Jump to content
  • Sign Up

Kittymarks - The derpy girl's comprehensive benchmark


Recommended Posts

E: Kitty's ended her testing due to lack of time and faster update cycle as she can't get her tests done between balance patches. But that means more time for actually playing those builds and Kitty's uploading lots of videos about that on her youtube-channel http://youtube.com/LadyKitty. And she's also occasionally streaming on them at https://twitch.tv/ladykittygw2.

E: Now that balance patch arrived, Kitty's starting her benchmarks again from 0 to account in the changes. As GotL was made into a might-spam trait and warrior's PS was nerfed, it's not almost a certainty (like 90% of squad running PS+druid+chrono in both subs) anymore that there's Empower Allies and Spotter in both subsquads since there's now some other good alternatives for both the slots that used to be reserved for 2nd PS warr and 2nd druid. So, Kitty takes a minimalistic approach with her benchmarks and drops EA and Spotter from her test setting, retaining banners and spirits as both boof whole raid squad and it's likely there's at least one (DPS or PS) warrior bringing banners and one ranger bringing spirits for the squad.These changes obviously mean that Kitty's numbers will now be even further "what you can expect at least from this build".

Kitty's this far found other benchmarks quite narrow and slightly non-realistic, so she started her own benchmarking project after PoF landed. Now she's close to completing them (just lacking power ele, condi warrior and some not-so-used builds), so guess it's a good time to share a bit. 126 benchmarks at the time of writing this and there will be about 180+ when Kitty's done.

The benchmark list: https://kittymarks.com/kittymarksRelated video channel: Kittymarks Youtube-channel

And like usual: don't be a build nazi. If the squad isn't speedrunning or doing No Updrafts-Gors, pretty much any build with over 20k DPS works for basic DPS role if played properly.

Note: these aren't "the best of 50 tries"-type benchmarks, but what Kitty could pull off within 8 tries (or less if Kitty felt like she did the best she could). Kitty's a derpy player with high ping, slow mind and reaction time and these benchmarks were done with a normal mouse and keyboard, so any player with full ascended gears (including poor multi-classers) should be able to do about as well at least. And speedrunners should do way moar <.< >.>

E: Kitty also added "Boss DPS Requirements"-tab with minimum squad total DPS numbers to avoid enrage.

Link to comment
Share on other sites

  • Replies 379
  • Created
  • Last Reply

I was like "ummmmm... what" when reading this. But actually despite all of the uh kitties, that's actually pretty amazing. Like if you're ever wondering for examplw... ohhh I wonder how much dps a power gs soulbeast or a gs/hammer DH can do, well dw kitty marks knows the answer to that question! Like you have all of the cancer and all of the love all there in full monty.I'm kinda blown away by this away, that must have taken forever to make, GJ.

Link to comment
Share on other sites

Also for anyone who says they can do more dps with so and so, thas pretty much perfect from what I've seen, thats normal max deeps for oceanic players, it's probably close to impossible to get much higher than that.Might be worth showing the metrics in arc dps if you record anymore, just so it so it protects you against the seething Americans who live right next to the servers.Also be worth using more formal language, people will take you more seriously.

Link to comment
Share on other sites

So many of these (possibly almost all of them) were done too quickly. You can't just jump into a build, give it 8 attempts, and call it good.As an example: I gave your greatsword/hammer dh build a few tries and managed to parse 15.5% higher after just 6 attempts, and I had 165 ping at the time (playing on EU from NA).

https://i.imgur.com/sV5ipMP.png

Until you hit the 250+ ping region and you're on a class with a lot of instant-cast skills (FA ele for example, every time you have to reattune to air as soon as you crit), ping really wont make much of a difference. I feel bad for those playing from asia/ocx but if you're just cross region NA->EU or EU->NA you don't have as much of an excuse.

Link to comment
Share on other sites

@Feanor.2358 said:As an ele main I just wonder why Kitty didn't test the actual meta power Weaver. All I see are some snowflake builds (FA, really... with that kitten air auto on staff). I encourage Kitty to try staff Fire/Air/Weaver with Bolt to the Heart.

Power elementalist builds are next on Kitty's list, just when her ping calms down a bit and she's not too tired to focus.

