70-79 Level 70 PvE BiS Gear Compendium

[Dreams]

Hey @Symandria the respect is mutual. It's awesome to find other people with the same interests. Don't take my directness as hostility.

I really recommend you look into simcraft because it is 100% transparent. It gives you a complete breakdown of which abilities were used in which order and breakdowns of buff uptimes and proc frequencies over tens of thousands of parses. I know Fx spent a long time tweaking the sim on my mage to accurately represent a 5 man encounter VS a TBC boss that would last ~3 minutes. To to be honest... Why would you ever trust the simmed stat values of a program you know for a fact doesn't take every minute detail into account?

In my personal testing for Ray of Frost the last tick always seemed the same base damage value as the one before it so I see no evidence of the partial ticks you indicate. I would test it again but at this moment my account doesn't have a subscription.

Here you go:

 

[Fx]

500.000 parse Frost Mage simulation with optimized secondaries

 

[Dreams]

Very nice opening damage of 50k, I don't know the circumstances that's achieved under, I do know I low balled mine a normal fight opener is more like 44k self-buffed including combat consumables.

Flask+potion+drums (would be higher with real bloodlust)

 

[Symandria]

Hey

First off let me say, I'm very impressed

I don't have time to review all the numbers and figures right now, but there is a very good chance using a more accurate model you guys have gotten better results!!

I may have to start using Simcraft looks very good! And Onlydreams since i wrote mine i do know every detail for how it works, but my guestimations that go into it don't have the level of precision that a real simulator has.


I still do have a question though and that is about ilevel and the gear pieces i selected While we both agree ilevels power growth is linear, I believe it still suffers from diminishing returns like any multiplier. Unless you are telling me each extra +1 avg item level adds * 1.0166 to the equation, then if you are already at say ilevel 154 your current bonus is 64.74% for a multiplier of 1.6474 so increasing the multiplier to 1.664 is effectively around 1.01% raw improvement to damage. So i think if you crunch the numbers you'll still find those picks are an improvement

P.S. I loved your short video proofs, I wonder if the coding is different in eu or if they changed it, the last tiny tick was very easy to see!

 

[Fx]

I still do have a question though and that is about ilevel and the gear pieces i selected While we both agree ilevels power growth is linear, I believe it still suffers from diminishing returns like any multiplier. Unless you are telling me each extra +1 avg item level adds * 1.0166 to the equation, then if you are already at say ilevel 154 your current bonus is 64.74% for a multiplier of 1.6474 so increasing the multiplier to 1.664 is effectively around 1.01% raw improvement to damage. So i think if you crunch the numbers you'll still find those picks are an improvement

I think you're mixing up two different themes in mathematics.

The logic you're using applies to Critical Strike. If you have 0% critical strike, 1% critical strike will equate to 1% more damage. But then you're at 1% critical strike, and so adding 1% crit won't give you 1% more damage, but, in fact, 0.99%. This is because it's not a linear function.

The ilvl damage increase, however, is linear, and so this logic shouldn't apply.

Either way, to use the example of Belt of Blasting, it is worth considering -if- you're not dropping below a whole item level decimal (.00). We already established this. We haven't looked at spec-specific total ilvls yet so we haven't worked out which items could be worth replacing. It's next on the to-do list.

The way we test items and enchants vs eachother is to just run simcraft with the two different builds and see which one comes out on top once we apply ilvl manually. And common sense / regular math of course.

 

[Dreams]

@Symandria
My current equipped ilvl is 160,18
If I replaced my tempest belt with belt of blasting it would drop by 1,625 like I mentioned earlier.
160,18 - 1,625 = 158,555
Meaning I drop two whole ilvl numbers.

Now, if my ilvl had been 160.625 or a higher decimal, I would only drop one whole ilvl by equipping belt of blasting, and then it would be worth it.

 

[Dreams]

I just did the item level decimal experiment again and took proper notes this time. I'll post in here:

Experiment goal: To figure out how the instanced PvE ilvl boost mechanic treats ilvl decimals.

Procedure:
1. Test openworld damage.
2. Calculate expected instanced damage in 3 different scenarios:
a) Your item level is rounded up.
b) Your item level is rounded down.
c) Your item level is used at face value including decimals.
3. Test instanced damage.
4. Compare instanced damage results to calculated instanced damage numbers.

Method: Sampling frostbolt damage. Using level 70 mage with equipped ilvl of 151,62

Openworld damage tests:
30 casts: They all hit for 1742

Expected instanced damage:
a) Item level is rounded up:
(152 - 115) * 1,666 = 61,642%
1742 * 1,61642 = 2815,8 (probably 2816 since damage doesn't work in decimals)

b) Item level is rounded down:
(151 - 115 ) * 1,666 = 59.976%
1742 * 1,59976 = 2786,78 (2787)

c) Item level decimal is used:
(151,62 - 115) * 1,666 = 61,008%
1742 * 1,61008 = 2804,75 (2805)

Testing instanced damage:
All casts hit for 2788

Conclusion: the ilvl boost takes your last whole item level number and completely ignores decimals.

