Forums » Suggestions
RFC: Next Generation VO System Requirements
While none of these are changes that are likely to happen "tomorrow", I think it's probably a good idea to start discussing how the game's minimum requirements may be changing. Some of this may occur later this year (although I'm not sure how much), so to be clear, this is not a "someday" discussion that is purely hypothetical.
I am posting this for feedback, so people are welcome to ask questions or comment. That's why it's here.
Moving along.. I've previously discussed how we're dropping the 32bit Windows version and the upcoming Windows installer updates, and those are likely the first major shift that will appear. I'm also considering dropping the 32bit Linux version, as it has barely been used since 2023, and getting rid of that would eliminate one more build that we have to maintain and support.
I suspect that MOST PLAYERS will not be impacted by this set of proposed changes below. There is a very small number of people who are likely to be impacted by these changes. We will be continuing to measure and monitor who would be impacted, and to what extent.. this isn't going to happen overnight.
More broadly than that, though, we should discuss "why", which leads to the..
Next-Generation Vendetta Online Engine
This is still in the planning stages, but basically:
1) We need to modernize how the engine and renderer work, to take advantage of certain features in modern hardware (like compute shaders).
2) When we do this, it will make some older hardware (say, pre-2010?) pretty problematic for running the game. It "might" still work, but it may become totally unsupported, or not look right.
It is the process of making everything "still work" on really old hardware that is a major development burden. We're trying to simplify things, so we can move ahead more cleanly.
Why is this required? Doesn't the game work now?
Yes, it's not really about the game As It Stands Today, it's about where we would like the game to go.
We need to set a "reasonable floor" for the hardware capabilities, if we're going to meaningfully change and update the game. Otherwise, it's very difficult to plan for new content, as it's a 25-year-old compatibility minefield. For example:
- The "Universe Redux" test, the Unknown System, has demonstrated we need to make some engine architecture changes to make things more efficient across a broad base of modern hardware.
- It's currently slower and more complicated to create new ships and stations and other content, which would be improved with features like modern "Physically Based Renderer" systems for prototyping.
- Instead of supporting a bunch of different renderers on different platforms, we want to migrate towards supporting one major render path that's used everywhere.
- We can add a lot of cooler, prettier stuff if we can work with at least slightly "modern" hardware capabilities.
We have specific goals for technical capabilities we want to have (and specific gameplay goals associated with those technical features); but I don't want to reveal things specifically, lest we have to cancel some feature and be accused of misleading anyone.
Does that mean I need super new hardware?
No, probably not. It just means we're likely to raise the minimum, so VO may not be quite so friendly to "Running on an Ancient Potato" as it has been historically.
So what kind of specs are we talking about?
I'm breaking things down into three categories: "Target", "Legacy" and "Unsupported":
* Target - This is basically the main focus of forward development, all new features will end up here.
* Legacy - This will be an upgraded version of the current game render tech, intended to be "compatible" with the new Target-level features. We will still fix bugs, but otherwise will be "frozen" in time and not improved past that point.
* Unsupported - Older hardware that may work.. or not. Good luck. We aren't fixing problems here anymore.
"Target" supported hardware:
GPU (Windows/Linux): Vulkan 1.3 compliant GPU with 2GB of video RAM (GTX960, AMD RX 400, Intel ARC). Recent drivers.
GPU (Mac): Mac M1 or better (Vulkan 1.3 on Metal via KosmicKrisp).
GPU (Linux): Vulkan 1.3 compliant GPU with 2GB of video RAM (GTX960, AMD Radeon HD 7000 on Linux with RAD-V).
GPU (Mobile): Vulkan 1.3 compliant SoC.
PC CPU: Intel Sandy Bridge (6th gen), such as i3 3250, or AMD Bulldozer, such as FX-4350 or better.
RAM: 4GB or more.
OS: 64bit Windows 10 or 11 (or Linux64, recent MacOS, Android 8+, etc).