@cat.8975 said:So many of these (possibly almost all of them) were done too quickly. You can't just jump into a build, give it 8 attempts, and call it good.As an example: I gave your greatsword/hammer dh build a few tries and managed to parse 15.5% higher after just 6 attempts. I only have 4 power infusions, and I had 165 ping at the time (playing on EU from NA).

sV5ipMP.png

Until you hit the 250+ ping region and you're on a class with a lot of instant-cast skills (FA ele for example, every time you have to reattune to air as soon as you crit), ping really wont make much of a difference. I feel bad for those playing from asia/ocx but if you're just cross region NA->EU or EU->NA you don't have as much of an excuse.

Kitty's ping averages between 200-250ms, playing on NA-servers as finnish player (connected thru mobile phone hotspot thru mobile network). And the ping isn't the main issue, but depression and meds slowing the brain and making focusing difficult. Though ping does like to mess up weapon swaps and subsequent skills a lot.One thing Kitty noticed from your test was that you used a large hitbox instead of small, which does make some difference. Your rotation was slightly more optimized, Kitty admits. If Kitty remembers correctly, she used 3 tries on that benchmark, so guess she better add hammer-builds to her "test more"-list.Kitty had already planned on giving more tries to builds that didn't get full 8 tries once Kitty's done with the bulk. Couldn't really do 8 tries on every build right away 'cause foods are very costy and Kitty's benchmarking time is currently very limited 'cause part-time work and inability to focus well enough 80% of time. And 8 tries on one weapon-combo/spec takes about 30-40 mins (depending on DPS, not including a couple rotation-pondering tries per weapon combo) so...you can figure why Kitty mainly saved full 8 tries to the most complicated to play/easy to mess rotation-builds during the initial benchmarking.

Link to comment
Share on other sites

That's the problem I noticed, too. In the end, I'm not entirely sure about this thing. It's certainly nice to see "benchmarks" for a ton of builds, but is it helpful when you have absolutely no idea where to place it between 100% god player and 0? With the qT benchmarks, at least I know that I'm sitting at x% of something that's probably as close to the theoretical maximum as you can get.

Link to comment
Share on other sites

@CptAurellian.9537 said:That's the problem I noticed, too. In the end, I'm not entirely sure about this thing. It's certainly nice to see "benchmarks" for a ton of builds, but is it helpful when you have absolutely no idea where to place it between 100% god player and 0? With the qT benchmarks, at least I know that I'm sitting at x% of something that's probably as close to the theoretical maximum as you can get.

Well, Kitty does usually out-DPS pugs in most encounters unless she's playing a build with long ramp-up time against quick ones at short-phased bosses or playing with very skilled ppls. Comparing the Kittymarks with qT's, Kitty's usually about 15-25% behind the equivalent qT-benchmark. So, guess Kitty's somewhere around 75-85% in that regard. Above an average pug, below hardcore speedrunners.Also, guess Kitty better mention that unlike qtfy, she doesn't use Pinpoint Distribution for her benchmarks as engis aren't that usual subsquadmate. Which does have some impact on condibenchmarks when comparing Kittymarks and qtfy's.

Link to comment
Share on other sites

@LadyKitty.6120 said:One thing Kitty noticed from your test was that you used a large hitbox instead of small, which does make some difference.

Hitbox size only matters for scepter builds, hammer/gs are unaffected.

Kitty had already planned on giving more tries to builds that didn't get full 8 tries once Kitty's done with the bulk. Couldn't really do 8 tries on every build right away 'cause foods are very costy and Kitty's benchmarking time is currently very limited 'cause part-time work and inability to focus well enough 80% of time. And 8 tries on one weapon-combo/spec takes about 30-40 mins (depending on DPS, not including a couple rotation-pondering tries per weapon combo) so...you can figure why Kitty mainly saved full 8 tries to the most complicated to play/easy to mess rotation-builds during the initial benchmarking.

Understandable.

Link to comment
Share on other sites

It will surely help me take you 100% more seriously if you referred to yourself as an actual person using the "i, me and myself" forms instead of referring to yourself as...a kitty...in the third person...food for thought.

Continuing from Aurellian's idea though, you can see that there is a problem when somebody (cat in this instance) can take one of your builds and pretty quickly do up to 15% better then what you're showing us. It doesn't make much sense to show your "100% efforts" when somebody can come along and do 115% because then the 115% becomes the new 100% and pretty much invalidates your benchmark because, effectively, you're not even performing optimally at something you're measuring yourself.

