Percent casting = 92%
Average latency = 0.147 (Australian lag)
IS uptime = 67%
MF uptime = 97%
So time casting has room for slight improvement, IS could be better, MF uptime is very good. What I'm interested in is finding out how much of an effect increased latency has on our dps, since I figure with my gear I should be pushing more.
Ingame ping is consistently around 250. Wrathcalcs has me modelling 4.9k at that lag. As you can see that's well below what I actually parse, which makes me think I've set up WC badly. Dropping latency down to a ping of 70 nets an extra ~500dps.
For interest's sake I looked at the parse and using the WWSCalcs sheet plugged in some numbers to see what would happen if I extrapolated my dps for that patchwerk parse to 100% casting time. It had me sitting at a theoretical max of 5880dps. Adorielle would the 100% casting time be impossible, as that would imply 0 latency? Or is it % casting of the possible time we could be, excluding latency?
Otherwise, are there any other parses with average latency & fight length in the same ballpark that people could post? I'd be interested in seeing if we could develop a rough 'lag offset' number for how much it does impact our dps.
Wow, this finally caught the bug that's been niggling at me. Yes, WWSCalcs assumes a certain latency with its %casting number. I'd thought I'd set that at 200ms. It's actually set as 20ms >.< So really, everyone's %casting numbers have been significantly lower than intended, though it's also the case that as your %casting got lower, the less effect this had on you.
With that caught, it makes me wonder how good of an indicator %casting is. I could just set it to assume no lag, which I suppose is fine. Everyone's numbers will go down (though not by much I don't think). You get a similar indicator out of Average Latency, though, as that's just time-not-casting/number of casts. Obviously as %casting gets higher, AL gets lower. I suppose it's worth it to see both, though, as they both give a different view on what you could be doing.
As a note, though, ingame ping has little to do with the latency field on WrathCalcs. I should probably separate it out to ingame ping and reaction time, as it's intended to be a sense of both, but they have wildly different effects (namely, you can get around ingame ping by queuing. You can't get around reaction time nearly as easily).
Ah that makes sense. While the queuing system has been a large improvement in the WoW engine fundtionality, it occasionally screws me when trying to switch after an eclipse proc. I have just come to accept that there'll always be 2 wraths cast after eclipse procs simply because the 1st one has already been finished by the time anything client side registers that eclipse has happened. So a catch all field to include lag+reaction seems very sensible.
I keep one eye on my cast bar while wrath spamming since I can notice the 0.8s cast bar easily enough and know to watch to see whether eclipse has procced over the next ~0.5s or so. Out of interest- what do people on low latency find is acceptable? One wrath after proc? Two? A miraculous 0?
Two is standard. Even though you get the NG proc as soon as the wrath finishes casting, it seems as though Eclipse won't proc until the spell hits the target. At that point you're finishing up the first wrath, and by the time the eclipse buff shows up and you switch buttons, you usually have the 2nd one queued up to cast.
Relatively new to this forum, but I've spent the time reading all the very informative threads. Thanks to everyone for all the informative posts. I'd like to get some advice on my current dps.
I just started 25 naxx and 25 vault runs about a month ago. I've been through naxx 3 times and vault about 5-6 times. Rolls have been lucky for me and I've replaced 6 pieces (including my first T7 piece).
Here's a pic of the WWS numbers in case the link expires
I see that my IS and MF time are low - especially the MF. I've read a few posts that suggested not reapplyin DoTs during eclipse and that was probably part of the low uptime. Should I change this and always reapply DoTs?
Fixed the wwscalc sheet and I can see that my % cast time is abysmal. I recall being stunned several times, but it's not likely I was stunned for 38% of the fight. Any tips for improving cast times would be appreciated - perhaps through macros?
My crit is also on the low side. My crit rating is 352. I seem to be having a battle with crit vs. haste and currently haste is what I'm focusing on. DPS on this fight was 2209. I happened to crit higher on the first trash and actually got over 3200 DPS. Should I try be trying to raise crit through gems and food buff?
I know I have to continue to gear up and work towards my remaining T7 pieces. I would appreciate any advice on things I could be doing to maximize my DPS (therefore being more useful to may raid) in the meantime.
Last edited by wongtonsup : 03/03/09 at 3:49 AM.
Reason: fixed obvious error in fight time for wwwcalc
I PM'ed Adorielle asking about the exact same thing and the reply was that "Fall" is number of casts, used to track number of GCDs spent casting Starfall. As you can see, it says you've spent more time casting Starfall than Wrath, which obviously isn't right. In a 4:30ish fight you should probably have a "2" in the "Hit" column and nothing in "Crit".
