DPS Spreadsheet - Page 41

Ricter · 04/28/08, 9:27 AM

After using the spreadsheet, it puts my unbuffed dps around 700, and my buffed dps ~1300. However, in practice, my dps is only ~950-1k.

I understand that part of it is because it's simply an "average", but I rarely break 1k, let alone get close to 1300. I've double checked my buffs, gear, enchants, and talents multiple times, and I'm sure they're all correct. I also know that part of it is that I'm not perfect at my rotations.

Now, as far as I know, recount's calculation of my dps compensates for my downtime (i.e. time I'm not attacking the boss).

Also, what "usage" of cooldowns generates the most dps? Spamming once they're up, or combining them if they're available near to each other?

Finally, I currently have version 2.4.0.3 I realize this is an older version, but when I downloaded the newest version and tried to import, my dps became bugged and wouldn't display, even though all of my items, talents, and buffs loaded properly.

Left · 04/28/08, 9:37 AM

Understand also that the model computes DPS on a perfectly immobile fight, where you have 100% uptime on all your buffs, and your cycle never changes. It also assumes that you never bandage, never vanish, never pause to do anything else, and simply never do anything other than DPS the boss.

In other words, it's a best possible case scenario. If any of the above is not true (ie you have an interrupted cycle, interrupted DPS, or interrupted buffs) you will not reach the potential in the sheet.

Recount, I believe, does take into account time that you are not DPSing. However, it isn't perfect at it. I believe there is a window before and after you start DPS that it counts (as part of figuring out when you started). It has no good way of knowing really that your 10 seconds of stopping DPS was getting air bursted (for example) as opposed to bandaging. So it can't completely and perfectly account for downtime.

Ricter · 04/28/08, 2:02 PM

Originally Posted by Left

Understand also that the model computes DPS on a perfectly immobile fight, where you have 100% uptime on all your buffs, and your cycle never changes. It also assumes that you never bandage, never vanish, never pause to do anything else, and simply never do anything other than DPS the boss.

In other words, it's a best possible case scenario. If any of the above is not true (ie you have an interrupted cycle, interrupted DPS, or interrupted buffs) you will not reach the potential in the sheet.

Recount, I believe, does take into account time that you are not DPSing. However, it isn't perfect at it. I believe there is a window before and after you start DPS that it counts (as part of figuring out when you started). It has no good way of knowing really that your 10 seconds of stopping DPS was getting air bursted (for example) as opposed to bandaging. So it can't completely and perfectly account for downtime.

Right, right. I understand that it's ideal conditions. My point wasn't clear. Do other rogues see a similar loss in dps (either % or absolute)? I want to make sure that it's because of non-ideal conditions, not because of any lack in my ability. It seems like a huge loss.

I knew part of it was recount as well, and good point on the air burst example.

Left · 04/28/08, 2:48 PM

Originally Posted by Ricter

Right, right. I understand that it's ideal conditions. My point wasn't clear. Do other rogues see a similar loss in dps (either % or absolute)? I want to make sure that it's because of non-ideal conditions, not because of any lack in my ability. It seems like a huge loss.

I knew part of it was recount as well, and good point on the air burst example.

Ah, sorry. Speaking for myself, I generally get 200-400 below my "ideal" fight numbers, depending on the fight and various peripheral conditions. In highly mobile or chaotic fights (eg, Archimonde, Kael) my numbers are much worse than that.

Dontmindme · 04/28/08, 9:25 PM

I've posted an alpha version with Left's cycles at rogue.raidcal.com
It should be noted that Envenom is not yet implemented, nor Ghostly Strike.

Instead of cut cycles, all cycles are considered cut cycles and the new term "Slack" is being used.
Positive slack is the average amount of time per cycle one cuts off.
Negative slack indicates that SnD will drop, on average, for that period of time.

Basically looking for unusual behavior at this point. It very well may contain bugs and some effects may not be implemented.

Left · 04/29/08, 12:08 AM

FYI, DMM, check out this post and following in the gear sheet thread. Should give a more precise model for [Shattered Sun Pendant of Might]:

Rogue Gear Spreadsheet

Also, I'll try to work on the envenom cycles when I have a chance... which won't be this week, unfortunately. Too much going on at work.

Ricter · 04/29/08, 10:13 AM

Originally Posted by Left

Ah, sorry. Speaking for myself, I generally get 200-400 below my "ideal" fight numbers, depending on the fight and various peripheral conditions. In highly mobile or chaotic fights (eg, Archimonde, Kael) my numbers are much worse than that.

