Fractal Softworks Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

Starsector 0.97a is out! (02/02/24); New blog post: Simulator Enhancements (03/13/24)

Pages: [1] 2

Author Topic: 0.9.1a weapon testing - missile launchers on Shrike  (Read 2860 times)

pairedeciseaux

  • Captain
  • ****
  • Posts: 340
    • View Profile
0.9.1a weapon testing - missile launchers on Shrike
« on: September 05, 2020, 02:02:46 AM »

This is a series of synthetic tests similar to those I’ve run a few weeks ago. Here we are looking at some medium missile launchers on Shrike. The idea was: see which weapon would perform best on Shrike (I usually put either Sabots, Reapers, or none), and see whether or not Salamanders would provide benefits.

TL;DR when in doubt use Sabots on Shrike

Starsector 0.9.1-RC8, Fleet Tester 1.0 mod, 6 concurrent game instances, battle duration measured using the in-game timer, AI vs AI (no human player). Also this time I didn’t “cheat” using aggressive personality to shorten test runs and emphasise weapon performance. Also, I endend-up having to consider the various battle outcomes in addition to battle duration as in some cases outcome was far from a 100% win.

So our test platform is a regular Shrike with
1 Heavy Blaster
1 medium missile launcher
2 IR Pulse Lasers
1 Ion Cannon
2 PD Lasers
20 vents
14 capacitors
steady personality

Medium missile tested in this configuration are:
  • Harpoon MRM Pod
  • Sabot SRM Pod
  • Annihilator Rocket Pod
  • Typhoon Reaper Launcher
  • Salamander MRM Pod

Also tested:
  • Pilum LRM Launcher, with 17 capacitors
  • No missile, with Flux Distributor and 16 capacitors

Part 1 - one Shrike against one Wolf
Spoiler
This is the Wolf Attack variant with steady personality. For each matchup I ran 24 repetitions.

Well, actually, I ran a few more repetitions in order to obtain exactly 24 samples for each matchup. That is, I got some Wolf retreat: 0 for Harpoon and Pilum, 1 for Reaper and no missile, 2 for Sabot and Annihilator, 4 for Salamander.



So those retreats were otherwise ignored.

From best to worst, the test results are:
(1) Sabot: total is 2683 seconds; median is 90
(2) Annihilator: total is 4297; median is 137.5
(3) Harpoon (5227; 223) ,no missile (5719; 248) and Reaper (6307; 280.5)
(4) Salamander (6899; 310) and Pilum (7107; 320.5)


[close]

Part 2 - one Shrike against one Tarsus
Spoiler
This is the Tarsus Standard variant with steady personality. For each matchup I ran exactly 24 repetitions. There was no retreat.

From best to worst, the test results are:
(1) Sabot: total is 993 seconds; median is 39.5
(2) Harpoon: total is 1102; median is 44.5
(3) Reaper (1152; 47) and Annihilator (1166; 47.5)
(4) Salamander (1346; 55.5), Pilum (1381; 55) and no missile (1338; 56)


[close]

Part 3 - one Shrike against one Prometheus
Spoiler
This is the Prometheus Super variant with steady personality. For each matchup I ran exactly 24 repetitions. This is the point where I changed methodology to account for various battle outcomes.

(here I think I messed-up my Pilum variant, it had 14 capacitors instead of 17)



From best to worst, the test results are:
(1) Sabot: 24 win; median is 136.5
(2) Reaper: 18 win; median is 219
(3) Harpoon: 8 win; median is 228
(4) Pilum: 7 win; median is 255
(5) No missile: (7; 344), Salamander (7; 430) and Annihilator (2; 395)

(median time is based on won battles only)


[close]

Part 4 - two Shrikes against one Prometheus
Spoiler
Same as part 3, but with an additional Shrike. For each matchup I ran exactly 24 repetitions. There was no retreat.

From best to worst, the test results are:
(1) Reaper: total is 1221; median is 50
(2) Annihilator: total is 1389; median is 57
(3) Harpoon (1505; 61)
(4) Sabot (1742; 71.5)
(5) No missile (1910; 79)
(6) Pilum (2056; 86)
(7) Salamander (2261; 92.5)


[close]

Summary

Both the one Shrike against one Wolf test and the one Shrike against one Prometheus test proved to be challenging for AI.

In all tests except the two Shrikes against one Prometheus case, Sabot provided best performance. In that one case, it was still better than no missile.

