Spell Casting Lag/GCD Bug

[cerbul]

Here is the issue that I am experiencing regarding spell casting: Even if I test out to make sure I don't press a spell too early before the GCD, sometimes my spells simply fail, even tho they do trigger a GCD. I've been realy paying attention to make sure I press buttons at the correct time. The issue is not there if I press the buttons AFTER the gcd finishes, but obviously that is a loss of playing performance, so is simply unnacceptable.
I've been literally able at some point to cast Devouring Plague about 6 times one after another, it triggered the GCD 6 times, but not a single time the spell was applied to my target.

I am confident that there are some problems there from Blizzard part. Some ppl sugest that it is a timingMethod problem, however timingmethod 1 and 2 for me makes all the spells take longer to cast, so that is not a solution for me.
However, if it is related to the cores of my Intel i7 920, how can I discover which cores are synced properly with which ones, so that I use only synced cores. Would this help in any way? Is the issue related to other problem? Is it simply a bug related to the moment when I press the spell? (i.e. I press the spell in a time winwdow where triggers the gcd but doesn't trigger the spell queue)? Any advice I can get on this? Anyone else experiencing this issue? Also, is not first time when I am pressing a macro that contains a /use trinket and then a spell, and the result is that the trinket doesn't trigger, but the spell does fire. Can't blizzard actualy trigger the GCD ONLY if the spell is taken into consideration, more exactly being added to queue?

Things I have tryed in order to get the problem solved:
- Reinstalled WOW (I know this is 99.9% useless, but I did it since I had something else to work on in the meantime anyway).
- Reinstalled all addons(I didn't copy them from the WOW I deleted)
- Tested this without any addons installed aswell, same problem.
- Tested with processaffinitymask values: 84,85,255,240.
- Tested with timingMethod default, "1" and "2".
- Tested with most services disabled.
- Tested with Windows media player opened.
- Tested without antivirus/firewall related software.
- Tested with ping to server window opened, to make sure is not an internet issue.
- Tested with min, max, equal to my world ms, a bit lower, a bit higher, custom lag tolerance.
- Tested with and without leatrix latency fix.
- Tested with and without vsync.
- Tested with and without Reduce input lag.
- Tested with and without Network Optimization.
- Tested with network checksum offload enabled/disabled.
- Tested all the steps adviced on the us battle net forums, both performance steps and the ones adviced to everything related to my issue.

Anyone has a fix for this? The moment the spells fail, I have my fps/ms stable, without any variations. Not even a tiny bit.

[Deleted]

Get hit capped. /thread

[Deleted]

Get hit capped. /thread

pretty much this.
you need 17% to never miss a spell.

also trinkets share cooldowns with other trinkets sometimes.

[Aleksandru]

A few tips:

Patch 3.3.2 (2010-01-02): Manually adjusting this cvar is no longer needed unless you're trying to run the game on specific cores. It is now uncapped so it will run on whatever core your OS deems as fit.

If you play a caster class (mage, priest, etc) you should enable "Custom Lag Tolerance" option from Interface > Combat, move the slider so that will be ~ your WORLD ms.

Install Leatrix Latency Fix (http://www.wowinterface.com/download...atencyFix.html)

[Zka]

Get hit capped. /thread

Seriously... Did you guys read what he wrote?

[cerbul]

I am able to read myself when I miss a spell, is not that. I have 17.01 % hit, so there is no issue there. To make it more clear, if I use Mind Blast, it will not go on cooldown, and the casting won't happen, though, the GCD does happen. So ye, nothing to do with hit cap.

About processaffinitymask: I started playing with it after I ran into spell casting issues, not before.

I tryed with lag tolerance higher/lower then my world ms, with same result. Only difference is that when I play with lower value, I have to press the spell much later to register, but that WON'T trigger a GCD. Nothing will happen in this case. I am talking about the case when i cast a spell, GCD triggers, but the spell is not happening.
I am playing by default with leatrix latency fix, I tryed without aswell.

[yurano]

Put Custom Lag Tolerance to 0 ms.

[cerbul]

Put Custom Lag Tolerance to 0 ms.

Ye.. that solution sounds like the perfect antivirus: "If you don't want to get viruses/trojans from the internet, simply don't connect to internet. Is 100% safe."

[Darthlegolas]

I've actually noticed this as well. I macro in my trinkets and Elemental Mastery to both Lava Burst and Lightning Bolt on some fights to make sure they are literally always on cooldown. Sometimes when Elemental Mastery is off cooldown and I hit the macros, everything stops. No cast going off, no EM triggering, nothing. Just GCD's and no casts.