This can quickly become an issue because putting out too many benchmarks that are not the theoretical maximum is quite counter productive, again at least with qt's benchmarks i can easily tell if i'm doing something good or bad simply because their number is a "maximum" or at least a number that gets close to that maximum. Lets take Condi PS for example, with a benchmark of 29.520, i can easily say i'm doing a good job if i get a dps number that is close to or on the target number. In this case i might take your benchmark of Condi ps and go over it with a 10% difference, but does that mean my benchmark is now the new 100%? Am i even performing optimally myself? Maybe someone comes along and dishes out an extra 10% over the number that i measured myself, and we could keep doing this for a few days/weeks until we really hit that theoretical maximum.

Now you've obviously put a lot of thought and effort into it and you can absolutely give yourself points for the time spent measuring your benchmarks and maintaining your website, absolutely well done. But if i may, i would urge you to take a slightly longer time focusing on a single class at once and really trying to get your max number out there so we have something to compare ourselves with. Blaming it on lag or "i can't focus 80% of the time" is not something you can keep waving as an excuse forever...i mean i don't have lag and i can focus pretty much all the time...should i be ashamed of being able to play under optimal settings/conditions? I would certainly hope that i needn't.

Link to comment
Share on other sites

Thank you kitty. It is very interesting to be able to compare some more obscure builds. Even though they do not represent maximum dps I do think it is a fair representation of how much the average player gets on the golem. I am looking forward to seeing how you refine it in the future. :)

Also it is important for everyone to keep in mind that these tests are exclusively done on a small hitbox and that can make a difference of anywhere between 5-15% dps. So if you would like to compare your dps with hers make sure you use the same standards. There is a section on her website that shows her setup.

Link to comment
Share on other sites

I think in the long run what "Kitty" is trying to show if nothing else is that while qT's work is well done and appreciated, as they themselves say the fights can be done on builds that may not be 100% up to snuff with their own.Someone here has said many times and its great.... Skill>build.So yea, just showing us that all things are possible.

(Although on your condi renegade rotation you say swap to dwarf? And Soulcleave sucks so much energy that razor is better, imo) ;)

Ok, now I'm confused.....your rotation on your site says to use Soulcleave yet in your video you use Razorclaw and its pretty much Fennec's rotation with a few extras thrown in? Is this a troll post....? If not, still a lot of work and hopefully it helps someone! ;)

Link to comment
Share on other sites

It's weird. On the one hand, when I do the QT tests I can get nearly as high as QT on some of their builds. On the other hand, there are some builds where I can't even get close, no matter what I do.

The scourge is an example of this. qT says it caps at 31k, but after over an hour of testing I can't break 25k. In spite of watching the videos and copying the rotation, it just doesn't work out.

Link to comment
Share on other sites

@LadyKitty.6120 said:And like usual: don't be a build kitten. If the squad isn't speedrunning or doing No Updrafts-Gors, pretty much any build with over 20k DPS works for basic DPS role if played properly.

Finally I can kick gs warriors without being considered an elitist.

Link to comment
Share on other sites

I will (probably) never understand how anyone reaches such high dps. I mean I'm a non-raider so I don't expect to understand it much, but I do fractals and smash through open world mobs with ease and I'm usually around 5-10k on any class in that aerodome test.

Just a different world/mindset I guess!

Link to comment
Share on other sites

First off, it's nice to see someone give multiple rotations a try.

Second, the only thing these benchmarks really show is how easy or fast some classes and their dps builds work and how hard it is to master them. Which granted is quite useful (unfortunately with only such a small amount of practice runs, this is literally looking at how someone completely new would perform).

The amount of time spent honing or practicing is negligible and to short to be taken seriously.

As others have mentioned, you are all over the place comparison wise. The quantify benchmarks are at least all on a similar level comparison wise (top 95th percentile).

Link to comment
Share on other sites

@Blood Red Arachnid.2493 said:It's weird. On the one hand, when I do the QT tests I can get nearly as high as QT on some of their builds. On the other hand, there are some builds where I can't even get close, no matter what I do.

The scourge is an example of this. qT says it caps at 31k, but after over an hour of testing I can't break 25k. In spite of watching the videos and copying the rotation, it just doesn't work out.

Depending on when you tested and if you have 18 +condi infusions 25k seems right.

Link to comment
Share on other sites

@Blood Red Arachnid.2493 said:

The scourge is an example of this. qT says it caps at 31k, but after over an hour of testing I can't break 25k. In spite of watching the videos and copying the rotation, it just doesn't work out.

