Special Forces Training Area- weird results?

[Messenger.7450]

Just wondering if anyone has experienced this.

I've recently (and currently) been doing some tweaking of my DPS build using the Special Forces Training Area. Standard Kitty Golem so it doesn't take too long. No additional effects; he just stands there doing nothing but taking hits because the test is purely about general DPS and not any special scenarios.

So I've got Old Build and New Build. They're not that different. Just changed some stats and infusion on some gear. No changes to Traits nor Skills. It's just fine-tuning and nitpicking of Attributes.

I do 3 tests per build I'm testing:

1. One without any buffs except for the boons the build itself provides.
2. As above but with food and utility buffs, since I always keep those up in PvE and WvW, and they're cheap and easy to do.
3. Just for fun, as above but with all boons turned on via Arena Console to see how high I can go. All of them: Might (25 stacks), Fury, Retaliation, Protection, Resistance, Stability, Aegis, Alacrity, Quickness, Regeneration, Swiftness, Vigor.*

My gear is ascended, not Legendary, so any stat switching I do is via Mystic Forge. For the tests, I did Old Build first, Tests 1 to 3, then switched stats, and New Build tested1-3. No going back to test Old Build unless I switch stats again.

I at least made sure the sequence of skills I started each test with was the same. After that, it's a mad dash to do the most damage. That said, you're given 5 DPS results by the golem per run, so you've got multiple data points to avoid mistaking any extremes as actual results of your test.

For Test 1, New Build has higher DPS than Old.For Test 2, New Build has higher DPS than Old.For Test 3, Old Build has higher DPS than New.

I was shocked. I repeated Test 3 for my New build several times. ALL OF NEW BUILD'S RESULTS WERE WORSE THAN OLD'S WITH ALL THE BOONS ON.

So I'm wondering how getting all those boons somehow made my DPS worse. Without them turned on from the Arena Console, New Build beat Old handily. This was consistent. After detecting this weirdness, I even turned off all boons and repeated Test 2 (only self-provided boons plus food and utility), and it matched previous results. Somehow, having all those boons gave me an anomaly.

So... Either this is a bug, or there's really way for boons (combined with specific stats?) to reduce your DPS?

---

• Now that I think about it, I didn't need to turn on the defensive boons...

[Khisanth.2948]

try it without alacrity

[Ayrilana.1396]

Difficult to point out specifics without knowing what the two builds were.

[Noah Salazar.5430]

prob one build was for open word, and your old one was for raid setting

what's that mean?

your new build genereting more self boons/vulanbirity causing more dmg, while in raid someone providing it to you, so you do more dmg by other skills while 25might and etc

[Messenger.7450]

@Khisanth.2948 said:try it without alacrityDid as you suggested. DPS went down, so it's not Alacrity causing the anomaly. Will post results below.

[Messenger.7450]

@Noah Salazar.5430 said:prob one build was for open word, and your old one was for raid setting

what's that mean?

your new build genereting more self boons/vulanbirity causing more dmg, while in raid someone providing it to you, so you do more dmg by other skills while 25might and etcActually they're both almost the same (open world) build. Minor changes only. Which is why I'm mystified about what's happening.

[Messenger.7450]

@"Ayrilana.1396" said:Difficult to point out specifics without knowing what the two builds were.

Old build: http://gw2skills.net/editor/?PKVAQlRwcYUsKWJWUX/tKA-zxgYhgO2RB57HEeJUgTwkHkwoCUV2A-e

New build: http://gw2skills.net/editor/?PKVAQlRwcYUsKWJWUX/tKA-zxgYhoAdsD57HEeJUgTwkHkwoCUV2A-e

The only change was the stats of the weapon.

Clarifications:

1. I only tested Rifle in all my tests.
2. No infusions or Sigils applied yet to focus on the change of weapon stats.
3. Opening sequence of skills are: Signet of Rage, Rifle 2, Rifle 3, Signet of Fury, Berserk, Gun Flame, Blood Reckoning, Wild Blow, Gun Flame, then I go nuts.