Make sure you've downloaded the latest version from the first page of this thread.
The number in the Crit column of the spreadsheet for SF/Wr/MF, should be the number of crits, not the crit percentages. In WWS hit the blue '+' tab, next to 'columns' to see the actual number of crits.
Thanks for that catch. I've updated the original post with the new wwwcalc numbers and it's pretty obvious my % casting time is abysmal. I was stunned several times, but not for 38% of the fight. Any thoughts on raising % casting? Macros? I've shied away from this due to the unpredicable nature of eclipse and because I didn't want to wind up clipping my dots with a dot macro. I'd be interested to hear what works for people.
I put a sample fight for me in to WWS calcs just to see what numbers I got and here are some things I noticed.
1. Your WWS says the fight was 4'30 which is 270 seconds
2. In a 141 second fight (on patchwerk) I had 50 SF casts, and 21 wrath casts (Note: I use the rotation that uses wrath to proc eclipse, you appear to be using starfire to proc it) But even with that you had a total of 82 casts of the two spells combined while I had 71 and about 3 minutes shorter fight.
3. The spreadsheet gives me an average latency of 0.1536 while yours is 0.9849 - not sure what you can do about this, maybe adriole can inform you if there is anything in your power here.
4. I don't think macros are a good idea for our variable rotations, the only exception is tying in a trinket to a spell cast to ensure maximum uptime.
5. I will SS the WWS calc and update this post later with it (can not do it from work) and also the WWS it relates too.
Thanks Eilt. Could framerate be an issue here? My framerate is usually around 10-20fps during raids. I also thought 20ms latency was too optimistic for me - I typically see 100ms or so, but even then, % cast is still extremely low.
Typically when I raid, I see some low FPS. This shouldn't affect much, since it's client side, except how quickly you see an eclipse proc. I usually spam the button for whatever I am casting due to the queueing system, since my casting bar isn't always accurate in terms of casting time. However, I also run with nearly 260ms latency almost always, so I'm not sure, theoretically, what the effect on my DPS that will be. I managed to post 3.1K on Patchwerk running about 1850 BUFFED sp (but low crit and haste). WWSCalcs says my casting time was nearly 88%, but I never stopped casting for anything and didn't run out of mana, so never even had a pot. I'm not sure, with my latency, that 90%+ casting time is going to be feasible, but I'll keep trying.
I filtered Uday's WWS log (the entire log, not just the PW fight) to show just SF and Clearcasting (= Omen of Clarity). I assumed that any clearcast within 0.5s of a Starfire was caused by the Starfire. This is optimistic. Some of those could have been caused by an instant cast.
The WWS reports 707 SF hits. My eyeball counting shows 49 OoC caused by SF. That is a proc rate of 6.93%.
If the actual proc rate is the 6% reported on wowwiki and wowhead, there is about a 13% chance of getting 49 or more procs in 707 trials. Not bad, particularly since some of those procs probably belonged to MF or IS.
I don't know what Uday's actual cast times were. If we use a very optimistic 1.5s, and assume 3.5 procs per minute based on actual cast times (not base times), there is about a 4.6% chance of getting 49, or fewer procs.
At 3.5 ppm based on
2s, chance of 49 or fewer < 1/50000.
3s, chance of 49 or fewer < 1/6e15
3.5s, chance of 49 or fewer < 1/6e22
I believe WrathCalcs and Rawr are using the 3.5s value. From looking at the code, I believe Simulationcraft is using the 3s value. If I'm correct, those are giving OoC too much credit. At low crit rates, in a mana-constrained fight, Rawr recommends SF-spam (with OoC), versus IS SF (without OoC).
I'm guessing that the real number is a fixed 6%, which is the number that feral cats report (and OoC has a much larger impact on them, so they have a large incentive to get it right).
Using the same method to look at MF. I see 7 OoC procs from 84 casts (8.3%).
Gear used in those tests should be exactly the same as what I'm wearing on live currently. I was surprised that the numbers were so high considering it's a much longer fight than live PW. One would think that a 2.30 kill with the high bloodlust uptime would reflect much higher DPS for me, but I don't think I've seen myself break 6k yet on live. Odd.
So, I'm seeing some of the other numbers being posted here and I know I must be doing something wrong at this point. I'm pulling respectable DPS compared to the rest of my guild but I'm not able to post numbers like the rest of the community here.