Kitty benchmarked it again yesterday for post-fix numbers (since qT refuses to do so before balance patch) and she got 25225 without infusions. Before the fix, Kitty got 30,5k. So qT's estimation of 6-7k DPS drop seems pretty correct. (though it still does some decent DPS).

Link to comment
Share on other sites

What I like about these benchmarks, compared to gtfy, is that kitty put alot more builds in the test, unoptimised as they are. And now the speed-runners-meta-builders-mid-maxers will ask: "So what? I just wanna know what does the best damage so I can run that." These kind of benchmarks give hope to those that don't precisely enjoy playing the meta. All those poor non-meta classes/builds that keep getting shut down by the "vets" when they mention their class as dps in raids, now have hope, that they could perfect a rotation and find a group to raid with.

And am I the only one that turns their nose from the smell of elitism when they read comments like "if you'd stop calling yourself kitty I might take more serious"? Like we're doing rocket science and not play MMO Role Playing Game. My bad, even physicists have more humour than some of these guys. But I think this attitude works when you're a child that wants all the toys for himself.

But I do agree with something. Kitty, you can't just come up with these excuses for ever about your unoptimised bechmarks on your free site that we can use or not. I mean, come on. You get it, right? Right?

Link to comment
Share on other sites

@MrRay.3027 said:But I do agree with something. Kitty, you can't just come up with these excuses for ever about your unoptimised bechmarks on your free site that we can use or not. I mean, come on. You get it, right? Right?

Well, even if Kitty tried 50 times, she prolly couldn't pull more than 1k better benchmark on most of these builds due to heavy tendency of messing up at some point and simply too slow visual reaction time for some swaps. Ofc using a gaming mouse would help a bit (and Kitty has a new Logitech G502 waiting for the time she's done with benchmarks and she can start raiding/fractaling properly again) but since not all peoples have fancy gaming mice and keyboards with macro keys, Kitty sticks to basic mouse with scroll, left- and right-click for these benchmarks so that even poor peoples can compare to these.

Link to comment
Share on other sites

Wow Kitty, this is really amazing work! And this just goes to show that there are so many viable options that's not just qT-copy paste. I will look more at this when I get home, but I feel that I can link this to loads of people who will find a build they're happy running. :D Now the main problem will probably find people who want to stick with Chrono/PS/Druid when there are sooooo many cool options around haha.

Love your work, really amazing and it'll be great to see some more updated ones when you get around to test a bit more :)

To all you depressive people who think 8 tries is too little, and that kittys can't be taken seriously: considering Kitty is one person that has tested way over 100 builds and landed at kittymarks that are definitely strong enough to be allowed in raids, I suggest you go and do it better. No? Didn't think so.

Link to comment
Share on other sites

@Ildrid Ildhjertet.2489 said:

To all you depressive people who think 8 tries is too little, and that kittys can't be taken seriously: considering Kitty is one person that has tested way over 100 builds and landed at kittymarks that are definitely strong enough to be allowed in raids, I suggest you go and do it better. No? Didn't think so.

So dismissive, people like qt and snowcrows put in hours and hours of effort over hundreds and hundreds of tries just to get one benchmark out, farting out a quick rotation on a dummy and calling it a day is hardly representative of anything.

Link to comment
Share on other sites

@Verenhimo.3296 said:So dismissive, people like qt and snowcrows put in hours and hours of effort over hundreds and hundreds of tries just to get one benchmark out, farting out a quick rotation on a dummy and calling it a day is hardly representative of anything.

I'm not saying they do a bad job, and I'm aware they do loads of testing, but they are a lot more than 1 person, and finding the most optimal build for anything is their focus after all. Having someone list a multitude of strong builds with high dps is something the community needs more of, and with Kittys ground work done maybe someone are willing to expand on it or help do additional testing on a few builds.

Link to comment
Share on other sites

@Verenhimo.3296 said:

To all you depressive people who think 8 tries is too little, and that kittys can't be taken seriously: considering Kitty is one person that has tested way over 100 builds and landed at kittymarks that are definitely strong enough to be allowed in raids, I suggest you go and do it better. No? Didn't think so.

So dismissive, people like qt and snowcrows put in hours and hours of effort over hundreds and hundreds of tries just to get one benchmark out, farting out a quick rotation on a dummy and calling it a day is hardly representative of anything.

I think he's more referring to the people playing down her efforts. Many people here are missing what she is doing - these are benchmarks done by a player using non-meta builds so that those who wonder sometimes what they can do at least have some form of mark to hit.