[cerbul]

Yeah I wish I was frapsing while this happened to me, but sadly I wasn't . It's luckily only happened once in arena where it really matters and it was against a DK of all things (DKs = to priest what an 18 wheeler = to an ant). I cast and my spells glitch. I can move, type, pretty much interact and do anything in the game and even press my abilities which start a GCD but nothing is ever casted. I think I've noticed it twice with DKs which made me think it was a bug with them, because it happened after Death Grip but it happened a third time when a DK was not present.

The time that it happened in arena I noticed something very interesting. I started moving and jumping/running from the spot where the glitch occurred and managed to make it a good 30-40 yards away, but when I died and my score screen came up, my body (or at least my camera) quickly panned over to the spot where the glitch occurred and took me back. I don't know what it is, but it's pretty stupid. It's happened a total of three times already and I'm just praying it goes away soon or doesn't happen often.

This kind of behaviour is almost certain related to packets loss. For me, I did all the testing with a ping cmd opened, so to make sure is not related to internet problems.

[cerbul]

My problem might be related to processaffinitymask, I've been getting close to perfect results with "84". Anyone can give some numbers that could solve this issue? Anyone having same problem and got fixed with a specific number for processaffinitymask? I can't get to make it work perfect no matter what value I use.

[grouch]

i am having a similar issue. i have this macro:

/cast Inner Focus
/use Shard of Woe
/use Jaws of Defeat
/cast [target=mouseover,exists][target=target,exists] Greater Heal

sometimes the first three will go off but usually they will not. this is reproducible but i can't always predict when it will happen. like the above poster said, if i execute the macro after my current spell finishes it works fine every time. if i execute the macro before the current spell is done then there's a very low likelihood that the first three commands will execute.

has anyone started a thread on the battle.net forums regarding this or run into a real solution?

---------- Post added 2011-09-07 at 06:08 PM ----------

it seems that if i put a pause between each command then it does in fact execute properly:

#showtooltip Greater Heal
/cast Inner Focus
/script debugprofilestart();while debugprofilestop()<25 do end;
/use Shard of Woe
/script debugprofilestart();while debugprofilestop()<25 do end;
/use Jaws of Defeat
/cast Greater Heal

this is obviously a hack but so far it seems to be working fine. i'd really rather not have to do this though.

[cerbul]

Actualy that is happening to me too. I macroed 2 cooldowns and a spell into a macro and I get moments when the final spell fails.
I also noticed that it only happens to have some spells cancelled if I press that spell a bit more early. If I cast it just as the current spell finishes, it will cast it corectly.
Biggest concern is that this happens aswell with simple spells, not only with macros.
I've been able to reproduce this very well at the start of the combat on my shadow priest.
To give you an example, this is how is behaving after latest small patch:
I start casting Vampiric Touch. Just as the spell is about to finish, I press Devouring Plague. Sometimes, I get a spike as I enter the combat, and the spell, doesn't appear highlighted (as I press the button). If I force it and press fast SW: Pain, I won't get my Devouring Plague casted, though Shadow Word: Pain gets applied. The problem is not only at start of the fight, can also happen during the fight, but what is needed is to cast a spell with cast time, won't happen if I cast instants. Anyone got any clue about what Blizzard did wrong and what they did to "partially" fix this issue?

---------- Post added 2011-09-15 at 06:49 AM ----------

Unfortunately adding "/script debugprofilestart();while debugprofilestop()<25 do end;" did not help with the issue, maybe a tiny bit. I still get it happening if I press the macro before the spell cast finishes. The problem with that script is that now I get a tiny freeze every time that is executed in the macro, making it impossible to move/etc for a tiny part of a second. So ye, not worth imo. Anyone got a better solution?

---------- Post added 2011-09-15 at 07:20 AM ----------

There is a forum post on us official of some person that has this issue, and what he did was to macro a cooldown and a spell into a macro. He started to cast shadowbolt, and jsut it was about to finish casting he pressed the key for the macro, and curiously, the cooldown got "another action is in progress error" while the other spell was executed properly. The name of the thread is "Spell Queue Lag".
For me, seems that any cooldown that modifies the haste gets not casted. Not sure why Avenging wrath was never skiped, thouh I tryed to put them in a total different order in the macro, to see if there is a pattern to it. I guess that Blizz has some bugs there related to haste problems. Forgot to add, he didn't get an answer specifically on the issue, but I think the silence shows that they got enough info now to test it out properly.