Display: 1920x1080 display (1080p) or better.
For best results, a GPU with Vulkan 1.4+ and EXT_mesh_shaders support will be recommended. This includes the GTX1650+, 20x0, 30x0 and so on, as well as similar AMD RDNA2 and Intel ARC GPUs.
Similarly, a modern Ryzen or recent Intel will be recommended on the CPU.
"Legacy" supported hardware:
GPU (Windows): DirectX 11 supported GPU with DX 11_0 compliant hardware, with 2GB of video ram (NVIDIA GTX 460, AMD Radeon 5870, Intel HD 4000. Our test case will be a GTX 660ti).
GPU (Mac): Intel Mac back to mid-2012, running the latest supported MacOS version (Vulkan-on-metal via MoltenVK).
GPU (Linux): I'm actually not sure yet. I would like to completely deprecate and remove the Linux OpenGL driver, and mandate Vulkan if possible (oddly, Linux has some of the best Vulkan support, thanks to Valve). Otherwise, OpenGL 4.5+.
GPU (Mobile): OpenGL ES 3.2 supported GPU, or iPhone 8+ iOS devices with Metal support (Vulkan-on-Metal via MoltenVK).
CPU: Any 64bit x86-64 compatible multi-core CPU with support for SSE3 (includes older Q6600, Phenom II, etc).
RAM: 2GB or more.
OS: 64bit Windows 10 or 11.
Display: 1280x720 or better, 1080p recommended.
"Unsupported" hardware that may still run the game, somewhat:
- Pre-DX11_0 hardware with DX11 driver on Windows: It may work, or just perform terribly, or look wrong. It may also work well enough for "station menu usage and chat", but not for really flying around and playing the game (or, it might "work" in an empty sector, and then be 4fps in an asteroid sector).
- 64bit Windows 7 or 8 may work, but are not guaranteed.
- Linux: I'm not sure yet, still figuring that out.
Can't we just "turn down settings" like we always have?
No, because we're talking about a fundamental change in rendering architecture. So, it won't quite work that way. This isn't just "more or less polygons on screen" or "less detailed textures", this is more fundamentally about how any of it gets on screen.
As an analogy, imagine the point in history when we went traveling by horse or carriage to using automobiles: Roads had to be smoother, gas stations had to be added everywhere, the entire paradigm of transportation had to shift.
The shift did more than just add automobiles, it also made horses less feasible or acceptable as a form of transport.
Or, to give a more specifically pointed explanation..
- If a modern user buys a modern, but low-spec laptop device, like a Chromebook or cheap Windows laptop, and then runs the current game on it (at the time of this posting).. that device will (currently) deliver a much poorer experience for the person, simply because we still support "old hardware".
- Conversely, if we update our renderer and engine to more modern requirements, relatively cheap (modern) hardware will be able to deliver a lot better-looking game, which we will have an easier time maintaining and improving.
- These benefits will also impact everyone else with hardware going back to at least 2015 (and probably back to higher-end GPUs from 2010 or so).
- People with more modern hardware (2018+) will have potentially a lot better fidelity and capabilities.
Won't it just be "slow" on unsupported hardware?
Not necessarily. It might work, or it might be fundamentally broken, or not run at all.
And, even if we try to make it "run" to some extent, it may be so much slower that it's basically pointless.
Can you say anything about how this will make things better?
Sure, we should at least have these capabilities:
- Higher fidelity graphics and visuals on all supported hardware, and especially on modern hardware.
- Potentially much larger sectors, with more content in them, higher object density, more complex types of objects (relevant to The Future of Mining).
- The same assets distributed across on all platforms, using XUASTC textures that cross-converts to the mandates of the local GPU. This is a *major* benefit for the developers, and simplifies adding new content (far less test cases).
- Physically Based Renderer material pipeline ("PBR"), allows for a lot of modern, high-fidelity material effects that should look much better than the current game. Also makes it much easier to prototype new assets.