In all tests, Salamander and Pilum had similar or worse performance than no missile.

In all tests except the one Shrike against one Prometheus case, Annihilator provided various degrees of improved performance compared to no missile.

In all tests, Harpoon provided various degrees of improved performance compared to no missile.

In all tests except the one Shrike against one Wolf case, Reaper provided various degrees of improved performance compared to no missile. It was especially effective in the two Shrikes against one Prometheus case.

The no missile load-out used here was always worse than having Sabot or Harpoon.

Moving from 1 Shrike to 2 Shrikes brings dramatic changes. Numerical superiority is good.

What do you think? Does this match your in-battle experiences?
Logged

Serenitis

  • Admiral
  • *****
  • Posts: 1458
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #1 on: September 05, 2020, 02:43:31 AM »

P. much a dead match for my experience.

Shrike needs to be in close to do its work effectively, so sabots are my 1st choice if available. Otherwise harpoon, annihilator or reaper depending on availability / how big a target I'm expecting them to face.

Salamanders don't fit Shrike very well as it wants DPS to work in a skirmish role. And it will be too close to effectively take advantage of the AIs engine paranoia.
Beam Shrike can use salamanders reasonably well. But that's a pure support build and doesn't do well on its own at all.
Pilum is not a good match, due to being a slow artillery-like weapon. It's a waste of OP putting those on front line skirmishers.
Logged

Grievous69

  • Admiral
  • *****
  • Posts: 2975
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #2 on: September 05, 2020, 05:39:45 AM »

Smh when Thaago constantly kept saying how Harpoons are equally as good as Sabots on Shrikes. The numbers don't lie.

One slight complaint I have is, who the hell uses IR Pulse lasers when you already have a Heavy blaster. It would make sense on a Pulse laser build (which goes better with HE missiles), but here only Ion cannons and PD are enough. Putting even more flux on the poor thing never goes well in actual battles.
Logged
Please don't take me too seriously.

Modo44

  • Commander
  • ***
  • Posts: 144
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #3 on: September 05, 2020, 05:54:10 AM »

Smh when Thaago constantly kept saying how Harpoons are equally as good as Sabots on Shrikes. The numbers don't lie.
The numbers change depending on fleet composition. This is a common fallacy of only 1v1 testing your ships.
Logged

Grievous69

  • Admiral
  • *****
  • Posts: 2975
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #4 on: September 05, 2020, 06:18:02 AM »

Well yeah obviously, but I still think the results wouldn't change too much with an actual skirmish battle (using the same builds). Now it might make a difference in a giant battle with capitals, but at that point you're not using Shrikes anyway.
Logged
Please don't take me too seriously.

pairedeciseaux

  • Captain
  • ****
  • Posts: 340
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #5 on: September 05, 2020, 06:24:55 AM »

Indeed, those tests can't replicate all real fleet battle situations, so one should never put too much faith in numbers. If you look at the results I got in the 2 Shrikes against 1 Prometheus test, Harpoon performed better than Sabot by a good margin.

In my experience AI is good at shooting volleys of Harpoons towards high flux and overloaded targets, and quickly destroy those targets. Sabots can't' do that, or at least certainly not to the same degree. Sometimes Harpoons are wasted, but it's true for all missiles.
Logged

Grievous69

  • Admiral
  • *****
  • Posts: 2975
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #6 on: September 05, 2020, 06:46:38 AM »

Which is why I'm not that crazy with tests where the testing side has greater numbers. Some weapons are just ''win more'' or a ''win faster'' button, with Harpoons being the best example of that. In campaign, most of the time you're either fighting with equal opponents, or get ambushed by a bigger fleet. You could say this is the case for every HE missile, but the thing with Harpoons is that every single ships that has them, WILL fire them at anything that's overloaded. So it's no wonder it has better time stats in skewed scenarios. Actually I'd say Harpoons are the closest missile to fighters, where increasing numbers have exponential growth in power, they're just a limited but more bursty version.

Pilums used to be literally current fighters, but they're kinda useless now. Only thing they do is confuse AI.
Logged
Please don't take me too seriously.

