Air to Ground radar?

Source?

Latest post in the “Unofficial Road Map” thread by Silver Dragon.

Well Yeah Doctrine and technology has change quite a bit. Now a days I doubt there is any Opfor with big concentrations of combat vehicles any more. But back in the cold war era finding this combat vehicles could be critical.

 

So basically yeah when we have the perfect means to find and kill vehicles, it is no longer relevant...

It was WW2 that changed the doctrines of marching etc. In WW1 all was about trenches, in WW2 it was Germany that developed a blitzkrieg, that evolved further after war to more distributed troops among allies. And in 60's the doctrine was already that you do not let enemy easily bomb your troops in marching with couple bombs.

And there is huge difference are you today fighting against a enemy at desert (billiard table) or in a forested area with full of ridges and dumps where to hide.

The DCS can't currently simulate a cold war missions because AI for ground troops is stupid, cheating and players wants an easy kill to feel superior.

DCS is great to learn the aircraft systems and procedures like how to release a bomb, how to acquire a designated target or how to perform a air refueling.

But when it comes to combat, or is extremely unrealistic because people prefer it that way and engine can't support that many various units there would be.

It literally is like 30 units instead 30000 units.

That is already something that is holding back the aircrafts features like A-G radar usefulness.

I would be happy if DCS would use dedicated 6 or 8 cores just for ground units. 3 or 4 to BLUFOR and 3 or 4 for REDFOR. Even dedicate 1-2 cores for infantry alone so we could have thousands of those on maps.

We already know that user aircraft can perfectly be run with single core, so let's get all other cores for everything else so we get major ground troops on maps and suddenly all the other features from EO or radars becomes far more differently used.

When the new weather system gets out, we likely see reduction of FLIR systems capabilities, and visually spotting targets to become more difficult.

So we want to use even limited A-G radar to find something at sea or ground like Bridges and buildings.

Source?

https://forums.eagle.ru/showpost.php?p=4041754&postcount=1551

 

Swap that for an area in Asia with a humid environment and a rainy season, an enemy who has a IADS and the ability to jam GPS, you might want to run interdiction missions with an aircraft with an a/g radar capability.

And to enemy that has passive radar detectors on ground, listening all radio and radar transmissions up in air, pin pointing aircrafts locations and informing those to their CAP.

Meaning, if you use radar, datalink, radio... Then the enemy knows your location (position and altitude) and can guide their fighters around you, prepare SAM NEZ traps etc.

We are leaving lots of all kinds tools that commanders would have in their possession to find and locate air and ground units.

My source is Chizh

Source?

Ru forums - I search for posts by Chizh and USSR_Rik from time to time.

So what's the holdup? Id love to read some details as to how and why the way ED are implementing AG radar that's taking so long.

I understand they dont want to do a fake-up like a line of sight filtered target array lookup for GMTI and SEA, but why not? And what else? And whats the holdup?

lets pray its not ruining the FPS, and if it does that it can be switched off :)

They have shown us nothing since 2015. Words are cheap.

First AG radar pictures are from 2015. Now is 2019, and as I understood it won't be ready this year. Maybe 2020?

Five years for developing one system...cool.

Yes and they said a few times that the guy who was responsible for the AG radar quit his job, that means basically someone else had to be found and this guy probably started at ZERO because it is easier than trying to understand what and why the guy did before.

You/we are talking about the same crap every month, the same guys are pissed about the same things every month - just stop buying EA modules - you can't handle this it seems! - please stop the endless crying.

Sorry for the harsh words but it's getting old.....

You/we are talking about the same crap every month, the same guys are pissed about the same things every month - just stop buying EA modules - you can't handle this it seems! - please stop the endless crying.

 

Sorry for the harsh words but it's getting old.....

This, so much this.

First AG radar pictures are from 2015. Now is 2019, and as I understood it won't be ready this year. Maybe 2020?

Five years for developing one system...cool.

Too many things and shifting priorities. Again a lot of work that is not useful anymore. Similar case to Wags comment on how all 3D work for F16 had to be dumped due to new engine and staffing changes, a similar thing happened for the radar, with one engineer working on it leaving the company (I remember something like that).

It is not easy because the development lifecycle for this system sometimes is so much slower that the core engine upgrading pace, which cause lots of problems. I really think they need to implement scrum methodologies in such systems, with minimal viable products that can be upgraded overtime (kind of EA style). The question is if its feasible to work with such a scheme for these systems, difficult to know without any further insight knowledge. Although my impression is that is feasible... with some concesions to realism in the first phases of each deliver.

Well if we get everything at a reasonable completeness level except the AG radar before end of the year, I will be happy equally

But when it comes to combat, or is extremely unrealistic because people prefer it that way and engine can't support that many various units there would be.

