got about a 440 DPS difference.

That is still a loss though, and I need to look into it more.

Izuul · 02/28/11, 11:15 AM

Currently the MM "Chimera Shot Careful Aim Behavior" setting is disabling ES and SrS during the SV test while set to "disable sting and chimera." The "Maximum speed to cast Aimed Shot" seems to be suppressing AS for SV as well.