Developers discuss the future of the RTS genre, diving into upcoming game Stormgate and the SnowPlay technology that makes it all possible.
Upcoming real-time strategy game Stormgate is a project from the recently formed Frost Giant Studios that aims to represent the future of the genre. Frost Giant Studios has many Blizzard RTS veterans at its helm working on the project who've worked on titles like Starcraft and Warcraft who are now setting out to tell a brand-new story. Stormgate will utilize SnowPlay technology, which is a proprietary RTS tool from the mind of James Anhalt that will make Stormgate even more responsive than past real-time titles.
The new project from the former Starcraft and Warcraft developers is a free to play title set many years in the future on Earth during a conflict between different factions. Two have been revealed for Stormgate so far: the Human Resistance and the Infernal Host, with more set to come in the future. While fighting in these sci-fi conflicts, building bases and armies, and harvesting resources, SnowPlay technology will help both players and spectators have a better, smoother experience thanks to more efficient technology.
Chief Architect James Anhalt participated in an email interview with Screen Rant to elaborate on the technical details of SnowPlay, and Frost Giant CEO and Production Director Tim Morten sat down with Screen Rant to discuss Stormgate's narrative, progress, and what players can expect in the future.
Screen Rant: You've said that Stormgate will be the most responsive RTS game you've ever worked on. Technically speaking, what helps make it the most responsive?
James Anhalt: Traditionally, games (and RTS games in particular) only run game logic at a rate much lower than the rendering rate. For example, StarCraft II originally ran the game logic (one game "loop" or "step") only 16 times per second (16Hz). When it shipped, we upped that to 22.4 Hz as the default rate. Some games run game logic as low as 8Hz and some games run parts of the game logic even less often.
Originally, SC2 ran all the Fog of War logic at only 4Hz! While this was in line with other games, I wasn't satisfied with that approach, so I rethought the problem and optimized it to be over 30 times faster. Now it runs at the same rate as everything else, provides a smoother game play experience, and takes less time doing it.
In Stormgate, we've upped the game logic to run at 64Hz, including Fog of War! This runs completely independent of the visuals provided by Unreal Engine 5, which can display the game as fast as the CPU/GPU allows. So instead of running player actions at ~44ms (44 milliseconds) intervals, the game responds to player actions at ~16ms intervals, potentially 28ms sooner than before.
How would you describe what rollback technology means? What makes it integral to more responsive RTS gaming?
James Anhalt: We are a deterministic game that only sends player actions to a central game server that then forwards those actions (and when to act on them) to each player in the game.
Traditionally, each player’s game needs to wait for these actions to be received to continue running game logic. This is sometimes referred to as "lock-step," where each player moves through the game at the same time. If this waiting didn't happen, then the games would get "out of sync" with each other and each player.
Our belief is that we can make our RTS more responsive and fun by implementing rollback. Please note that this is a major endeavor that we’re actively working on. We plan to implement rollback in Stormgate once we have tested it thoroughly and made sure that it improves the experience for our players.
So, how does it work? Rollback netcode, which made a major impact in the fighting game community, means the game doesn't need to wait if a network hiccup momentarily delays players’ actions from reaching the server. It also means if nothing "new" is happening in the game (no players are sending actions), we don't have to send "no actions on game loop 42" to the players because they won't be waiting for it anyhow.
How often do players send actions? RTS players refer to actions per minute or APM. It can range from 50 to around 350 for professional SC2 players, which translates into less than 6 per second, or 1 every 10 game logic steps for Stormgate. By increasing the game loop rate the game needs to handle fewer actions, which reduces the chances to experience network jitter.
When an action does "come in late," the game “rolls back” to the time it was supposed to respond to the action, then fast forwards to the current time without causing a hiccup to the player (see below for why we have the "extra" time to do this.) In traditional RTS games, these hiccups are so jarring that extra latency is built in to cover them up, so add another 1 or 2 game loops (at 22.4Hz) or 88ms. In Stormgate we can run closer to "the edge" (and even speculatively into the future) with no extra built-in latency, resulting in the game feeling more responsive.
You have a lot of experience working to make RTS games a smooth experience for players. How would you say what you've learned in the past has factored into your work on SnowPlay?
James Anhalt: The primary learnings are around determinism, optimization, and memory.