And to be honest, GW2 player fixation on golem benchmarks is beyond reasonable , and I don't mean any disrespect to qT/SC and their efforts. What they offer is the equivalent of Simulationcraft, and even then isn't nearly as good. For those who don't know what Simulationcraft is, it's a program used in WoW to simulate in a closed environment a fight using controllable buffs, actions and gear and done by a computer. It's perfect execution in a situation that never comes up, exactly like what qT benchmarks are but done by a computer so there is 0 error.

However, even with that tool, that is not how benchmarks and class performance is rated. Players use Simulationcraft to see what the class can do in (mostly) single target 0 movement scenarios to see the theoretical maximum and use it as a guideline for what to expect in a raid and how to gear. The real important information comes from Warcraftlogs which shows class performance in actual raid scenarios and how that simulated DPS translates into real world application. While a class can simulate very high, it sometims doesn't translate into very good DPS overall as mechanics can very much change the lay of the land. In the end, no one cares about the simulated DPS, just what warcraftlogs shows and how the class is rated there because it's real data, showing how they will perform in the actual fight.

Yes GW2 is different, but the theory is the same.

Nice job kitty, while there could have been more tries to get more data, it's very refreshing to see that you have tried multiple builds and posted the videos. Yes, better players can get more, but at least it's a starting point.

Link to comment
Share on other sites

@Verenhimo.3296 said:

@"Ildrid Ildhjertet.2489" said:

To all you depressive people who think 8 tries is too little, and that kittys can't be taken seriously: considering Kitty is one person that has tested way over 100 builds and landed at kittymarks that are definitely strong enough to be allowed in raids, I suggest you go and do it better. No? Didn't think so.

So dismissive, people like qt and snowcrows put in hours and hours of effort over hundreds and hundreds of tries just to get one benchmark out, farting out a quick rotation on a dummy and calling it a day is hardly representative of anything.

Well, if nothing less, Kitty's benchmarks do show what lots of off-meta builds can do at least which other benchmarkers don't bother with. True, Kitty's performance does vary heavily and she usually fails to pull off decent damage (in the eyes of mainers) with the more complicated builds with such low amount of tries and thus her benchmarks aren't the best for comparison. But then again, one point here is how reliably an average person can pull off good DPS.And like many peoples who've run with pugs know, rather few pugs really bother with mastering the complicated rotations and thus fail really hard in actual raid situations when their semi-learned rotation gets interrupted and then they panic and asdefonrhgso.Kitty has actually spent some time before at golem, trying to master condiengi, condireaper and power DH metarotations and she's actually been in raids and fractals with them, unlike with 80% of the builds tested. But unless Kitty started playing one of those as her primary main, she won't be able to pull them off to maximum in actual gameplay situations just 'cause especially condireaper metarotation is very easy to mess up. Just if you knew... Or constantly trying to memorize when condiengi's skills come off the CD while trying to focus on doing the boss mechanics...durr, Kitty can't do.For those reasons, Kitty tries to keep her rotations relatively simple (yet decently effective) so that she'd also have a chance of actually pulling them off properly in raids. And guess there's some other ppls, too, who can't pull off or are just too plain lazy to hone their rotations to perfection.

Not to forget that there exists these thingies called "peoples who want to try a new class" and get scared upon seeing the super-effective-yet-superlong rotations. Seeing simple yet decent rotations might actually lower ppls' threshold of trying out stuff. And...do we need to be always optimized and super-effective to score the kills?Kitty thought the main point is actually getting the kills, not how fast you get them. Unless you want to be cool and try to kill the bosses with half of enrage timer left. For that kind of peoples, sure it's good to mixmax optimize. But when trying to do that with pugs (Kitty only raids with pugs atm 'cause her guildies aren't ready for raids yet), that doesn't work so often.For some hilarious example, Kitty once joined a no-greens VG metagroup. Though the whole squad was experienced at no greens, we failed 3 times (around 20-30% HP left) before trying the normal way. And Kitty got to see how those experienced no-greeners couldn't get past first split.A few days later, Kitty did VG the normal way with some non-experienced random guild group (ofc that squad didn't have druids or warriors). And it was smooooth though DPS wasn't anything fantastic, just good enough to have 10 secs left on timer at kill.

Sorry about possibly messed up wall of text. Kitty managed to forget lots of stuff she intended to write during the process.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...