Megas

  • Admiral
  • *****
  • Posts: 12117
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #7 on: September 05, 2020, 08:50:10 AM »

And when player has an advantage, the enemy usually runs, which means autoresolve.  Autoresolve is generally better than actual fighting because it is much faster, no risk of casualties, and player can use trash ships that have guns (like the Colossus 3 with a bunch of light weapons placed on mounts as free storage or recently recovered clunkers with a bunch of looted weapons thrown on them) and kill most of the enemy fleet.
Logged

Thaago

  • Global Moderator
  • Admiral
  • *****
  • Posts: 7173
  • Harpoon Affectionado
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #8 on: September 05, 2020, 11:32:00 AM »

Smh when Thaago constantly kept saying how Harpoons are equally as good as Sabots on Shrikes. The numbers don't lie.

...

I'm also constantly saying that the Sabots are the duelist build while the Harpoons are for fleet situations, and that all my results are from large fleet battles where I compare the post battle damage report. These results are time trials of duels and line up well with my experiences. Numbers don't lie, but they also only give data on what they are based on. Unfortunately, misinterpreting limited data and then claiming that the case is closed is a time honored tradition in fooling people.

These tests are a fantastic set of data! Harpoon vs Sabot lines up well with my experiences, but I'm a bit surprised that Pilum and Salamander do so poorly. The 2v1 hint at the situation with the sabot vs other missiles: sabots are great at producing opportunities for a kill, which is super important for a ship that struggles to break shields like the Shrike. However when there are other opportunities for a kill, IE the 2v1, then the kill missiles shine. I'm also surprised that in this scenario the Annihilators do so well, but I suppose Prometheuses have narrow shields and are a big target, and Anni DPS is pretty good!
Logged

intrinsic_parity

  • Admiral
  • *****
  • Posts: 3071
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #9 on: September 05, 2020, 02:17:07 PM »

It feels like there's a lot of underlying variance with such a small sample size. That's not necessarily a criticism of the test but more a comment on the complexity of the game. There's so many other variables that dictate the outcome of these combats (AI decisions and stuff), and it's hard to know how much of the results are actually due to the choice of missile, especially for results that only difference by 1-2 (shrike vs wolf results don't really say anything IMO).

That being said, salamanders and pilums clearly just don't solve the problems that the shrike has. Either that or they just get countered so hard by decent PD that they are useless without flares or something to distract the PD. I'm also wondering if the 1x shrike with annihilator pod got shut down by PD, but two shrikes with pods got through and were able to win easily because of that. 1x annihilator sucked super hard and 2x was amazing. There could be some variance involved there as well.
Logged

Megas

  • Admiral
  • *****
  • Posts: 12117
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #10 on: September 05, 2020, 03:53:19 PM »

I use Salamanders on Shrike (P) if I do not have Sabot Pods (and I usually do not have Sabot Pods early since they are military and not found on any bundle pack), since they can brawl like a poor-man's Medusa, and Salamanders are useful when brawling smaller stuff in a duel.  Salamanders are more useful early in the game.

Also, small Salamanders cost 5 OP, compared to 10 on the useful medium missiles like the pods.
Logged

RustyCabbage

  • Captain
  • ****
  • Posts: 347
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #11 on: September 05, 2020, 08:07:45 PM »

Nice work. The results are pretty unsurprising, I think, but it's good to see confirmation.

intrinsic_parity

  • Admiral
  • *****
  • Posts: 3071
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #12 on: September 05, 2020, 10:20:50 PM »

I use Salamanders on Shrike (P) if I do not have Sabot Pods (and I usually do not have Sabot Pods early since they are military and not found on any bundle pack), since they can brawl like a poor-man's Medusa, and Salamanders are useful when brawling smaller stuff in a duel.  Salamanders are more useful early in the game.

Also, small Salamanders cost 5 OP, compared to 10 on the useful medium missiles like the pods.

Ironically, the performance against the wolf was very poor for the salamanders, the wolf got away more often than any other choice, and the time to kill was worse on average than having no missile. This could be pure variance, or it could be some AI artifact. Maybe the wolf plays a lot more passively to avoid getting hit by the salamander, and thus is harder to kill and more able to disengage and retreat. Regardless, this testing doesn't support the idea that salamanders should be used on the shrike in pretty much any scenario, at least under AI control in situations with small numbers of ships. The performance with no missile appeared to be the same or better in every test.