Starting with determinism–conceptually, it’s simple. Executing the same actions in the same order on each machine should result in the same outcome. In practice, you run up against undefined compiler behavior, compiler bugs, hardware bugs, and if you are using anything not made with determinism in mind you can't trust that it is actually deterministic.
With each RTS I've worked on, we've built better systems in a way that make it easier to maintain determinism and easier to detect when determinism fails and where. In Stormgate, the deterministic game simulation is completely independent and can be sandboxed and protected from the non-deterministic visualization.
It's very hard to write non-deterministic code in SnowPlay compared to previous projects. To date, our only "desyncs" happened because of bugs in the desync detection code and a memory corruption we weren't detecting. We had to add special debug cheats to cause desyncs on purpose to make sure it was actually working (and stayed working) correctly!
Then there’s optimization. What you optimize for is important. In an RTS game with large numbers of units, it is important that the logic "scales" well. A lot of problems in games can initially look like what we call n-squared problems, where if the number of units doubles, the time quadruples instead of doubles. This is bad. We avoid this by looking for solutions that are linear or even sub-linear.
There is also a trade off between fixed costs and variable costs. In SC2, I feel like I over-optimized for variable cost at the expense of fixed costs, or optimized for the worst case. For example, the fog of war is very cheap per unit, but to achieve that the fixed costs are higher than needed. So over the course of the game we spend more relative time doing fog of war at the beginning of the game when there are few units compared to later when there are large armies.
In Stormgate, I've taken a more balanced approach and try to target the usual case. So with fog of war, the time it takes is also related to how far spread out your units are with almost no fixed cost, not just how many units there are plus a big fixed cost as in SC2. This lets us run game logic faster on average which makes it easier to support rollback and fast forward in replays at higher speeds. I’m also happy to share that the per unit cost is about half of what is in SC2–with twice the sight range!
Finally, there’s memory. This goes hand-in-hand with optimization, but modern CPU speeds have outpaced memory speed increases, so “calculate more, use less memory” is the rule of thumb. One of the benefits of doing things from scratch at Frost Giant is being able to take a high-level view of how things fit together. We can take advantage of redundant data and shrink memory usage from a holistic view instead of micro-optimizing the memory usage of specific structures.
Our pathing mesh, for example, is yet another example of a system that uses less memory than SC2, but is capable of supporting maps with twice the complexity and takes over for multiple other systems like placement and large map-sized grids, which SC2 relies on for speed in some operations. Since we need to locate each unit in the pathing mesh, it is better to store the ground height there than in a separate grid. If you were to benchmark the grid vs. mesh solution in isolation, you would find the grid solution is faster by far, but if you take into account you have to locate the unit in the mesh anyhow, the mesh height lookup is as fast as or even faster than in the grid case, plus we've now used much less memory . . . which helps everything else!
I’ll add something we’re very excited about, both for our developers and for our future user-generated content creators. SnowPlay is going to empower everyone to create and iterate on their in-game creations very quickly. Time is always the biggest limiting factor, and the more time and polish you can pour into a game, the more fun it’s going to be. I like to say it doesn't let us make a game any faster, it just lets us do more polish with the time we have. Determinism, optimization, and memory also play into increasing the rate at which we can iterate. Replays and rewind are great tools for debugging and polishing gameplay, too. The faster those are, the less time wasted and the more time goes into actual polish–those crucial tweaks and changes that make the game more fun.
How will SnowPlay allow spectators of games to have a better experience? How does SnowPlay help players enjoy consistently good performance from the game even when playing on computers with lower specs?
James Anhalt: First, we can talk about performance. SnowPlay runs the game logic completely independent of the visual frame rate. The game logic is highly optimized to use as little CPU and memory as possible. In fact, even though we run the game logic about 3 times as often as in StarCraft II, it likely runs in less total CPU time. For example, 4 game logic updates in SC2 might take 10ms each, or 40 ms total out of 176ms–or 22% of a CPU core running game logic. Whereas in Stormgate, we run 12 game logic steps during that same 176ms time interval, but each one may only be 2ms, or 24ms total, resulting in about 14% less CPU usage overall. With today's multi-core machines, the visuals and the game logic should hardly impact each other compared to previous RTS games.
Lower-spec machines should benefit, too. The networking model for games like SC2 is often referred to as "lock-step," with the implication that everyone's game moves in lock-step with everyone else. So a "slow" computer in the game could cause a bad experience on other computers in the game, or conversely "fast" computers might make problems on "slow" computers even worse.