[Messenger.7450]

Original test that weirded me out.DPS/Damage/Time

Old Build:Test 1 (no consumables): 5099/200300/39, 5001/400086/80, 5022/604439/120, 4935/800296/162, 5014/1000249/199; average DPS 5014.2Test 2 (w/ Spicy Moa Wings and Tin of Fruitcake): 6037/204285/34, 5595/406837/73, 5183/602526/116, 5205/800934/154, 5350/1008515/189; average DPS 5474Test 3 (consumables + all boons): 9870/203316/21, 9347/400799/43, 9292/600661/65, 8986/801203/89, 9290/1001843/108; average DPS 9357

New Build:Test 1 (no consumables): 5831/200827/34, 5454/408832/75, 5072/600071/118, 5033/801702/159, 5145/1003135/195; average DPS 5307Test 2 (w/ Spicy Moa Wings and Tin of Fruitcake): 6083/204143/34, 5595/400829/72, 5539/601134/109, 5512/801641/145, 5401/1000473/185; average DPS 5626Test 3 (consumables + all boons): 8953/204850/23, 8869/401236/45, 9336/601605/64, 9025/800239/89, 9400/1002394/107; average DPS 9116.6

I really don't get how New Build beats Old in Tests 1 and 2, but loses in 3 when it's got tons of boons for doing more damage.

I started testing New build over and over since I got these results. These seems to be quite a variation in results, but when I took the average of all the DPS results, they match what you see above.

I've got 10 Spirit Shards to change my Rifle back from Berserker's to Assassin's so I can do more tests using Old Build.

[Dawdler.8521]

That is negligable difference.

Your average dps in test 3 is heavily skewed by the 1000 point difference in the first run, the other results are roughly equal (with 3 of 5 runs higher dps on the new build).

If those first run numbers are repeatable (ie you didnt just fail with the new build), then it just looks like the simple matter of crit rate beating power in short runs when buffed to the sky.

I'm not a pve raiding pro though.

Edit got the power backwards

[LuRkEr.9462]

The difference is easily mistakes or slowness in rotation, something as simple as cancelling an auto attack chain early. That being said the dps is so low that it looks like a large difference, but when you get to where your doing 25k+ on a golem thats not much at all.

[Khisanth.2948]

That is a 1% difference between builds.

The difference between the high and low roll of the weapon strength of a rifle is 22% ...

[Messenger.7450]

Dawdler.8521, LuRkEr.9462, Khisanth.2948:Thank you for pointing out that the differences are actually pretty small. I've continued doing tests over and over and over, including switching my build back to the older version and going through a whole new slew of them. I see your points because doing a lot of tests showed quite a variation in results.

I did a total of 40 tests to evaluate my build, and then to figure this inconsistency out. I organized it on a spreadsheet and took averages. Etc.

It turns out my Old build was actually the stronger one. The reason it initially looked like New beat Old in Test 1 and 2 yet lost to Old in Test 3 was because when I did Old's Test 1 and Test 2 (only done once per test for Old), I actually got low results. This made New Build look better except in Test 3. The variation in results I found showed that it can go either pretty high and pretty low. So mystery solved.

The big lesson I get here is that you can't just go 1 round with the golem if you want to accurately measure your build's DPS even if it gives you 5 data points per go. You can get extreme results that skews how you view the whole thing. Better to go at least 3, or maybe even more.

[Cyninja.2954]

So let me get this strait, you did 40 test runs to see if exchanging 1 items stats from berserker to assassin was worth it?On a rifle warrior build.Running sinister trinkets and berserker runes.On a 3 signet build with not a single condition damage ability either on the rifle nor on your utilities.

Okay. Hope you at least had fun.

FYI: the benchmark for semi useful performance of builds with all boons bottoms out at 25k absolute minimum, 30k dps average and 35k+ dps for good builds. Should you require ideas or inspirations for builds, do feel free to check at least metabattle and read through some of the descriptions of the high ranked builds there.

[Blood Red Arachnid.2493]

I will encounter stuff like this on occasion. There's a few things to consider when testing out different rotations, gear types, etc.

(1): Make sure you don't forget anything. There's been at least a dozen times when I was comparing gear and I realized I forgot to put the proper runes/sigils on them.
(2): Quickness does not affect all skills evenly. There are a couple of skills in the game that are animation locked in one way or another. For example, Rocket Charge on Scrapper is a DPS net positive only when not under quickness, because it doesn't get faster with the boon.(3): Alacrity has similar differences with quickness. Though instead of it being animation locked, it has to deal with skills that aren't tied to normal recharge.
(4): Conditions don't scale with the same proportions that power does. Tests with maximum might on an otherwise mightless build will change the proportion of damage done between power and conditions. This change won't be reflected in early tests.