- A modern GPU-driven renderer, making use of Compute Shaders, is much more efficient at using GPU resources, and allows a great deal more content on-screen at the same level of performance (compared to an older architecture that is more CPU-driven).
- More optimized CPU-driven side will make use of high-performance "SIMD" instructions, based on availability. This is called "Dynamic Dispatch", and it means we can have the game only require SSE3, but still make usage of more recent instructions (like AVX2) on devices that support them, without breaking backwards compatibility. This can make certain code 4x or 8x faster, or more.
When will this take place?
I don't know exactly, but again, this is a meaningful part of our planning matrix and not some "hypothetical never-never concept". This issue is (and has been) a bit of a blocker on some forms of game progress for some years.
We'll probably at least do some significant updates to the "Legacy" renderer this year, to get a good idea of a minimum baseline and how well that works, before working on the next-generation renderer. Or, that development might be concurrent.
Changes will not happen all at once, there'll likely be a number of staged changes, as we update different platforms.
I am posting this for feedback, so people are welcome to ask questions or comment. That's why it's here.
Moving along.. I've previously discussed how we're dropping the 32bit Windows version and the upcoming Windows installer updates, and those are likely the first major shift that will appear. I'm also considering dropping the 32bit Linux version, as it has barely been used since 2023, and getting rid of that would eliminate one more build that we have to maintain and support.
I suspect that MOST PLAYERS will not be impacted by this set of proposed changes below. There is a very small number of people who are likely to be impacted by these changes. We will be continuing to measure and monitor who would be impacted, and to what extent.. this isn't going to happen overnight.
More broadly than that, though, we should discuss "why", which leads to the..
Next-Generation Vendetta Online Engine
This is still in the planning stages, but basically:
1) We need to modernize how the engine and renderer work, to take advantage of certain features in modern hardware (like compute shaders).
2) When we do this, it will make some older hardware (say, pre-2010?) pretty problematic for running the game. It "might" still work, but it may become totally unsupported, or not look right.
It is the process of making everything "still work" on really old hardware that is a major development burden. We're trying to simplify things, so we can move ahead more cleanly.
Why is this required? Doesn't the game work now?
Yes, it's not really about the game As It Stands Today, it's about where we would like the game to go.
We need to set a "reasonable floor" for the hardware capabilities, if we're going to meaningfully change and update the game. Otherwise, it's very difficult to plan for new content, as it's a 25-year-old compatibility minefield. For example:
- The "Universe Redux" test, the Unknown System, has demonstrated we need to make some engine architecture changes to make things more efficient across a broad base of modern hardware.
- It's currently slower and more complicated to create new ships and stations and other content, which would be improved with features like modern "Physically Based Renderer" systems for prototyping.
- Instead of supporting a bunch of different renderers on different platforms, we want to migrate towards supporting one major render path that's used everywhere.
- We can add a lot of cooler, prettier stuff if we can work with at least slightly "modern" hardware capabilities.
We have specific goals for technical capabilities we want to have (and specific gameplay goals associated with those technical features); but I don't want to reveal things specifically, lest we have to cancel some feature and be accused of misleading anyone.
Does that mean I need super new hardware?
No, probably not. It just means we're likely to raise the minimum, so VO may not be quite so friendly to "Running on an Ancient Potato" as it has been historically.
So what kind of specs are we talking about?
I'm breaking things down into three categories: "Target", "Legacy" and "Unsupported":
* Target - This is basically the main focus of forward development, all new features will end up here.
* Legacy - This will be an upgraded version of the current game render tech, intended to be "compatible" with the new Target-level features. We will still fix bugs, but otherwise will be "frozen" in time and not improved past that point.
* Unsupported - Older hardware that may work.. or not. Good luck. We aren't fixing problems here anymore.
"Target" supported hardware:
GPU (Windows/Linux): Vulkan 1.3 compliant GPU with 2GB of video RAM (GTX960, AMD RX 400, Intel ARC). Recent drivers.