That's what I needed to know. So running ~1k of 1.3k isn't that bad. Thanks for your time and sorry for being unclear originally.

Left · 04/29/08, 10:47 AM

Preliminary bug report in the Alpha release: while using a Mutilate cycle and increasing points in Imp. Eviscerate such that the cycles model selects a 3 finisher cycle, the overal DPS estimate actually drops when the new "higher DPS" cycle is switched to.

Steps to Reproduce

Load the Alpha with default gear
Switch MH weapon to "Blade of Serration"
Switch OH weapon to "Fang of Kalecgos"
Load the "Mutilate" talent build
Examine the cycle recommended (3-5s/5-5r) and DPS estimate (1679)
Change "Improved Eviscerate" to 3 talent points, leaving all else the same
Examine the new cycle recommended (4-5s/3-5r/4-5e) and the new DPS estimate (1610). Note the decrease.
(I also verified in the cycles model on the "Buffed Cycles" sheet that this cycle is listed as having the highest DPS.)

Expected Behavior
The cycle that the cycles model shows as producing the most DPS should not produce less DPS than when a different, theoretically lower DPS, cycle is in use.

Source of Problem
I believe the source of the problem is in "Buffed DPS" cell I10:

=IF(AND(instatk="mut",'Buffed Cycles'!P283=0),0,E10*evisec/M3*bGlobDam*(1+DDmult)*(IF(AND(NOT(IF(Talents!$O$25="Instants",1,0)),Q17>0),1+Talents!I37*20%/(30*Q17),1)))

I'm not sure what all the rest of the equation is doing, but it appears that there is a reference (shown in red) to treat the calculation differently for Mutilate. This is wrapped in an IF statement (shown in blue). (It's also referencing a cell that appears to have been in the old Mutilate block but is now just an arbitrary part of the cycle model.)

Proposed Fix
Like I said, I don't understand the whole formula, but if you take out the IF conditional surrounding Mutilate it seems to work properly again. Here's the formula as it should be (without the conditional):

=E10*evisec/M3*bGlobDam*(1+DDmult)*(IF(AND(NOT(IF(Talents!$O$25="Instants",1,0)),Q17>0),1+Talents!I37*20%/(30*Q17),1))

Following this change, the displayed DPS correctly increases by 5-10 or so when moving to 3/3 Imp. Evis. and getting a new cycle recommendation.

Ashere · 04/29/08, 3:18 PM

Someone on the European forums found something odd running a 0/40/21 spec, using T6 equipment: it started to outperform full-Combat again.
Doing some napkin math, I got to the conclusion that a 1.8 dagger while hitcapped would yield about as much energy from Combat Potency as Relentless Strikes yields in a 5s/5r rotation, and a bit more with a faster offhand and some haste, I still find it pretty hard to imagine it could become that good. So I'm wondering if anything has changed in regards to the modeling of talents involved in such a build.

On a sidenote: if such a build actually WOULD be viable, then it would be so in 25-men raids, definitely not in 10-men content, since hemo would be overwriting itself too frequently, not using all it's charges before being re-applied.

Jakani · 04/29/08, 5:06 PM

Originally Posted by Ashere

Someone on the European forums found something odd running a 0/40/21 spec, using T6 equipment: it started to outperform full-Combat again.
Doing some napkin math, I got to the conclusion that a 1.8 dagger while hitcapped would yield about as much energy from Combat Potency as Relentless Strikes yields in a 5s/5r rotation, and a bit more with a faster offhand and some haste, I still find it pretty hard to imagine it could become that good. So I'm wondering if anything has changed in regards to the modeling of talents involved in such a build.

On a sidenote: if such a build actually WOULD be viable, then it would be so in 25-men raids, definitely not in 10-men content, since hemo would be overwriting itself too frequently, not using all it's charges before being re-applied.

If the hemo debuff is included in the analysis, it's already accepted that a hemo build offers more Raid DPS than a full combat build (up to one rogue). The real question is: does 0/40/21 beat 11/27-8/22-3? I'm pretty sure tri-spec still comes out on top there. Do you have a link to the post with its findings?

Dorvan · 04/29/08, 6:14 PM

Couple of apparent bugs with the new alpha. Gear/talents were plugged in from my armory:

The World of Warcraft Armory

with default buffs.

With my gear, a 3+s/4+r/4+e cycle was suggested. Shifting talent points into Imp Evis didn't have any impact on my DPS though, even though the suggested cycle didn't change when I did so. More precisely, when I removed 3 points from Vile Poisons, my DPS dropped by about 8 (I believe from 1355 to 1347). Then when I added those points to imp Evis, my DPS remained at 1347, even though it was suggesting a 3+s/4+r/4+e cycle beofer and after the talent change.

