Jump to content


Photo

Shadowpriest Theorycraft 3.3 Edition - I get by with a little help from my friends


  • This topic is locked This topic is locked
253 replies to this topic

#241 Klokworkk

Klokworkk

    Glass Joe

  • Members
  • 1 posts

Posted 28 June 2010 - 09:13 PM

Muqq and izzolite have argued that there is no "haste cap". There is however, in my opinion, a point of intersection. This intersection is the point where Spell casting speed out paces player skill. This intersection is not simply a static point that is applied to all players, but instead a sliding scale whereby players of varying skill will reach this intersection at varying points. Remember, as you increase your haste rating three things happen. Your Damage over Time spells and Mind blast have a shorter duration/cast time, your channeled ability (Mindflay) becomes shorter, and your Global CoolDown (or GCD) lowers. As each of these three conditions change, you, as the player, must also change. You need to react faster, being more aware of your timers, and be ready to recast abilities sooner, and sooner. For some players, this may not happen until they reach the GCD thresh hold (the point at which spell casters can no longer reduce their GCD...this), which is 1.0 seconds. Some players will reach this point well before that. This is something that only the individual player can determine, and is something that should be constantly tested.

There is also the matter of dps downtime as a result of excess haste. As I said before, and has been stated for some time, the GCD cap for a spell caster is 1.0 seconds. This is a problem that is more or less exclusive to Shadow priests and Balance Druids.

Vampiric Touch, and Mind Blast have a base ( or non-haste modified) cast time of 1.5 seconds. Your GCD is baseline (non-haste modified) at 1.5 seconds. As you increase your haste through items and gems, your VT cast time, and duration begine to decrease, this applies also to Mind Blast. Your GCD also begins decrease. The problem at high haste/gear levels lies in the ability for spells, and not the GCD to continue to cast faster even when the GCD can no longer be hastened. At this point, you will find yourself with dead time. This dead time, over the course of a fight, can be rather detrimental to your performance. There is nothing that can be done about this. Once you reach 1.0 second GCD, you are NO LONGER ABLE to increase your cast frequency. As many Shadow priests who use Engineering are beginning to learn, once you reach around 1200 haste (this is not a far fetched number.. in less than 50% heroic gear I all ready have 1204 haste), activating hyper-speed accelerators brings your VT/MB to around .90-.83 seconds...dependent on the actual level of haste from gear/gems you all ready have. Consider that Hyper speed accelerators provide 340 haste (or approx 10.5%). Bloodlust provides 30% as a comparison. The question I have been trying to flesh out is whether the benefits to DoT's outweighs the loss of cast activity caused by going under GCD cap every 60 seconds + the 45 seconds of Bloodlust.

Edit: As "the not so evil" pointed out in the post below, Lag will play staggering role in how haste affects your dps gains from haste. This is something that can not be prevented, and will always have a detrimental affect. While it is possible to mitigate minor latency changes, such with cast bar mods like Quartz+MF2 which can provide a buffer zone (the red section in the cast bar that changes) for spell queuing and MF ticks. At the end of the day, if you are player who consistently has lag spikes, or generally high latency, your "intersection" is most likely to be markedly lower than a player of equivalent gear/skill with a lower latency. When testing your haste value in this fashion, always consider things in your game environment that will directly impact your ability to smoothly and quickly cast spells. The ability to maintain a high "activity" percentage is a cornerstone of good dpsing.


While your statements here are fairly true, I think you're looking at this in a bit different then Muqq/Izo/etc. Most of the discussion that occurs on this topic is mostly to do with optimal playing. In every theorycrafting scenario haste will improve your DPS even over the 1640 mark. The main problem/discussion point past that concerns MF Vs. MB and generally depends on the player at this point.

Regarding cooldowns, there is a point where just outright popping everything is going to hurt you more then help you, you are right about that. The common theme with said cooldowns is to stagger them out. You and I are trolls for example, on a fight like Festergut we could stagger our Haste CDs to Bloodlust -> Berserking -> EngiGloves -> Potion. The only wildcard there would be Black Magic (which even if it procs during other cooldowns, it would still technically be a DPS increase).

I do agree with the basic point that the player makes the gear, and the player is going to have to adjust himself accordingly. However, given numbers alone Haste (as it is right now) is always going to be a DPS increase stat versus other possible choices.

#242 Shelak

Shelak

    Glass Joe

  • Members
  • 20 posts

Posted 30 June 2010 - 01:53 AM

Bowchikabow, in a perfect scenario a haste cap is reached when the main nuke reachs 1sec cast time. Since you only have that during bloodlust, you can continue to gather haste until you reach that 1sec MF outside of bloodlust, which takes more haste than current possible on gear.

Something different, I was TogC25 last week on icehowl (the big yeti thing if the name isnt correct), with all dots up, mind blasting here and there because I was missing refreshment but I didnt get the buff. Same thing for the retribution paladin we had with us. I havent heard about a bug with that yet. Anyone of you know more?