 

[Symandria]

@Fx Respectfully you seem to be wrong. I believe ilevel is not linear in the way you think it is, and it can be proven one way or the other

Let's say we have this equation: Int * (1 + crit) * (1 + haste) * (1 + (1 + ((avgilevel - 115) * 0.0166)
Yes, I know that it's a slight oversimplification of the way int crit and haste work since they all have a base weight of 1 in this sample equation. But our focus is ilevel so it should work.

Now let's assign the value 1000 to Int, .30 to haste, and .20 to crit.

At ilevel 115 we get: 1000 * 1.3 * 1.2 * 1 = 1560
At ilevel 150 we get: 1000 * 1.3 * 1.2 * 1.581 = 2466.36 //Here we see the linear increase you expect! 35 ilevels give us 58.1% damage increase

At ilevel 151 we get: 1000 * 1.3 * 1.2 * 1.5976 = 2492.256
//Here with a 1 ilevel gain we see an increase of (2492.256 / 2466.36 = 1.0105). This says that we gained 1.05 percent damage for this ilevel increase.

At ilevel 161 we get: 1000 * 1.3 * 1.2 * 1.7636 = 2751.216
//10 ilevels gives us 2751.216 / 2492.256 = 1.1039, an increase of 10.39%.


Now let's work it the other way and see what it would be if it were in fact a linear increase!

If it were a linear increase each time we gain an ilevel we would gain 1.66 percent damage so the equation looks like this:
Int * (1 + crit) * (1 + haste) * (1.0166 ^ (avgilevel - 115))

At ilevel 115 we get: 1000 * 1.3 * 1.2 * 1 = 1560
At ilevel 150 we get: 1000 * 1.3 * 1.2 * 1.7793 = 2775.708
At ilevel 151 we get: 1000 * 1.3 * 1.2 * 1.8089 = 2821.785
//1 ilevel gives us an increase of 2821.785 / 2775.708 = 1.0166, or an actual increase of 1.66%.
At ilevel 161 we get: 1000 * 1.3 * 1.2 * 2.1326 = 3326.821
//10 ilevels gives us an increase of 3326.821 / 2821.785 = 1.1790, or an actual increase of 17.9% which does match up with 1.0166 ^ 10 = 1.1790

So in practice you can take your ilevel and see if it acts whether ilevel is boosting your damage by 1 + ((avgilevel-115) * 0.0166) or 1.0166 ^ (avgilevel - 115). Then you will know if it's linear exponential growth or diminishes like an additive multiplier.


@Onlydreams By my calculations which granted may be imperfect you gain 2.21% damage from the belt of blasting, at ilevel 160 you have a damage multiplier of 1.747 from ilevel so each increased ilevel will give you another 1.0095% damage, so losing 2 ilevels is trading 2.019% damage. And for the shoes you gain 1.25% and would lose another ilevel for a trade off of - 1.0098% (this number is a little larger since you would be 2 item levels lower already.).

Once Fx verifies it is a additive and not exponential multiplier, you can compute the stat value of the stats for the items and see if in fact the boots and belt are both in fact upgrades If it is in fact an exponential multiplier then the boots and belt are bad.

 

[Dreams]

By my calculations which granted may be imperfect you gain 2.21% damage from the belt of blasting

Indeed the value that you assigned to belt of blasting is imperfect. First of all it's assuming that haste has an overly inflated value far from its actual value and therefore it's also assuming that you would use haste gems.
The raw stat upgrade from equipping http://www.wowhead.com/item=30038/belt-of-blasting over http://www.wowhead.com/item=34557/belt-of-the-tempest if we take the correct stat weightings into account and therefore gem intellect is 19 SP, 4 crit, 1 haste. My mages buffed spellpower is 1015 meaning the 19 spellpower is a bit less than a 1,9% spellpower buff.
Now it might be more or less true, like @Fx has been trying to get into my head, that a +1% intellect buff is ROUGHLY a +1% damage increase. But as we all know there's a 5% difference between intellect and spellpower.
You are correct in assuming +2 average ilvl is ~2% throughput.
So unless you can convince me that 1 haste rating and 4 crit rating can increase my damage by about half a percent, I'm still not convinced of the belt of blasting.

 

[Resto]

No response from @Resto and @Tuskbreaker

I'm not an active user on this site, come on when I need something then get off. I'm also not like majority of this bracket who sit and bicker about dumb shit, so yeah. No response

 

[Dreams]

I'm not an active user on this site, come on when I need something then get off. I'm also not like majority of this bracket who sit and bicker about dumb shit, so yeah. No response

I think it says a lot about a person if they're not willing to back up their claims at all.
Also if you think the whole topic is just "dumb shit" why even post to begin with?

 

[Fx]

@Symandria I'm pretty sure we're just talking past one another.

