Okay so I tested Onslaught Mk1 vs Onslaught XIV, basically 2 of each, in alternating formation, with 3 LP Brawlers (me in one of them) on the sides to corral the flankers, against double Ordos, in 0.98a-RC8. The results were that...they performed about the same. One finding from testing this setup in 0.98a-RC7 was that the Mk1 pretty much doesn't really need PD, since its Heavy Adjudicators provide PD toward the front and it has vambraces on the sides. However, the XIV does need at least some PD. So after some testing I put a Flak Cannon in the front center medium ballistic, despite it being a premium location, to provide that PD (it was typically enough). The rest of the medium ballistics went to Heavy Mass Drivers (HMD), so that's 4 on the XIV, and 7 on the Mk1. The XIV does have small ballistics though so it also had 4 Light Needlers. 2 of the HMD's on the Mk1 and 2 of the Light Needlers on the XIV were side-facing, but the rest could all hit forward. Large ballistics on both were Hephaestus in all 3 slots.
The average after 3 runs (so, 6 of each) was:
Ons XIV:
weapon total shield armor hull hits count fired tot fl dam/fl uptime
HMD 47940 37325 1923 8693 255 4 372 52103 0.92 1.41
heph 43124 11568 13895 17661 544 3 851 93647 0.46 1.07
TPC 39736 25626 6404 7707 276 2
LN 37020 31419 1268 4334 908 4 1404 56147 0.659 1.77
har pod 22184 9057 4331 8797 51 4 144
total 190004 114994 27821 47191 2033 2771 201897
Ons Mk1:
weapon total shield armor hull hits count fired tot fl dam/fl uptime
HMD 85185 64789 3867 16529 514 7 761 106470 0.8 1.65
H Adj 50216 15136 5231 29849 253 2
heph 47492 12978 17047 17467 665 3 1019 112035 0.424 1.29
har pod 5350 1422 925 3004 10 2 48
total 188242 94325 27069 66848 1442 1827 218505
The Mk1 fired around 20-25% more often than the XIV for whatever reason (maybe because it tended to ram the enemy fleet more often). This meant that the 3 Hephs on the Mk1 did more than the same 3 Hephs on the XIV. Nevertheless, the 4 HMD + 4 LN on the XIV pretty much equaled the 7 HMD on the Mk1. For the Harpoon Pods, the XIV has 70 more OP than the Mk1, so it could fit EMR and ECCM, but I couldn't make them fit on the Mk1, so those were essentially throwaway missiles on the Mk1 due to lack of OP (it did have elite Missile Spec though, while the XIV only had regular Missile Spec). The Mk1 got more damage from the side HMD's than from trying to buff missiles.
So overall, they seem to perform about the same against Ordos. Based on observing them in combat, though, the Mk1 definitely charged forward more often and closer to the enemy fleet, so it would probably pair better with aggressive ships (Retributions, SO ships, etc.). Whereas the XIV also charged forward but stayed at a relatively longer range (despite officers being set to "Reckless"), so I would expect it to pair better with ships that tended to keep a longer distance. In fact, sometimes the Mk1 charged forward so aggressively that it blocked the line of fire of the XIV, despite me trying to order it back, which probably reduced the XIV's weapon uptime somewhat.
This does make me wonder, should we be taking advantage of the shiny and fancy new simulator more? We can simulate full up 240 DP s-mod and alpha core Remnant fleets if we want (just summon - or farm - a pile Alpha cores in your cargo hold). While that doesn't include fleet skills, the s-mods probably make it a wee bit harder than normal remnant fleets. We can put singular ships up against a couple cruisers or straight up fight an Alpha Radiant with 3 s-mods. We can also let people run the exact same opposing fleet from different save files, and change the ratio of Paladins versus Tachyon lances on various Remnant ships.
We could, but for me, directly doing a campaign battle is the most accurate to what the player would encounter in a campaign, because, well, it really is what the player encounters in a campaign. This includes not just enemy fleet skills but stuff like dealing with objectives and the enemy fleet AI's behavior with respect to that.