With SnowPlay, each computer isn't in lock-step with the others, rather they each do the best they can and run the game with as little latency as possible given their CPU, GPU, and network conditions. Each machine and user should thus squeeze more performance out of their respective setup. (I've tried to get our team (and myself) away from calling our networking model lock-step, but old habits are hard to break, so please forgive us if we slip up from time to time.)
This same tech allows mass spectating of a game without affecting those playing in the game. This is another feature we are actively working towards that we plan to introduce post-launch. The spectated game could be paused for spectators, rewound, resumed, and played at faster-than-real-time until it catches up with the live game (or a preset delay in online tournaments to prevent snooping or stream sniping) while the game continues being played.
Each viewer could also control the viewing experience for themselves or follow along with someone else "casting" the game. This even works in reverse . . . if the live game is paused for a short time, spectators may never even notice if the pause was shorter than their current delay!
All of this works together to create an environment that provides players with consistently good responsiveness when compared with other RTS games, regardless of their hardware.
Is there anything else you want players to know about SnowPlay or what it means for the future of RTS gaming?
James Anhalt: Above all else, we want players to have fun playing Stormgate. The more responsive the game is, the more interactive and immersive the game becomes. The more we can iterate, the more we can polish the game. I see SnowPlay’s primary purpose as letting the player get the most fun possible out of Stormgate. As someone who’s been working on RTS games for a long time, it’s been fun to start fresh from a clean slate again (and yes, a lot of work). I am hopeful that SnowPlay will provide the headroom to create fun and innovative game experiences for (and by) our players for years to come.
Can you talk a little bit about the evolution of Stormgate? What core thesis did you have for it at the beginning, and how did it grow from there?
Tim Morten: I'd say the thrust of how we approached the game from the beginning was to try to take what we learned from working on the RTSes that we've worked on in the past, and apply it to make a new RTS that could reach an even bigger audience. Some of the things that really stood out to us as opportunities were primarily focused around social gameplay. Historically, real-time strategy has been such a solitary experience. If you're playing 1v1 competitive, you're alone against the competitor. Solo campaign is also tremendously popular, but again, tremendously solitary.
For Stormgate, we took a cue from the success of the cooperative mode that got added to the last expansion of StarCraft II: Legacy of the Void, which had unexpectedly become the most popular mode in the game. We took that as a signal that players really enjoy playing with their friends. It gives them a reason to come back; it gives them a reason to stay. We thought about whether there was an opportunity for us to apply social capabilities to all the modes of the game, not just that explicitly co-op against AI mode.
Stormgate will have cooperative campaign support and community building, even for 1v1 players directly within the game, but it will also support 3v3 competitive. And then we're increasing the number of players for the open-ended co-op mode to three players as well. That theme kind of carried throughout each of the game modes and really became central for us.
In many ways, Stormgate represents the future of what RTS games can be. What does that future look like to you, and what's most important to keep RTS games popular and fun going far into the future?
Tim Morten: For popularity, we're trying to also take a cue from Elden Ring. I think one of the things that has limited the popularity of RTS games in the past is the intimidation factor. There is a very different perspective and interface than players coming from other genres are used to, and particularly the first time or the first few times that you play an RTS, getting used to those differences can feel very intimidating. There's a lot of interface, there's a lot of units, and you're seeing so much more of the world than you typically see in other genre.
We're trying to provide an on ramp for new players coming into the game that doesn't hit them with all of that at once, but gives them a chance to acclimate and feel comfortable in circumstances where they're up against an AI opponent or in a match with support from others. That way, they're not just feeling the weight of the world on their shoulders and can take a moment to wrap their head around RTS.
I listened to a podcast where you were talking about how even the controls are so different. Just moving the little center roller bringing you to blackness is very disorienting and confusing for new players.
Tim Morten: Yeah, it is a common thing that people scroll off into the fog of war where they can't see anything. And because they're not necessarily used to dealing with the mini-map, they don't realize that there's an easy way to get back to where the action is. That's just one of those many differences in an RTS game that people aren't used to.
I want to talk about Stormgate's lore a little bit. Can you tell me a little bit more about the game’s two main factions that have been revealed so far, the Human Resistance and the Infernal Host?