Also, the spreadsheet seemed to be overvaluing bloodlust brooch/berserker's call. They were coming out around 40 DPS above options like the Warp-Spring Coil or the DST....not sure what's going on there, but even if my gear's not top-notch that can't be right.

FWIW, I was using the spreadsheet in OO 2.1

Left · 04/29/08, 8:12 PM

Dorvan, see my post three posts above yours for the likely source of the first bug you described. I can't speak to the second, mostly because I don't know what you meant by "They were coming out around 40 DPS options like the Warp-Spring Coil or the DST". You seem to be missing a word. Lower? Higher?

Dorvan · 04/29/08, 8:56 PM

Oops, I forgot an "above" in there which I've edited into the original post (though I might argue you should infer that from the statement that the AP trinkets were being overvalued

). I'll try your fix to the first problem as soon as I get to where my spreadsheet is...looks like I just kinda spaced on that one :P

Ashere · 04/29/08, 9:39 PM

Originally Posted by Jakani

If the hemo debuff is included in the analysis, it's already accepted that a hemo build offers more Raid DPS than a full combat build (up to one rogue). The real question is: does 0/40/21 beat 11/27-8/22-3? I'm pretty sure tri-spec still comes out on top there. Do you have a link to the post with its findings?

found it:
WoW-Europe.com Forums -> Combat Swords rogues, want 20+ free DPS?

Note: I'm rather sceptical about this as well. There would have to be a huge amount of Combat Potency procs on top of the ones to compensate for Relentless Strikes, to compensate for the talents missing compared to trispec: Malice, Ruthlessness, Murder and 1-2 points in Dirty Deeds. And trispec is already behind combat in full T6 with haste & hitcap, so it has to compensate for that too.

Cyn · 04/29/08, 9:52 PM

The spec does 'outdps' pure combat, only when the hemo buff to the raid is considered personal dps, but we allready knew this.

The other thing is that the default dirty deeds model just isn't realistic, assuming you have +20% damage for 35% of the fight is just flawed, for one, it's already shorter because you are doing more damage(!), also considering that it's stacked with mage/warrior executes, makes it even less significant.

Draiko · 04/29/08, 11:33 PM

Originally Posted by Dontmindme

Draiko ---
I need a lot more to go on than that. Neither item has gem slots which is the usual culprit as people lose their meta-gem requirement. Is the Upgrade Macro showing this? Is this a change by hand? Which version of the spreadsheet are you using? PC or Mac? Open Office? Are you sure you are looking at the right items and didn't change anything else at the same time?

Sorry it took so long to reply to this:

I don't have a meta, this is what I am seeing both from the upgrade macro (click the one next to gloves) and when I manually change my gloves from one to the other. Using version 2.4.0.7A (I think, no earlier). Running on a PC (Windows XP). MS Excel. I'm positive I have not done anything else to change the results.

Still having this issue too, any ideas?

Thanks.

Edit: My unbuffed DPS drops by 0.12, but my buffed DPS goes up by 2.80.

Domo · 04/30/08, 2:29 AM

I have been using this DPS spreadsheet for quite sometime. Unfortunately, after a long try figuring out the reason why my DPS in raid can never be even close to what the spreadsheet has given me. According to the spreadsheet it shows a 4s/5r cycle which I tried to maintain on every boss fight. Of course, I am hitting the boss from behind, and popping trinket + bf etc. whenever CD is up. I have checked my recount to check the melee (white dmg) to see if it matches what the DPS sheet calculate, but it seems to be slightly different. For example, DPS sheet shows 65% of white dmg, 25% SS, and 5% others. It seems that according to recount it's showing I am getting 59% white dmg, 27% SS. Is there a catch or timing in between the SS? What exactly is the catch?

The last one question is regarding to the "SnD Cut" that was defined on the spreadsheet, I am still a little confused about it; "snd cut means wasting snd time instead of combo points, instead of waiting till snd runs out, you refresh it as soon as possible.". Does this mean I should always refresh and spam my SS to build combo point whenever possible to keep my SnD up even when it still have 5-7 seconds left on it or should I waste my SnD time then build up the combo point when I need to?

If anyone could please help me out I will really appreciate this. Thanks in advance

drumbum · 04/30/08, 2:29 AM

Originally Posted by Jakani

If the hemo debuff is included in the analysis, it's already accepted that a hemo build offers more Raid DPS than a full combat build (up to one rogue). The real question is: does 0/40/21 beat 11/27-8/22-3? I'm pretty sure tri-spec still comes out on top there. Do you have a link to the post with its findings?