GPU (Mac): Mac M1 or better (Vulkan 1.3 on Metal via KosmicKrisp).
GPU (Linux): Vulkan 1.3 compliant GPU with 2GB of video RAM (GTX960, AMD Radeon HD 7000 on Linux with RAD-V).
GPU (Mobile): Vulkan 1.3 compliant SoC.
PC CPU: Intel Sandy Bridge (6th gen), such as i3 3250, or AMD Bulldozer, such as FX-4350 or better.
RAM: 4GB or more.
OS: 64bit Windows 10 or 11 (or Linux64, recent MacOS, Android 8+, etc).
Display: 1920x1080 display (1080p) or better.
For best results, a GPU with Vulkan 1.4+ and EXT_mesh_shaders support will be recommended. This includes the GTX1650+, 20x0, 30x0 and so on, as well as similar AMD RDNA2 and Intel ARC GPUs.
Similarly, a modern Ryzen or recent Intel will be recommended on the CPU.
"Legacy" supported hardware:
GPU (Windows): DirectX 11 supported GPU with DX 11_0 compliant hardware, with 2GB of video ram (NVIDIA GTX 460, AMD Radeon 5870, Intel HD 4000. Our test case will be a GTX 660ti).
GPU (Mac): Intel Mac back to mid-2012, running the latest supported MacOS version (Vulkan-on-metal via MoltenVK).
GPU (Linux): I'm actually not sure yet. I would like to completely deprecate and remove the Linux OpenGL driver, and mandate Vulkan if possible (oddly, Linux has some of the best Vulkan support, thanks to Valve). Otherwise, OpenGL 4.5+.
GPU (Mobile): OpenGL ES 3.2 supported GPU, or iPhone 8+ iOS devices with Metal support (Vulkan-on-Metal via MoltenVK).
CPU: Any 64bit x86-64 compatible multi-core CPU with support for SSE3 (includes older Q6600, Phenom II, etc).
RAM: 2GB or more.
OS: 64bit Windows 10 or 11.
Display: 1280x720 or better, 1080p recommended.
"Unsupported" hardware that may still run the game, somewhat:
- Pre-DX11_0 hardware with DX11 driver on Windows: It may work, or just perform terribly, or look wrong. It may also work well enough for "station menu usage and chat", but not for really flying around and playing the game (or, it might "work" in an empty sector, and then be 4fps in an asteroid sector).
- 64bit Windows 7 or 8 may work, but are not guaranteed.
- Linux: I'm not sure yet, still figuring that out.
Can't we just "turn down settings" like we always have?
No, because we're talking about a fundamental change in rendering architecture. So, it won't quite work that way. This isn't just "more or less polygons on screen" or "less detailed textures", this is more fundamentally about how any of it gets on screen.
As an analogy, imagine the point in history when we went traveling by horse or carriage to using automobiles: Roads had to be smoother, gas stations had to be added everywhere, the entire paradigm of transportation had to shift.
The shift did more than just add automobiles, it also made horses less feasible or acceptable as a form of transport.
Or, to give a more specifically pointed explanation..
- If a modern user buys a modern, but low-spec laptop device, like a Chromebook or cheap Windows laptop, and then runs the current game on it (at the time of this posting).. that device will (currently) deliver a much poorer experience for the person, simply because we still support "old hardware".
- Conversely, if we update our renderer and engine to more modern requirements, relatively cheap (modern) hardware will be able to deliver a lot better-looking game, which we will have an easier time maintaining and improving.
- These benefits will also impact everyone else with hardware going back to at least 2015 (and probably back to higher-end GPUs from 2010 or so).
- People with more modern hardware (2018+) will have potentially a lot better fidelity and capabilities.
Won't it just be "slow" on unsupported hardware?
Not necessarily. It might work, or it might be fundamentally broken, or not run at all.
And, even if we try to make it "run" to some extent, it may be so much slower that it's basically pointless.
Can you say anything about how this will make things better?