Yes ideally people should run against the exact same opposing fleet (and I generally use the exact same test fleet per game version for all my testing, so that the conditions are uniform across all the player fleets I test). In lieu of that though, for testing I take steps to ensure that the test fleet I use closely approximates the "average" enemy fleet. For double Ordos, based on Tranquility's
Fleet Testing mod, the statistics of the "average" double Ordos for 0.98a-RC7 (100000 samples, assuming 320 FP and 100% ship quality), relative to the one I'm using for testing, are below:
Stat Average Test Diff
eff DP 736.38 756 +19.62
Alpha 20.357 21 +0.643
Beta 9.736 11 +1.264
Gamma 9.378 9 -0.378
D-mods 7.996 5 -2.996
Radiant 1.615 2 +0.385
Nova 1.719 2 +0.281
Apex 3.744 4 +0.256
Bril 7.687 7 -0.687
Scin 7.049 7 -0.049
Fulg 11.55 11 -0.55
Glimmer 8.237 7 -1.237
Lumen 8.227 8 -0.227
Flash 3.251 3 -0.251
Lux 4.182 3 -1.182
Spark 6.812 8 +1.188
Term 7.488 8 +0.512
The stats for 0.98a-RC8 were essentially the same. So the double Ordos test fleet that I'm using right now is slightly more difficult than the average, though by just a bit (roughly 3%), but should overall be pretty close the "average" double Ordos fight that players should expect to see.
I do use the simulator to test out different builds though, and the new one is definitely a big improvement over the original. For example, I'll do stuff like, spawn 4 Maws, delete the middle two using Console Commands (so that the remaining 2 are far apart), then test out ship A against one and ship B against the other, to have two one-on-one fights going on side-by-side simultaneously. That makes it easier to get a sense of just how each ship is faring relative to the other.
Regardless of how one chooses to do their testing, as long as they're clear about the conditions that they tested under to generate their results and thus how they came to their opinions or conclusions, I think it's fine.
Although I will admit I'm used to assuming, everything else being roughly equal, a higher hull to shield damage ratio is better. Certainly for something like an Afflictor or Doom, that holds. Here, where the MK I is doing 40% more hull damage, if that was the only stat I had available to look at, I'd assume it was killing about 40% more ships.
Well, sort of. It's more complicated than that. Yes in some sense all that "matters" is hull damage, since killing a ship just depends on its hull and not its shields or armor, but you need to get through the shields and armor first. The Mk 1 definitely has a lot of anti-hull damage, so once enemy shields go down, its DPS is going to go way up, so it will be finishing off enemy ships more quickly than other player ships. It's basically a kill-stealer. (So is any flagship I pilot, since I the player am much better at gauging if it's okay to finish off a target than the AI so I tend to be much more aggressive at finishing off targets.)
However, shields can recover, and armor can be broken in only one place, or the ship can take awhile to die, and turn to present fresh armor. So generally, if I hand you a ship's damage breakdown, and don't indicate how many ships it faced, or how long it took to deal that overall damage, it's really hard to interprete.
Yes, you want to have as many bits of information as possible in order to best interpret what happened. A ship's damage breakdown by itself isn't really enough; it's better to know something more about the numbers. In my numbers above for example, since the ships are in the same battle, you know that they took the same amount of time to do the damages recorded, and they were also side by side.
Let's take a step back, although this will get a bit into the weeds of how battles work from a damage/DPS standpoint and the principles I use for optimal fleet compositions to minimize battle completion time. Back in 0.95.1a, I took the battle results from a bunch of battles fighting the same double Ordos test fleet using different player fleets, and plotted the total shield/armor/hull damage dealt. The result was that total armor and hull damage dealt was pretty much the same across runs (so things like enemy ships turning to face fresh armor, or hull regen from elite CE, etc. were relatively insignificant), but total shield damage dealt increased linearly based on battle completion time. That is to say, there was a linear relationship between the battle completion time and the total shield damage dealt,
regardless of player fleet composition or the specific weapons used, at least within the bounds of my testing. The relationship was roughly 1k more shield damage dealt for every additional second of battle. This can be interpreted as how quickly the enemy fleet regenerates flux during the fight, so the longer the battle takes, the more total flux you have to overcome. However, on a damage per second (DPS) basis, it was always the case that higher DPS meant shorter battle time, even though the total damage dealt would be lower. That's illustrated in the attached graphs.
Under this framework, each ship is just a platform to place weapons that do damage, and then you see how much damage each weapon did; all the maneuvering and shields and so forth basically boil down to increasing how much damage weapons do. To get faster times, the fleet needs to have higher DPS, subject to it matching the ratio of DPS types for a given completion time. So that's what I meant about achieving the "ideal" mix of shield/armor/hull damage dealt (or, DPSshield, DPSarmor, and DPShull, which is equivalent for a given battle). If the fleet has too much anti-shield weapons, for example, then it just means that more of the anti-shield shots end up hitting armor/hull, thus reducing their overall effectiveness, until the overall DPS matches the mix of DPS for the fleet's battle completion time.