Tim Morten: Absolutely. The Human Resistance are a band of humans in the future on Earth have lived through a lot of technological progress, but also the semi-apocalyptic arrival of this alien race of Infernals. There's a lot of commonality with us as humans, but at the same time, they have mechs and technology that's well beyond what mankind has today. But it's an obvious advancement from technologies that we already have, so that faction should feel relatable and fun to play because of all the advancements that have happened.
The Infernal Hosts are culturally very different from us, and they're obviously biologically very different from us. They are intent on subjugating us and taking resources from our planet. They have their own weapons that to us are magical, and therefore very different. We don't necessarily understand how they work or how they're powered, so they are a very challenging foe that almost overruns our planet.
That creates this tension where mankind is fighting for survival, while the Infernals are fighting to accomplish their goal of subjugation. And they have their reasons that we'll learn more about through the story. But it creates this great sense of difference between the two perspectives.
Can you reveal anything new about the game setting? I know you said it's a far future Earth after an apocalyptic event, but is there anything else you can reveal about the environment?
Tim Morten: In Stormgate's future Earth, many of the trends that we've seen start in current day Earth and are taken to extensions. Certainly, climate change is something that is impacting mankind; overpopulation is something that is impacting mankind. And these are things that had led to the creation of a research program in Stormgate, called the Sigma program.
This program was focused on ways for mankind to find new habitats, whether that's through space travel or through portaling to other worlds. But basically, it's ways for us to find a backup to Earth. It was actually one of the Sigma research programs that led to the opening of the Stormgate that allowed the Infernals to storm Earth.
I think I know the answer this, but can you reveal how many factions there will be total? Or is that still a secret?
Tim Morten: We don't have a hard cap. We will definitely have more than two, but we don't have a hard cap on that number.
I know that building community for this game and receiving feedback from players, even before the final version is anywhere close to release, is really important to you. Can you speak a little bit to the importance of player community and how it relates to the game's development?
Tim Morten: One of the inspirations for us to make Stormgate is the attachment that we feel to the RTS community. Tim Campbell had come off of Warcraft III and the early days of StarCraft II, but he also worked on Red Alert 2 and Generals on the Command & Conquer side. For myself, I spent a little bit of time on Command & Conquer, but especially on StarCraft II. We really feel an attachment to the players that we've made games for, so it just makes sense to us as we develop Stormgate to try to include them in our process and to validate our ideas with them as we're starting to implement things.
We've had really great interaction through Reddit posts, where we post a topic and provide an opportunity for players to share their thoughts, then circle back by recapping what we saw in those replies. To the extent that we have conclusions, we share those conclusions with the community. And we've also done a fair number of direct feedback sessions with prominent figures in the RTS community, whether that's pro-players or streamers and casters. It's just so helpful to us to really understand their thought process because as a developer, there's always a little bit of getting so close to something that your perspective narrows. Re-broadening it by hearing those outside player opinions was just super valuable to us.
I know it's still early days, but can you think of any feedback that's already made significant changes in how the development process was going?
Tim Morten: Definitely. One of the early thoughts we had was that if players love playing the game socially, maybe we should design the game so that the only way you can play the game is socially. I think that was an understandable conclusion reached, but it was not a correct conclusion. We definitely got feedback that suggested, "No, actually, a lot of us still want to have the option to play solo if we want to play solo." So, we've made sure to continue to empower that with Stormgate.
What stage of development would you say you're at right now? Is there any timeline that you guys are working with?
Tim Morten: Currently, we are finishing pre-production on 1v1 competitive and working on what we would internally refer to as "a vertical slice," where we take some specific assets and bring them up to production quality for visuals. We take some specific game mechanics and take those up to production quality to help us validate that we've answered the creative questions that we wanted to answer for the modes, and that we are ready to start producing content for that mode.
That's an exercise we still have to go through for campaign and co-op, and it's an exercise we'll still have to go through for 3v3. It's hard to predict exactly when that phase will conclude for each of the modes, because so much of it is iterative and feedback-based, but we are going to start some testing next year. Beta will begin, and we'll have an opportunity for players outside the team to experience the modes in the order that we get them ready. And it'll be super valuable to us then to be able to use that feedback to tune the work that we've already done.
That's as much of a timeline as exists right now.
I know you've really been trying to strike a balance with Stormgate between providing a solid amount of challenge and in-depth strategy while also making the game accessible to players who are new to RTS. You compared it to Elden Ring and said it's much more accessible than previous titles. Can you talk a little bit more about how Stormgate is striking that balance, and what that challenge has been like?