Sure, we should at least have these capabilities:
- Higher fidelity graphics and visuals on all supported hardware, and especially on modern hardware.
- Potentially much larger sectors, with more content in them, higher object density, more complex types of objects (relevant to The Future of Mining).
- The same assets distributed across on all platforms, using XUASTC textures that cross-converts to the mandates of the local GPU. This is a *major* benefit for the developers, and simplifies adding new content (far less test cases).
- Physically Based Renderer material pipeline ("PBR"), allows for a lot of modern, high-fidelity material effects that should look much better than the current game. Also makes it much easier to prototype new assets.
- A modern GPU-driven renderer, making use of Compute Shaders, is much more efficient at using GPU resources, and allows a great deal more content on-screen at the same level of performance (compared to an older architecture that is more CPU-driven).
- More optimized CPU-driven side will make use of high-performance "SIMD" instructions, based on availability. This is called "Dynamic Dispatch", and it means we can have the game only require SSE3, but still make usage of more recent instructions (like AVX2) on devices that support them, without breaking backwards compatibility. This can make certain code 4x or 8x faster, or more.
When will this take place?
I don't know exactly, but again, this is a meaningful part of our planning matrix and not some "hypothetical never-never concept". This issue is (and has been) a bit of a blocker on some forms of game progress for some years.
We'll probably at least do some significant updates to the "Legacy" renderer this year, to get a good idea of a minimum baseline and how well that works, before working on the next-generation renderer. Or, that development might be concurrent.
Changes will not happen all at once, there'll likely be a number of staged changes, as we update different platforms.
Sounds great to me, testing legacy hardware on a 660ti might distort thinga a bit since the 660ti absolutely smokes the HD3000/HD4000 and the GT 720/HD6570 that you find in many SFF desktops.
Sounds great to me, testing legacy hardware on a 660ti might distort thinga a bit since the 660ti absolutely smokes the HD3000/HD4000 and the GT 720/HD6570 that you find in many SFF desktops.
(Rewriting my earlier response to be a little clearer)..
Quick sum up:
- "Target" is what we're actually building content for.
- "Legacy" should render properly (albeit not the same as Target), but we cannot guarantee performance.
- "Unsupported" may not render properly at all.
The GT660Ti is definitely much faster than integrated GPUs of the era. But, I think we'll be able to scale reasonably to meet some of the integrated GPUs. At least, I hope so.
Our primary "baseline" target for Vulkan 1.3 is the Qualcomm Adreno 740 (used in the Quest 3's Snapdragon XR2 Gen 2), which is a pretty capable modern mobile chip, but is nothing like an RTX4090 or "big desktop" GPU. As a result, I don't think our baseline will be pushed so high as to be impossible for some less-capable Legacy GPUs with diminished visual-quality settings.
We do have an in-house Intel HD 5000 (I think), so that'll be a reasonable bare-minimum test. The Intel HD3000 is not a "hardware 11_0" device, it's a 10_1 device, so that falls under "Unsupported" and not "Legacy".
That all being said, baseline content will be defined by the Target (Vulkan 1.3 on Adreno 720), and Legacy will be intended to be as compatible as possible, but without guaranteeing any particular performance.
We'll also do what we can to stretch the reasonable performance as far "back" as possible, probably using increasingly ugly settings (and I suspect we'll succeed in the goals of my post). But, if we eventually have to say "The game is just not-playable on an HD4000", that's what we're going to do. Legacy will not be driving the overall target here, that isn't what "Legacy" is for.
At some point, the in-game scenes are what they are, and I'm not going to cut a huge number of asteroids, or drastically shrink space stations, to make them functional on a circa-2012 bare-minimum iGPU like the HD 4000. (That's actually what I'm trying to get away from).
So, while we'll do our best to make things function as broadly as possible, Legacy expectations should be aligned to the above reality.
I hope that makes sense.
(Rewriting my earlier response to be a little clearer)..