#243 Xtian

Xtian

    Von Kaiser

  • Members
  • 30 posts

Posted 01 July 2010 - 08:37 PM

Something different, I was TogC25 last week on icehowl (the big yeti thing if the name isnt correct), with all dots up, mind blasting here and there because I was missing refreshment but I didnt get the buff. Same thing for the retribution paladin we had with us. I havent heard about a bug with that yet. Anyone of you know more?

Replenishment buffs the 10 raiders with the lowest mana pool percentage. If there were 10 other people with less mana than you (by percent of their whole mana bar), then you wouldn't get the buff.

#244 Althor

Althor

    Great Tiger

  • Members
  • 792 posts

Posted 02 July 2010 - 07:46 AM

Replenishment buffs the 10 raiders with the lowest mana pool percentage. If there were 10 other people with less mana than you (by percent of their whole mana bar), then you wouldn't get the buff.


This is almost correct but there is a caveat. Namely, that it won't try to select someone who already has a Replenishment buff from someone other than yourself.

i.e. If Player A has the lowest percentage of mana but they already have a Replenishment buff from say Paladin B then Priest C's Replenishment won't choose Player A to give Replenishment to. It will instead check the next lowest etc.

#245 Shelak

Shelak

    Glass Joe

  • Members
  • 20 posts

Posted 02 July 2010 - 12:19 PM

Exactly. And to make this even more strange, I had replenishment at the other 2 bosses before, but none at icehowl. Something I didnt observe in any other 25man raid before. I try to get there this ID again and grab webstats to confirm whats going on.

#246 enqi

enqi

    Glass Joe

  • Members
  • 21 posts

Posted 22 July 2010 - 03:17 PM

Bowchikabow, in a perfect scenario a haste cap is reached when the main nuke reachs 1sec cast time. Since you only have that during bloodlust, you can continue to gather haste until you reach that 1sec MF outside of bloodlust, which takes more haste than current possible on gear.

Depends almost completely on the brain and server lag of the player.
Haste pulls ahead the lower latency and the better you are at clipping and keeping dot uptime.

But yes, if you're considered a stable player who knows your game and have a decent connection getting haste will be a dps increase all the way to 1sec mind flay w/o bloodlust compared to crit since VT/DP scales very very well with haste aswell.

I believe crit doesn't pull ahead until 1600~ unbuffed haste rating with a 50/50 (server/brain) latency setting in rawr with BiS setup. It can be argued with but I for example is going for dfo/cts and the normal sp/haste in yellows etc to reach the maximum amount of haste.

#247 Bowchikabow

Bowchikabow

    Von Kaiser

  • Members
  • 65 posts

Posted 22 July 2010 - 04:10 PM

I don't want to sound argumentative, however. Despite the paper math that indicates increasing haste to be the best increase in dps, an armory check of the top 50 US priests recorded on WoL are utilizing +/- 1200 haste, with many barely being in 1100 range. These are the people who are topping world charts against others of their class spec....consistently. These are also players who are sitting on their Best in Slot gear, or near enough that what pieces they have yet to obtain will not drastically alter their current dps. This tells me that if haste being the answer is based on the math, maybe the math being done is being wrongly interpreted.

As is said often: "The numbers don't lie". In this case, the numbers appear to be saying "someone missed something". I will be honest, I don't know specifically what that might be.

In my previous post I agreed with Izo in that there is no attainable haste cap, from a strictly numbers standpoint. My contention is that individually, the player must constantly evaluate the benefit they are getting from haste, and to understand that they will likely not see their best numbers by simply getting closer and closer to the GCD cap.

I am very curious to know how it is that the numbers we are getting from simulations do not, at the higher levels, match actual numbers?

One such example:World of Logs - Real Time Raid Analysis

shadow priest leaderboard for Festergut 25man Heroic.
"Better to keep your mouth closed and be thought a fool than to open it and remove all doubt"

#248 tedv

tedv

    Observation: I am awesome

  • Allied Members
  • 1,796 posts

Posted 22 July 2010 - 07:23 PM

I am very curious to know how it is that the numbers we are getting from simulations do not, at the higher levels, match actual numbers?


I believe the answer is that people have much worse latency and reaction time than they suspect. Allow me to tell a short story...

My guild has two really solid shadow priests, myself and Risocko. Risocko has slightly better gear than me, but I noticed that on the tank-and-spank fights, he did better damage than could be explained by the gear difference alone. Skada said he was doing slightly more damage than me on Vampiric Touch and Devouring Plague, and a good deal more with Mind Flays.

Obviously I'm unwilling to perform sub-optimally, so I looked at the World of Logs parse. He wasn't getting Power Infusions or Tricks of the Trade and we were both using potions as appropriate (including using pre-pull potions). Eventually I opened up the combat log browser to look at the exact time stamps of our DoT refreshes and Mind Flay ticks.

Basically what I found was that whenever I cast a Mind Flay, I was consistently 150 ms too late in casting the next spell. I have 100ms of latency to my server, but when I watched my habitual cast sequence in the next raid, I could tell the problem was me. I had trained myself to start the key press (physically touch the key) at the moment I wanted the ability to get sent to the server, but it took a fraction of a second to actually finish pressing the key.