Each ilvl gives 1.67% bonus damage, but not necessarily 1.67% output/throughput/whatever you want to call it. And even at the highest ilvl, it's still not relevant because you're still gaining more than what a gemslot would provide if you were to make that trade-off. Maybe if we were able to reach 170+ would it be a consideration.

It's also irrelevant for calculations, because we just run simulations and then add the total ilvl bonus to each result to see which items are better.

 

[Symandria]

The raw stat upgrade from equipping Belt of Blasting over Belt of the Tempest if we take the correct stat weightings into account and therefore gem intellect is 19 SP, 4 crit, 1 haste. My mages buffed spellpower is 1015 meaning the 19 spellpower is a bit less than a 1,9% spellpower buff.
Now it might be more or less true, like @Fx has been trying to get into my head, that a +1% intellect buff is ROUGHLY a +1% damage increase. But as we all know there's a 5% difference between intellect and spellpower.
You are correct in assuming +2 average ilvl is ~2% throughput.
So unless you can convince me that 1 haste rating and 4 crit rating can increase my damage by about half a percent, I'm still not convinced of the belt of blasting.

You say that the belt adds 1.9% + 5 secondary stats (probably about 0.25%) throughput damage = 2.15%~ throughput, vs ~2% throughput damage. That definitely sounds like an upgrade to me But no need to guess, that's what a fancy simulator is for, plug in all the numbers and if it checks out for the boots and belt update them

P.S. Why not go shimmer? It adds far more to DPS than Glacial Insulation, by freeing you to cast while moving without interrupting your spell or spending a GCD, and it comes with 2 charges for even more mobility!

@Fx So run the simulations and add the total ilevel bonus to the each result If the sims are good then a winner should emerge.

 

[Dreams]

@Symandria Yes about 1,9% more spellpower. A bit less if we take ring+trinket procs into account and add their average static SP gain to the equation. But you're right: we should just sim it.

P.S. Why not go shimmer? It adds far more to DPS than Glacial Insulation, by freeing you to cast while moving without interrupting your spell or spending a GCD, and it comes with 2 charges for even more mobility!

Don't be too quick to dismiss the added survivability from glacial insulation if you're 3-5 manning t6 content. There's a lot of random CC on tbc encounters that mages can just block, and getting a free shield every time you do so is going to make your healers job easier. (obviously irrelevant if you aren't running really small groups)
Also you should change talents depending on the fight. For example on archimonde and shahraz I play shimmer. On fights like Rage Winterchill and Kaz'rogal it's crucial not to pick shimmer. Etc.

 

[Fx]

So run the simulations and add the total ilevel bonus to the each result If the sims are good then a winner should emerge.

It's not so easy to build a character from scratch in simcraft. Takes some time.

Simcraft seems to think the two items are roughly equal. The DPS difference is within the error margin.

 

[Blackberryy]

@Fx @Onlydreams finally used simbot for my rogue and wow, instead of trying to keep up with so much math, i can finally just sit back and compare whatever i want. Glad you 2 tipped me off

 

[Fx]

finally used simbot for my rogue and wow, instead of trying to keep up with so much math, i can finally just sit back and compare whatever i want. Glad you 2 tipped me off

You need to double-check things though. There are some issues regarding some procs not working and some socket bonusses being doubled etc, so you have to manually fix some things before you can run accurate sims.

 

[Manaki]

Phew it's definitely heating up in here!!! Someone turn down the heating, quick!!

 

[Symandria]

@Onlydreams TBH I've tried a ton of fights solo where survivability isn't just important it's life or death. My pet doesn't tank, there is no healer behind me. And time and time again I've gone back to shimmer, on the hard fights I tried switching (especially when a stun was involved and I could blink out of it) to the extra 11% damage reduction and shield, what I found was as a squishy mage it just didn't matter you needed to not be hit by baddies or things go bad super fast.

Simcraft seems to think the two items are roughly equal. The DPS difference is within the error margin.

Forgetting margin for error which item was slightly better, you can always increase the test sample size if you need more data points

 

[Fx]

Forgetting margin for error which item was slightly better, you can always increase the test sample size if you need more data points

No, you're actually looking at a ~0.1% dps difference. Thinking that warrants spending tons of gold on a rare craftable when you don't play like a computer seems... foolish? Ignoring the error margin, which is also ~0.1%, do you really think you can play -that perfectly- and make use of such a dps increase? I really doubt any player in the entire world would be able to.

If it were any tangible increase (like 0.5%) I'd change it, but this is so minuscule and so irrelevant that it's not even worth considering even if it did give that ~0.1% increase (which I don't think it does, that's why there's an error margin).

Additionally, on shorter fights (<2.5 mins) the difference is smaller, while on long fights (over 3.5 minutes in length; there is only 1 fight in all of TBC that does this, which is KJ) it's slightly more favourable for Belt of Blasting. So if you can play like a computer for over 3.5 minutes straight, you can consider using it. In the meantime us mortals might want to stick with the easier to get, just as powerful option.