No, it definitely is NOT accepted that Hemo beats Combat Swords at the T6 gear level. In fact, the spreadsheets have told us the opposite. This is covered in the first post of the Roguecraft 101 thread.

That said, I've never seen 0/40/21 beat Combat swords no matter what gear level. Tri-spec Hemo has always been superior to Combat Potency Hemo.

Originally Posted by Cyn

The spec does 'outdps' pure combat, only when the hemo buff to the raid is considered personal dps, but we allready knew this.

See above.

The other thing is that the default dirty deeds model just isn't realistic, assuming you have +20% damage for 35% of the fight is just flawed, for one, it's already shorter because you are doing more damage(!), also considering that it's stacked with mage/warrior executes, makes it even less significant.

Dirty Deeds is not included in a 0/40/21 build.

• Rerox · 04/30/08, 2:49 AM

Originally Posted by Domo

I have been using this DPS spreadsheet for quite sometime. Unfortunately, after a long try figuring out the reason why my DPS in raid can never be even close to what the spreadsheet has given me.

First of all, check the buffs and debuffs in the sheet and compare them to what you really get in your raid. That's one hell of a difference in some raids and it can be the cause for several hundred dps difference.

Originally Posted by Domo

The last one question is regarding to the "SnD Cut" that was defined on the spreadsheet, I am still a little confused about it; "snd cut means wasting snd time instead of combo points, instead of waiting till snd runs out, you refresh it as soon as possible.". Does this mean I should always refresh and spam my SS to build combo point whenever possible to keep my SnD up even when it still have 5-7 seconds left on it or should I waste my SnD time then build up the combo point when I need to?

SnD cut means to cut short SnD time in favor of wasting combopoints, when you have accumulated 5 CPs, yes.

But most fights, it is more useful to wait for one or two energy ticks, when you have 5 CPs and SnD still running, instead of a) wasting CPs or b) cutting SnD early. As long as you are not energy capped! As long as the energy ticks only fill up your bar it's no difference if you wait and use the energy later. Just cut SnD before you would waste energy.

drumbum · 04/30/08, 3:01 AM

After doing some poking around, it seems the spreadsheet is treating Hemo incorrectly. Right now, when you enable the "Include Hemo Debuff DPS Estimate" option, all it does is add a static +42 to your average/min/max MH and OH damage. If I'm missing another reference to this cell, let me know, but by searching I wasn't able to find any other reference.

There are several problems with this. First, it's not safe to assume every attack you deal will consume a Hemo charge because many will be used by the rest of the raid. It would be more suitable to calculate your personal DPS assuming no Hemo debuff exists, then include some value such as 42*10*(1+average_crit_chance_of_raid)*(1-armor_reduction). Unfortunately it's pretty difficult to estimate this accurately so you have to make some assumptions.

Second, this causes the Hemo debuff to be overcounted when MHavg is used to calculate the damage done by the instant damage from Hemo. Since Hemo is 110% weapon damage, for example, it's also multiplying the 42 damage by 110% although this has been proven not to be the way it works in game.

I'm not sure if this explains why Hemo is coming up ahead of combat swords in the alpha, but it's at least something.

drumbum · 04/30/08, 3:04 AM

Originally Posted by Rerox

SnD cut means to cut short SnD time in favor of wasting combopoints, when you have accumulated 5 CPs, yes.

But most fights, it is more useful to wait for one or two energy ticks, when you have 5 CPs and SnD still running, instead of a) wasting CPs or b) cutting SnD early. As long as you are not energy capped! As long as the energy ticks only fill up your bar it's no difference if you wait and use the energy later. Just cut SnD before you would waste energy.

Well, just to clarify a bit on this, since the person you responded to said he used a 4s/5r cycle, then he would want to always refresh SND with 4 combo points, never 5, even if there is a lot of time remaining on it.

Although in practice there's very little difference between 4s/5r and 5s/5r in damage output, so it wouldn't really matter much. It is a bigger difference on the shorter cycles though.

todemax · 04/30/08, 3:46 AM

Originally Posted by drumbum

After doing some poking around, it seems the spreadsheet is treating Hemo incorrectly. Right now, when you enable the "Include Hemo Debuff DPS Estimate" option, all it does is add a static +42 to your average/min/max MH and OH damage. If I'm missing another reference to this cell, let me know, but by searching I wasn't able to find any other reference.