It doesn't seem like a lot, yet it somehow seems to creep its way into whatever you're trying out. Working with Thief? Alacrity changes how initiative works with utilities. Mesmer? Boons affect clones and phantasms differently, also depending on how often you shatter. Revenant? Similar issue with Thief, except its energy instead of initiative. There's also animation locked skills all over the place.

Though it sounds like this issue was resolved in other ways, it helps to consider these whenever something... odd happens.

[Messenger.7450]

@Cyninja.2954 said:So let me get this strait, you did 40 test runs to see if exchanging 1 items stats from berserker to assassin was worth it?It was originally just 6 to see which worked better between Assassin's and Berserker's. Then I encountered the discrepancy in DPS results that's the topic of this thread, which is what led to all those extra tests to investigate which figures were correct, all the while I was still evaluating the tweaking of my build.

On a rifle warrior build.Running sinister trinkets and berserker runes.On a 3 signet build with not a single condition damage ability either on the rifle nor on your utilities.Arms trait line and Berserker spec means I do a lot of bleeds and burns in rapid succession. I was only testing rifle here just to see the power-based and crit-based damage of the build, but I also use a longbow, and I usually have Superior Sigil of Earth on both because of my high crit rate.

Okay. Hope you at least had fun.Hey, I learned stuff! :-) Like one single run versus the practice golem is not enough of a test to evaluate a build because of how results can vary. Gonna do at least 3 in the future as the need arises, probably more to be sure.

FYI: the benchmark for semi useful performance of builds with all boons bottoms out at 25k absolute minimum, 30k dps average and 35k+ dps for good builds. Should you require ideas or inspirations for builds, do feel free to check at least metabattle and read through some of the descriptions of the high ranked builds there.I'm not a raider. :-) I'm PvE who does WvW casually for dailies (but at least tries to contribute to his realm).And I'm not going to be happy following someone else's build, even if it means I'm sub-optimal compared to Metabattle's standards. Doing things your way may mean poorer performance, but winning without originality is a pretty empty experience for me.Doesn't mean I don't try to get my homebrews to at least work better- hence why I occasionally go to the Special Forces Training Area.

[Dawdler.8521]

@Cyninja.2954 said:FYI: the benchmark for semi useful performance of builds with all boons bottoms out at 25k absolute minimum, 30k dps average and 35k+ dps for good builds. Should you require ideas or inspirations for builds, do feel free to check at least metabattle and read through some of the descriptions of the high ranked builds there.Yeah that really depends on the purpose.

A condi build which can do like 5k dps on a golem is absolute super OP cancer in WvW according to the forums.

[Cyninja.2954]

@Dawdler.8521 said:

@Cyninja.2954 said:FYI: the benchmark for semi useful performance of builds with all boons bottoms out at 25k absolute minimum, 30k dps average and 35k+ dps for good builds. Should you require ideas or inspirations for builds, do feel free to check at least metabattle and read through some of the descriptions of the high ranked builds there.Yeah that really depends on the purpose.

A condi build which can do like 5k dps on a golem is absolute super OP cancer in WvW according to the forums.

Which build would that be? Condi immobilize druid?

If I assumed a berserker/assassin/sinister, 3 signet, rifle warrior was in any way a WvW build, I would have made the connection to WvW. Not seeing a WvW build here though.

[Ayrilana.1396]

@Cyninja.2954 said:

@Dawdler.8521 said:

@Cyninja.2954 said:FYI: the benchmark for semi useful performance of builds with all boons bottoms out at 25k absolute minimum, 30k dps average and 35k+ dps for good builds. Should you require ideas or inspirations for builds, do feel free to check at least metabattle and read through some of the descriptions of the high ranked builds there.Yeah that really depends on the purpose.

A condi build which can do like 5k dps on a golem is absolute super OP cancer in WvW according to the forums.

Which build would that be? Condi immobilize druid?

If I assumed a berserker/assassin/sinister, 3 signet, rifle warrior was in any way a WvW build, I would have made the connection to WvW. Not seeing a WvW build here though.

I think they were being sarcastic or that’s how I took it.

[Schimmi.6872]

If you have arcdps, you can use the log-feature to get much more details from your test, this might help to find the differences.

[Jijimuge.4675]

Looking at the stats, it doesn't surprise me. One of the clearest differences I can see is that build 1 has lower power but higher precision (and a trivially higher ferocity) compared with build 2. Thus, build 1 will gain more (relatively speaking) from each stack of might it gains,...and 25 stacks of might will add up.

It's just a question of the way the maths works IMO with how crits provide a multiplicative effect, but might's effect is additive to power.