Quick sum up:
- "Target" is what we're actually building content for.
- "Legacy" should render properly (albeit not the same as Target), but we cannot guarantee performance.
- "Unsupported" may not render properly at all.
The GT660Ti is definitely much faster than integrated GPUs of the era. But, I think we'll be able to scale reasonably to meet some of the integrated GPUs. At least, I hope so.
Our primary "baseline" target for Vulkan 1.3 is the Qualcomm Adreno 740 (used in the Quest 3's Snapdragon XR2 Gen 2), which is a pretty capable modern mobile chip, but is nothing like an RTX4090 or "big desktop" GPU. As a result, I don't think our baseline will be pushed so high as to be impossible for some less-capable Legacy GPUs with diminished visual-quality settings.
We do have an in-house Intel HD 5000 (I think), so that'll be a reasonable bare-minimum test. The Intel HD3000 is not a "hardware 11_0" device, it's a 10_1 device, so that falls under "Unsupported" and not "Legacy".
That all being said, baseline content will be defined by the Target (Vulkan 1.3 on Adreno 720), and Legacy will be intended to be as compatible as possible, but without guaranteeing any particular performance.
We'll also do what we can to stretch the reasonable performance as far "back" as possible, probably using increasingly ugly settings (and I suspect we'll succeed in the goals of my post). But, if we eventually have to say "The game is just not-playable on an HD4000", that's what we're going to do. Legacy will not be driving the overall target here, that isn't what "Legacy" is for.
At some point, the in-game scenes are what they are, and I'm not going to cut a huge number of asteroids, or drastically shrink space stations, to make them functional on a circa-2012 bare-minimum iGPU like the HD 4000. (That's actually what I'm trying to get away from).
So, while we'll do our best to make things function as broadly as possible, Legacy expectations should be aligned to the above reality.
I hope that makes sense.
*shakes fist in "ancient potato hardware"*
Just kidding, I think we will all be satisfied so long as we can keep playing the game. I have a touchscreen HP Envy 23 All-in-one that is a dedicated VO box, and I'd be sad if I had to replace it, but would it disgruntle me enough to stop paying y'all for the ability to fly my 'Dent? HELL NO, VO FOREVER!
Just kidding, I think we will all be satisfied so long as we can keep playing the game. I have a touchscreen HP Envy 23 All-in-one that is a dedicated VO box, and I'd be sad if I had to replace it, but would it disgruntle me enough to stop paying y'all for the ability to fly my 'Dent? HELL NO, VO FOREVER!
We certainly appreciate that :). And, for what it's worth, we'll do what we can (within settings and such) to give your device as much basic functionality as possible, for as long as possible.
As a quick followup, I've seen a few people with low-end devices have been understandably gloomy from my post. But, I want to note a couple of more optimistic points:
- We're probably going to make the existing renderers a bit faster and more efficient (even on older computers), before we build the new Vulkan renderer.
- When we go about doing that, we may find enough "headroom" to keep things working on older devices.
So, I don't know for sure if any hardware used by a current player is going to become "unplayable". We won't know that until we start testing with the updated renderers, and then build the new engine and content.
A lot of the reason for this post is more to set a clear set of design goals around what we're "targeting" and why, so that people understand if things get slower, or worst-case, become unplayable on some particular piece of hardware.
But whether anyone's hardware is actually unplayable? That remains to be seen.
- We're probably going to make the existing renderers a bit faster and more efficient (even on older computers), before we build the new Vulkan renderer.
- When we go about doing that, we may find enough "headroom" to keep things working on older devices.
So, I don't know for sure if any hardware used by a current player is going to become "unplayable". We won't know that until we start testing with the updated renderers, and then build the new engine and content.
A lot of the reason for this post is more to set a clear set of design goals around what we're "targeting" and why, so that people understand if things get slower, or worst-case, become unplayable on some particular piece of hardware.
But whether anyone's hardware is actually unplayable? That remains to be seen.