Also I'm pretty sure the testing used the 10 OP salamander pod which fires twice as fast as the 5 OP small missile.
Logged

SCC

  • Admiral
  • *****
  • Posts: 4112
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #13 on: September 05, 2020, 11:40:44 PM »

10 OP Salamander and 7 OP one both fire every 25 seconds, but the big one fires two salamanders at a time.

pairedeciseaux

  • Captain
  • ****
  • Posts: 340
    • View Profile
Re: 0.9.1a weapon testing - missile launchers on Shrike
« Reply #14 on: September 06, 2020, 01:04:47 AM »

The 2v1 hint at the situation with the sabot vs other missiles: sabots are great at producing opportunities for a kill, which is super important for a ship that struggles to break shields like the Shrike. However when there are other opportunities for a kill, IE the 2v1, then the kill missiles shine.

Right.  :)  And that's precisely and good use-case for Shrike or a larger ship such as Gryphon, in my experience.

It feels like there's a lot of underlying variance with such a small sample size. That's not necessarily a criticism of the test but more a comment on the complexity of the game. There's so many other variables that dictate the outcome of these combats (AI decisions and stuff), and it's hard to know how much of the results are actually due to the choice of missile, especially for results that only difference by 1-2 (shrike vs wolf results don't really say anything IMO).

All those tests had a fairly high dispersion. I often had the impression to measure random noise  ;D but looking at the final time distributions, it does make sense (to me at least). So I think I got enough samples for those simple tests.

More information on the Shrike vs Wolf test:
Sabots: minimum is 26 (lowest of all), maximum is 401, mean is 111.8, median is 90, standard deviation is 81.1.
Annihilators: minimum is 61, maximum is 439, mean is 179, median is 137.5, standard deviation is 119.2.
Salamanders: minimum is 112, maximum is 397, mean is 287.5, median is 310 (ouch!), standard deviation is 72.8 (lowest of all).

What did you mean when you wrote "especially for results that only difference by 1-2"?

I'm also wondering if the 1x shrike with annihilator pod got shut down by PD, but two shrikes with pods got through and were able to win easily because of that. 1x annihilator sucked super hard and 2x was amazing. There could be some variance involved there as well.

See Thaago's comment above.  :)  With numerical superiority, ships gets more opportunities to strike.

Also worth mentioning: the Prometheus has 2 Railguns and 1 Light Needler. So kinetic damage and a higher range than Shrike. So a nightmare for AI Shrike, and probably an unpleasant experience for a human player.

Also in 1v1 situation, if target is stronger Annihilators are wasted against enemy shield.

---

Just to clarify: in these test only medium sized missile launchers were used (or no missile at all). Though I would still consider a load-out using a small-sized missile launcher as valid as long as missiles are useful. I won't run tests using those options as it's a lot of work.

Pilum not being effective (or even relevant) on Shrike, I agree.  :)

I use Salamanders on Shrike (P) if I do not have Sabot Pods (and I usually do not have Sabot Pods early since they are military and not found on any bundle pack), since they can brawl like a poor-man's Medusa, and Salamanders are useful when brawling smaller stuff in a duel.

Do you use Salamanders on AI Shrike(P) and/or player-piloted Shrike(P)? What benefits do Salamanders provide in these contexts? Are you using the trick mentioned by Serenitis here:

Salamanders don't fit Shrike very well as it wants DPS to work in a skirmish role. And it will be too close to effectively take advantage of the AIs engine paranoia.

?

Ironically, the performance against the wolf was very poor for the salamanders, the wolf got away more often than any other choice, and the time to kill was worse on average than having no missile.

I am disappointed by this result, but not entirely surprised. The naive expectation is: Salamander would catch Wolf and make Shrike's job easier. In practice Wolf would disengage when Salamanders are chasing it, and dodge those without too much effort. So rather than shortening battle duration, Salamanders are make battle duration longer.

I wonder whether using other frigates such as Lasher would gave the same results or not, but, again, that's a lot of work. Also I didn't want to search too hard for a test scenario that would make Salamander shine. So maybe Wolf is precisely the edge case where Salamanders on AI ship underperform... or maybe Salamanders are just not effective against agile targets. This would require a test campaign dedicated to Salamanders to narrow it down.

Though I have experienced real in-game battles were Salamander would catch ships of any sizes on the frontline and thus put those ship in danger. And Salamanders are also able to distract several enemy ships to some degrees, which can lead to more opportunities. So I know Salamanders can be effective. But are they 10 OP worth on a 80 OP ship? I do not have a definitive answer yet, but it is unlikely.
Logged
Pages: [1] 2