Each weapon slot generally had a weapon that had the highest DPS on most ships. In 0.97a, for large missiles it was the Squall, for medium missiles the Harpoon Pod, for large ballistics the Hephaestus (technically the Mjolnir did have higher DPS but costed too much flux on most ships), for medium ballistics the HVD or HAC depending on the ship (for 0.98a it looks like it'll be the HMD, which has outperformed the Heavy Needler, Heavy Autocannon, and HVD in all the side-by-side testing I've done thus far), for small ballistics the Light Needler, and so forth. So naively, you might think, oh, that's easy then, just put the highest DPS weapon on each weapon slot.
The issue is that few if any ships actually match the ideal DPS profile. The Squall/Harpoon Gryphon, for example, is overly anti-armor/hull and not enough anti-shield. The HVD/IRAL Eagle is good at anti-shield and anti-hull but bad at anti-armor. And so forth. So if you're just testing one ship, i.e. spamming a bunch of a single ship, you're basically forcing that ship's weapon profile to try to match a particular DPS profile that it may not be "inherently" suited for based on the provided weapon slots. That's what I found back when I was testing flagship Onslaught + cruiser spam (spamming the same cruiser to fill out the rest of the fleet other than the flagship Onslaught) -- every cruiser (and indeed, every ship) had an "inherent" damage profile that was different than what the fleet needed, so there were inefficiencies with the weapons when trying to minimize battle completion times.
The better fleets are, of course, mixed fleets. In this case you can mix and match different ship's weapon profiles so that the overall amount more closely matches the "ideal" DPS profile. This is one reason why Gryphons + Conquests worked better than all Gryphons or all Conquests. All Gryphons was too much anti-armor/hull, so a lot of the Harpoons were wasted on hitting shields, while all Conquests had a lot more anti-shield but not as much anti-armor/hull punch on a per DP basis. Having several of each ended up being a lot better.
Keeping this framework in mind, the rest is fairly straightforward:
For instance, can you tell me how many ships were engaged by the XIV and how many were engaged by the MKI? If I gave you those MKI and XIV damage numbers and told you the MKI engaged 6 Brilliants and the XIV engaged 5, the numbers mean on thing, and if the MKI engaged 5 Brilliants and the XIV engaged 6, they mean another. The damage stat line could conceivably happen in both.
Not really, I look at full battles so that it's purposely an aggregate of the results against all ships, which gives a clearer understanding of ship performance against whole enemy fleets than testing it against one ship. If one ship was up against 5 Brilliants and another had 6, that would be reflected in their armor/hull damage numbers.
This is why I was trying to tease out the amount of time spent shooting at shields versus armor versus hull. If the MKI happened to spend 10% more time shooting at hull, that might imply shields were only up for 90% of the time they were up for the XIV side, which might imply that side of the battlefield had to deal with 10% total less shields. For whatever reason (fewer ships? ships with different shield to hull ratios? ships cycling in and out so they can recharge more flux?).
Yes it's true that for minimizing battle completion times, what we're really interested in is time to strip shields, time to strip armor, and time to strip hull, but what's actually measured is the damage done to each instead. So we have to infer times based on the battle completion time, and I subtract 45 seconds to account for the time it takes for the two fleets to reach each other at the beginning of each battle. With that, I don't directly know the time spent on each, but I do calculate the damage/hit, damage/shots fired, and damage/flux ratios. I also do a lot of side-by-side comparison testing since this means both faced identical conditions. For battle variability (i.e. a given ship might have just happened to have seen more of one or another), that's why I ended up using double Ordos and not single Ordos, and using multiple ships and/or multiple runs to average that out. I generally force the AI to fire all weapons by linking their weapons to a group with PD, and prevent them from fluxing out by controlling their positioning.
Depends on the testing setup. If it is a single ship being deployed, then I stand by the statement - a higher hull to shield ratio is better (even if the ratios we are comparing are less than 1, specifically 0.3 is better than 0.2, and I recognize against Ordos shield damage is going to be higher than hull unless you're soloing with a Doom or the like). If it is a fleet in aggregate (i.e. the overall fleet did less shield damage), then I think the statement also generally stands as that typically implies a shorter completion time.
If it's a full fleet, yes higher hull to shield ratio is better but that's because the same hull damage was done but less shield damage was needed to reach armor/hull, which meant higher shield DPS, and yes lower completion time. On an individual ship within a fleet basis though, you can't use this to conclude that a ship is "better" than another ship because a ship could be contributing more of its damage toward shields instead. Otherwise, when comparing ship A vs ship B within the same fleet, you lead to degenerate solutions where the "optimal" individual build is for the ships to be all Thumpers or something waiting for somebody else to get rid of shields and armor so they can quickly do the hull damage before the enemy ship explodes, even though that messes up the fleet's overall battle completion time.