This effectively turned a 1.8s mind flay into a 1.95s mind flay, which costs 8% of its damage. It's even worse during Bloodlust, when it turns a 1.2s mind flay into 1.35s, reducing mind flay damage by 11%. Vampiric Touch and Devouring Plague also ticked less because they were getting refreshed later. This easily accounted for a 5% drop in damage, all from being a few milliseconds too late.

In the end, I've started pressing my buttons around 250ms before the Mind Flay cast bar gets in the red latency bar displayed by Gnosis (or Quartz for those of you who haven't switched yet). This is often 400ms before the official end time of Mind Flay. I'm still learning just how aggressive I can be and not lose ticks, but my damage has improved such that I'm generally competitive with Risocko, and the player who deals the most damage is determined by random elements of the fight. (For example, whether the slime pools near you get triggered on Rotface.)

#249 moowalk

moowalk

    Don Flamenco

  • Members
  • 283 posts

Posted 22 July 2010 - 09:43 PM

I find if you're using MSBT, it's best to turn off spam throttle, or whatever it's called. This means I get real time feedback on whether I'm clipping dots or not, which is the best way to improve timing.

To reduce spam, set threshold such that SW:P ticks aren't shown.

#250 enqi

enqi

    Glass Joe

  • Members
  • 21 posts

Posted 23 July 2010 - 11:54 AM

I don't want to sound argumentative, however. Despite the paper math that indicates increasing haste to be the best increase in dps, an armory check of the top 50 US priests recorded on WoL are utilizing +/- 1200 haste, with many barely being in 1100 range. These are the people who are topping world charts against others of their class spec....consistently. These are also players who are sitting on their Best in Slot gear, or near enough that what pieces they have yet to obtain will not drastically alter their current dps. This tells me that if haste being the answer is based on the math, maybe the math being done is being wrongly interpreted.

As is said often: "The numbers don't lie". In this case, the numbers appear to be saying "someone missed something". I will be honest, I don't know specifically what that might be.

In my previous post I agreed with Izo in that there is no attainable haste cap, from a strictly numbers standpoint. My contention is that individually, the player must constantly evaluate the benefit they are getting from haste, and to understand that they will likely not see their best numbers by simply getting closer and closer to the GCD cap.

I am very curious to know how it is that the numbers we are getting from simulations do not, at the higher levels, match actual numbers?

One such example:World of Logs - Real Time Raid Analysis

shadow priest leaderboard for Festergut 25man Heroic.


You'll have to be enlightened of the fact that the good shadowpriests do not get buffs such as PI nor Tricks, if you view the shadowpriests record you also notice that they do not have WoL #1 ranked warrior pulling out 20k on Festergut nor do their mages pull ahead which in my opinion, means something is very wrong.

If you go to our guild logs on WoL, you'll see that Ryts (My fellow shadowpriest, I'm pretty low on gear atm, still rolling with RotU) is pulling out numbers far bigger compared to gear and buff diffrence than you could calculate from the people you'll most likely find in the top50. We do not get PI, we do not get Tricks, we do not get FM but we still perform accordingly, you can also account to the fact that like I mentioned in my previous post, the value of haste is very brain/server latency related due to the clipping and dot uptime.

Very few shadowpriests still clip, why? It's not as much of a dps gain as it was in Ulduar, it doesn't net you more than a few hundred anymore.


There's many factors involved, most "top" shadowpriests I've seen thus far use a nochannel macro and just spam away because their server latency is 10 but their brain is immensely slow regarding multitasking keeping track of the surrounding and managing to clip and keep dot uptime perfectly.


I will give you numbers at the end of WotLK when I got my BiS list acquired.

#251 XXxx_Totmacher_xxXX

XXxx_Totmacher_xxXX

    Glass Joe

  • Members
  • 2 posts

Posted 26 July 2010 - 03:39 PM

Possibly a good time to bring up Snowfall key press. The Add-on changes the initiation of the action to the press of the key vice the release of the key. It's one of those out of sight out of mind add-ons you completly forget you have.

SnowfallKeyPress

#252 Guest_aboshan_*

Guest_aboshan_*
  • Guests

Posted 28 July 2010 - 07:03 PM

Just a mere suggestion about the utility of mind blast.

It's still a useful spell to avoid mind flay clipping before refreshing dots. So when VT has 2 second left, I cast mind blast (1 sec cast with appropriate haste rating) and 1 sec to cast VT. There in no way you can pull a 1,5 sec MF with regular gear and since most players has a 1,8 sec mind flay, 2 ticks require 1,2 sec channeling. Since MB does more dmg than a single MF tick, I use quite often.

#253 Phlo

Phlo

    Glass Joe

  • Members
  • 7 posts

Posted 13 August 2010 - 12:56 PM

Shouldn't the Totem of Wrath of Air add 5% spellhaste, and therefore be added itself to the "raid buffs/debuffs" chart?

#254 Snowy

Snowy

    Mitt Romney?

  • Moderators
  • 10,236 posts

Posted 12 October 2010 - 01:07 PM

Not that this thread has seen use for a while, but discussion about 4.0 mechanics should start in a new thread. This one will be locked, to avoid confusion of people looking for help and seeing old mechanics instead.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users