Tim Morten: It's super important to us that we don't water down what makes real-time strategy great, and a big part of what we believe makes real-time strategy great is having depth and a high skill ceiling. At the same time, it's possible for us to think about how we lower the skill floor without bringing down the skill ceiling.
Really, what that means is that if you're a new player coming into the game, you don't have to master or even understand all of the depth that is eventually available for you to explore. You don't have to have a high actions-per-minute rate. Perhaps you don't even have to undertake all of the same tasks that an advanced player will take.
We see 1v1 competitive as a mode primarily focused on high skill competitive gameplay, so that's a mode where perhaps not all players will land. We saw even in StarCraft II that the majority of players prefer to play against AI rather than to play competitively. We see the higher player count modes being the modes that will appeal to new players coming into the game, and that's because there's an opportunity to rely on your friends and an opportunity to perhaps have your friends carry while you're still working things out.
But those are also modes where we will support heroes, or hero characters separate from the general army of your game. Those heroes each have different playstyles, so there's an opportunity for us to have heroes that don't require armies that are as big or perhaps don't require all of the same macro tasks in terms of managing your economy. But there are trade offs there. Anytime that simplifications like that are made, it creates an opportunity for more advanced players to really appreciate the things that controlling their own macro tasks to the most detailed extent, or having an army that has more units or different force compositions, afford them as advance as players to really excel and show off their skills.
I would say game modes and the differences between heroes are ways that we're exploring to create a lower skill floor for new players coming in without impacting that skill ceiling for advanced players.
What elements of Stormgate are you most excited about either from a developer standpoint, or in terms of getting to see player reactions to it eventually?
Tim Morten: There's so much there. First and foremost, the new universe and the new story are very exciting for me. Along with that, it's the way that we're presenting it. With Starcraft II and past Blizzard RTSes, there was so much time between story installments that players were kept waiting. We experimented a little bit towards the end of Legacy of the Void with the Nova covert ops missions, delivering content on a quarterly basis instead of having multiple years between drops. That really seemed to go over well, and that's something that I'm excited to try to apply from the beginning to the whole story so that we don't have those big gaps in story.
The social gameplay features, I believe, can be revolutionary. I really believe that being able to play with your friends in a way that encourages you to stick with the game and to keep coming back can transform the experience and cause RTS to become a lot more popular. That's something I'm really excited about. Integrated eSports, which will reduce the friction for players to participate in tournaments, even if they're not at the top Grandmaster professional level. Even just on a neighborhood basis, or who's the best player in your family or on the block, will help. Having the ability to participate in competitive tournaments more easily, I think, will be very revolutionary. Those are probably my top three, I guess.
Can you talk a little bit more about how that eSports aspect will work in the game?
Tim Morten: Absolutely. For StarCraft II, we had an automated tournament system, and that had some scheduled tournaments that players could sign up to participate in. But that wasn't tied to any ability to organize your own tournaments, and it also wasn't tied to any tournament leagues. The World Championships for StarCraft happened through signups completely outside of the game, as did all of the community tournaments.
We're putting the ability within the game to self-organize tournaments, and to participate in community tournaments that others have organized outside the automated tournament system, all the way up to the professional eSports that we will support. It creates the ability for players to see the availability of those directly in the client, but also participate in those from directly within the client with no need to go out to third-party websites or discover what's going on independently.
Is there anything else you want players to know about Stormgate?
Tim Morten: The last thing I'll throw out there is that user-generated content has been part of the DNA of real-time strategy for decades. It's how a lot of our team members got involved in developing games to begin with, so that's one more aspect of Stormgate that we are very passionate about. We're working on a custom level editor and providing the ability for players to seamlessly share content with each other, so that's something else that's in the works.
Deven McClure is a Senior Gaming Writer covering news, reviews, features, and interviews for Screen Rant, specializing in the simulation genre. A lifelong lover of video games, she began writing for Screen Rant in 2021. When she’s not working, you can undoubtedly find her playing her latest video game obsession – most likely something with farming and crafting. She’s been playing games since early childhood, first becoming obsessed with Animal Crossing for the Gamecube at 8 years old, and has loved learning about them ever since. Originally from California where she studied arts and child development, Deven moved to New York in 2017. After her move, she attended Gotham Writing School with a focus on article and television writing. She now lives in Brooklyn with her boyfriend and cats, Timmy and Tommy – named after the two adorable Animal Crossing characters, of course.