By changing the raidcrit way to the right in the gear buffs tab, and by changing the armor of the boss, one can assure oneself, that the "Include Hemo Debuff DPS Estimate" option is in fact including both these values, and is not just static as you suggest.

drumbum · 04/30/08, 4:52 AM

Originally Posted by todemax

By changing the raidcrit way to the right in the gear buffs tab, and by changing the armor of the boss, one can assure oneself, that the "Include Hemo Debuff DPS Estimate" option is in fact including both these values, and is not just static as you suggest.

Good catch, I see it now. It's adding 10 charges of Hemo in sell X8 on Buffed Cycles sheet. However, this means that the Hemo debuff is added twice for your own attacks. It seems to me like the +42 should be removed entirely from your MHavg and OHavg values.

Kag · 04/30/08, 5:04 AM

Originally Posted by Dontmindme

In further news...
I've decided to occasionally upload my current working version (Excel only). This should have all fixes (and some upgrades) up to the date listed. Of course it could also have undiscovered bugs from new coding. The Excel version still works in Open Office last I knew (just need to deal with the Copy buttons). Mac is a little more problematic.

I believe it "may" be made to work with as few as the following steps:
1) Go to Tools-->Macro-->Visual Basic Editor. In the Visual Basic Editor, go to Tools-->References and uncheck Microsoft XML, v6.0 if it will let you.
2) Cut the Import from Wow Armory button on the Gear _ Buffs sheet

I explained in an earlier post what changes needed to be made to the PC version to make it work on the Mac without loosing any functionality in the PC version. I think the biggest problem is that out of the hundreds of buttons used in the spreadsheet, two of them are not of a type that is cross platform. It takes just a couple minutes to replace them (Armory Import and CreateLookup) with regular buttons that work just as well.

I think I found a way to keep the msxml code unmodified (i.e. it doesn't need to be removed completely like it is in the current Mac spreadsheet and the code doesn't need to be conditionally compiled like I suggested previously and the msxml6.dll reference does not need to be removed). On a PC do the following:

1) Download and install msxml6_SDK.msi

2) Use the midl command line tool (available with the Microsoft Windows SDK v6.1 (Windows SDK for Windows Server® 2008 and .NET Framework 3.5)) to compile the msxml6.idl file. You may need to update the PATH environment variable to include the folders that contain needed exe and dll files:

Go to My Computer->Properties->Advanced->Environment Variables->Path and add the paths of any exe or dll files that midl complains about (also add the path for midl itself). e.g.:

midl.exe       C:\Program Files\Microsoft SDKs\Windows\v6.1\Bin
cl.exe         C:\Program Files\Microsoft Visual Studio 9.0\VC\bin
mspdb80.dll    C:\Program Files\Microsoft Visual Studio 9.0\Common7\IDE

At the command line prompt enter:

cd C:\Program Files\MSXML 6.0\inc
mkdir defaultoutput
midl msxml6.idl /out defaultoutput /I "C:\Program Files\Microsoft SDKs\Windows\v6.1\Include"

3) Copy the created msxml6.tlb file from the defaultoutput folder on the PC to the Macintosh (place it in the same folder as the RogueDPS.xls spreadsheet which has the fixed buttons), and rename it to msxml6.dll

The tlb file is a type library that contains all the interface information for the msxml6 library so that the Visual Basic stuff in the spreadsheet will compile and you can also browse the library. The tlb file doesn't contain any code though, so Import from Armory will still not work on the Mac (you'll get a "Run-time error '429': ActiveX component can't create object" error if you click "Start Import") but at least you now have a worksheet that can be used on a Mac or PC without any changes.

Ashere · 04/30/08, 5:30 AM

Originally Posted by todemax

By changing the raidcrit way to the right in the gear buffs tab, and by changing the armor of the boss, one can assure oneself, that the "Include Hemo Debuff DPS Estimate" option is in fact including both these values, and is not just static as you suggest.

hmmm, I'm pretty sure DMM fixed this a few versions back, so that:
-"hemorrhage" from the gear buffs page adds 42 dmg to each MH and OH hit
-"include raid debuff estimate" from the talents page adds in 10 hemocharges/hemohit with armor & crit adjusted values
-"include raid debuff estimate" from the talents page causes the calculations to ignore the hemorrhage buff from the gear page
So basically, double counting shouldn't happen anymore, while making proper estimates for hemo-rogues would use the option from the talents page instead of the gear buff page.

Was this reverted at some point? Since this means the spreadsheet has dropped the reliability of hemo figures if it was :-/