In this particular case, the Mk1 has more anti-hull (due to Heavy Adjudicator instead of TPC), but the XIV has more anti-shields. So in combat, the XIV will naturally contribute more shield damage, while the Mk1 will naturally contribute more hull damage, and you need both types to defeat enemy fleets. So you can't use it as a basis to say one ship is better than another.
Although, having run the fights, do you get a feel for how much dead time is spent between waves, and is it equal between sides? I mean, given the variation in damage you mentioned, there's got to be wide variation in what engages the XIV and MKI ships, and presumably it varies a lot within and between fights. Is a DPS comparison valid for individual ships if one ship is engaged by capitals while the other is engaged by destroyers? Although, given your Doom statistics, perhaps they are just engaging destroyers and a few cruisers each?
I don't really know how much dead time is between waves. The closest estimate is
here, where I tested TPC with or without s-modded Expanded Magazines, by having the standard TPC alongside one with 1.5 faster ammo regen (the ship had un-s-modded Expanded Magazines so both had the ammo amount bonus). The result was that the s-modded Expanded Magazines TPC did around 25% more overall damage, even though on paper it should've done 50% more overall damage. So there was clearly some amount of time where the buffed TPC was just waiting at full ammo.
Having said that, *some* amount of dead time is good, because it enables the AI to vent if it's high on flux, but any time spent at flux = 0 means flux is being "wasted" on not hitting the enemy, so I usually try to push the AI forward as much as possible. Also, the flanks tend to get less ships, and the ships they get tend to be weaker. This is why I switched over to testing ships with Gryphons on the flanks, with the ship being tested in the middle, for 0.96a and 0.97a (LP Brawlers were too inconsistent due to the "back off on Eliminate order" bug). The Gryphons on the sides basically absorbed that dead time so that I could measure how the ship being tested fared against the bulk of the enemy fleet, without worrying about the testing being thrown off from one of the ships being tested having to run off to chase down a frigate or having dead time due to that. For 0.98a I'm hoping to move back to using LP Brawlers on the sides for this, to free up more DP to use on the ship being tested.
Even when I'm using the Doom, the other ships do still get their amount of capitals, simply because I mess up quite often (i.e. try to kill a Radiant, fail, have to back off to vent, so the other ships have to go in and kill it while I'm backing off and venting), or because the Doom isn't as good at killing the Nova, and generally just because I can only be in one place at a time.
Also, you can see that even though this dataset is using a very different testing setup than the previous one, the results were still broadly similar. The previous formation was:
LP Brawler - Gryphon - Onslaught (XIV or Mk1) - Gryphon - Onslaught (the other one) - Gryphon - LP Brawler
with the flagship Doom somewhere out in front, while this one is:
1 or 2 LP Brawlers - Mk1 - XIV - Mk1 - XIV - 1 or 2 LP Brawlers
with me in one of the 3 LP Brawlers. Yet the results still ended up being that the XIV did more shield damage while the Mk1 did more hull damage, in roughly similar proportion, with their relative totals being pretty similar. This is despite the Onslaughts doing less than 1/3 of the player fleet's overall damage in the previous setup while doing the overwhelming majority of the damage in the current testing setup, and despite the Onslaughts using somewhat different weapons this time around. So basically, the other ships in the fleet didn't really have a large effect on the comparison. Of course, the previous setup is one that is actually achievable in the campaign whereas this fleet is not, but is just for testing.
Interesting. I think you are saying the 1 sigma standard deviation of shield damage dealt by the Onslaught XIV is something like 20,000 10,000 or larger? With similar values for the MK I? Yeah, in that case I don't think we can say anything at the 25% level, let alone the 10% level.
Yes more or less. Some of the numbers I wouldn't read too much into because this is averaging across a range of numbers, with some of the variations quite large. So I'm looking more at the overall gist.
Damage value might be small for the kinetics, but the number of shots isn't.
Oh heh that's true. I think I usually consider it more as what is the overall damage/flux ratio that I get out of each weapon. That's in some sense not completely accurate (since I'm just adding up total = shield + armor + hull, and each point of shield is not "worth" the same as each point of armor, etc., in terms of how easy it is to inflict) but it's good enough for me, especially when I'm comparing weapons of the same damage type.
I haven't really tried to mine the DCR data to figure out how much each weapon was hitting shields vs armor vs hull. I think shields would be the easiest since there's no hit strength to worry about. I generally assume that all weapons are firing, so they should all be seeing about the same ratio of shield/armor/hull as each other, with some caveats. If so, then the same proportion of hits should be going toward shields. In that case, it's a matter of figuring out how much damage is done per shot to shields (taking care to account for TPC not getting the Ballistic Mastery bonus, so it gets a *(1+0.05+0.10+0.07)/(1+0.05+0.10+0.07+0.15) modifier, accounting for Tactical Drills, CR, Cybernetic Augmentation, and Ballistic Mastery, respectively), and then looking at the ratio of shield damage to this hypothetical amount of damage. Doing this with that data, I get:
weapon total shield armor hull hits damEach damtype damVsSh * hits sh/dam
har pod 22206 4809 5482 11914 39
TPC 21116 13641 3638 3836 154 250 energy 222.63 34285 0.398
heph 18612 5160 6868 6583 254 120 HE 60 15240 0.339
he need 13514 12133 359 1022 317 50 kinetic 100 31700 0.383
li need 11170 9667 492 1012 308 50 kinetic 100 30800 0.314
HAC 9813 8475 284 1054 111 100 kinetic 200 22200 0.382
LAC 3757 3383 112 262 87 50 kinetic 100 8700 0.389
So most of the weapons had a shield / theoretical damage to shield ratio of around 0.39, except for the Heph which was at 0.34 and Light Needler which was at 0.31. So those two seemed to hit armor/hull more often than the others for whatever reason, all the more puzzling because the Light Needler and Heavy Needler hit around the same number of times, but the Heavy Needler consistently did more shield damage per hit, even though both have base 50 kinetic damage per projectile. (This implies the Heavy Needler hit shields more, while the Light Needler hit armor/hull more.) This was true for all of the samples. Don't have a good explanation as to why. Guess I'll have to see if this trend appears in the other tests.
(One might ask what's the significance of the 0.39. Well, assuming that the Ordos average shield efficiency is 0.5, then that means you get 0.39/0.5 = 0.78. Then if other damage bonuses amount to say 20%, then that's 0.78/1.2 = 0.65. Then this means that 65% of the hits were to shields, and 35% of the hits were to armor or hull, on average.)
When I find another Alpha Core I'll try out your advice, but I'm not going to expect too much. Most of my fleet is built around being annoying at a distance. So having a capital ship that loves to give the enemy fleet a hug probably won't ever work out well.
Yeah if the rest of your fleet is meant to be used at a distance, the regular Onslaught may be better. It seems like the Mk1 is more for if the rest of the fleet can properly support a ship that dives head first into the enemy fleet.
Also a 230 second time when the fleet went for the right most objective instead of approaching head on
Yes for testing under standardized conditions purposes I usually just take the battles where the enemy fleet goes straight down. If it goes to the right I just reset and start over. I used to take them all and just note which direction the enemy fleet went, but it got too cumbersome to keep track.
Realistically the better fleets I test, this doesn't matter much (they can surround the enemy fleet just the same, just that I have to move them diagonally forward), this basically affects (improves) the worse fleets more since I don't have to sit through tests where they're unable to contain the enemy fleet and end up taking a long time.
In 0.98a, it seems like a stealth change is that the enemy admiral AI now makes additional decisions on where to send the enemy fleet. For the diamond objectives, it used to always send the fleet straight down or to the right. (This was recognizable by where it sends its initial frigates. One to each side -> main fleet will go straight down. One to the left and one to the bottom -> main fleet will go to the right.) In 0.98a, though, there's at least one additional case where it'll send half its fleet to the left and half its fleet to the right, and possibly other cases as well. This increases variety for battles but makes it harder to test under the same conditions.
During all my testing, I had not lost a MKI to any of these small Ordo fleets, while I've lost numerous XIV (both shield shunt and without).
I think it depends on how you're controlling them. Generally if a ship dies frequently I consider it that I need to improve the build and/or figure out how to control it better. In this case, I actively control them (via Rally, Rally Civilian Craft, and Eliminate) to move them forward or to stay put (and have the rest of fleet moves forward) as needed based on how much enemy pressure they're under and whether or not they're fluxing out. I think in testing I've actually had the Mk1 die more often because it'll continue going past its commanded "Rally Civilian Craft" point into the enemy fleet, thereby getting too much enemy pressure and dying, whereas I have to work a lot harder to get the XIV to keep going forward.
Author
Topic: [!!SPOILERS!!] New academy quest line ship [!!SPOILERS!!] (Read 12474 times)