 

It literally is like 30 units instead 30000 units.

 

That is already something that is holding back the aircrafts features like A-G radar usefulness.

 

I would be happy if DCS would use dedicated 6 or 8 cores just for ground units. 3 or 4 to BLUFOR and 3 or 4 for REDFOR. Even dedicate 1-2 cores for infantry alone so we could have thousands of those on maps.

 

We already know that user aircraft can perfectly be run with single core, so let's get all other cores for everything else so we get major ground troops on maps and suddenly all the other features from EO or radars becomes far more differently used.

 

When the new weather system gets out, we likely see reduction of FLIR systems capabilities, and visually spotting targets to become more difficult.

So we want to use even limited A-G radar to find something at sea or ground like Bridges and buildings.

Not sure where to start, but will try hard to be nice... As a general rule we are terrible pilots. We fly solo, employ atrocious intercept tactics, and rely on terrible AI to survive. I reckon 99% of players have no idea about what intercept quadrants and positional advantage are, how to plan an intercept for desired lateral separation, the relationship between aspect, ATA, cut, HCA, etc., much less how to plan an intercept to hit a desired employment point. No real pilot would commit against multiple bandits head on, allowing themselves to get within WEZ.

FWIW there are exceptions. If you take the time to learn section BFM/SEM tactics, know first/second fighter roles, comm brevity, know the difference in your responsibility as the engaged fighter vs. free fighter, know how to keep your lift vector out of phase to set up a forward quarter shot if the engaged fighter can't employ after the first turn, there are groups with ex-pilots that fly that way... but it's a small group.

To some extent you are correct, recent conflicts have involved unchallenged air dominance, strategic JDAM strikes and CAS missions. A-G is not the primary sensor in such missions but hardly irrelevant. Its hugely important for ship strikes, without radar the harpoon is basically useless. Even GT and GMT are likely more useful in DCS given the lack of civilian traffic. Ground radar is very good at picking up large metal things hidden in foliage that a tpod would miss. But with one exception, recent conflicts have not been large scale ground engagements with unprotected tank columns in the desert. That isn't really due to doctrine so much as the relative force capabilities, but i digress.

A-G radar has a few other uses besides picking up 30000 ground units though. DBS and SAR are good examples given the types of missions we fly. We love tpods around here, even the gimped versions we have are nice for finding that JSOW target from 10nm out :lol:. SAR produces a lower resolution image than optical sensor, but it is all weather, limited in range only by the curvature of the earth and your altitude. And, as I said before, foliage that obscures hardened targets is penetrated by radar, making it particularly useful for producing an image from standoff range that can be shared via datalink for coordinated attack with SLAM-ERs without ever coming within range of that SA-10.

Everyone seems to forget about AGR and CCIP accuracy, or TA for night low altitude. Neither of which suck.

But honestly dude I don't know where to start on the psare 6-8 cores for ground unit AI. I am not going to drone on about the limitations of DX11, high overhead drivers, or the fact that the majority of DCS players don't have a "spare 8 cores." To wit I will just say that the devs will soon have the ability to better utilize multi core processors, it is among the principal drivers for the move to vulcan api.

Be careful what you wish for, however. I hate to drone on about how bad we are, but we are. Given the lack of coordination, knowledge and skill, my confidence in our ability to engage and suppress a modern integrated air defense is approximately 0%. Are you volunteering to be the guy who spends his afternoon getting a few dozen guys organized, taking off, lining up to top off, launching four TALD decoys, spending 20 min in the stack, landing, then maybe launching to help refuel inbound? We can't all lob aim-54's from 40 miles out, for every strike mission there's 10 dudes launching decoys, and 5 or 6 refueling other Hornets.

I'm not going to get into the development feasibility, I think is beyond the level of this thread. But your average DCS player would be pretty happy to have a 6700k. Good CPU, mine hit 4.9ghz and was damn near as good for gaming as my 9900k. Unfortunately, like most chips until very recently, it had only 4-cores... so that spare 6-8 might be an issue. Also, I seriously question your assertion that 1 core is enough for everything so use all the other cores for AI, but it's an irrelevant point. It also ignores the fact that you still have to draw those units, so its GPU resources are well. But again, irrelevant point.

AG radar is taking a while, I'll give you that. So is TWS, AZ/EL, scan centering, bump acquisition, SCAN RAID, EXP modes, L&S stepping, spotlight, among others. We'd all like it faster but it is what it is, bitching isn't going to help, especially when we don't have a clue what we're talking about :music_whistling:

Not sure where to start...

That deserves to be sicked and/or put as a last point in rules section :)

That deserves to be sicked and/or put as a last point in rules section :)

I liked the "we all suck" line personally...

Well if we get everything at a reasonable completeness level except the AG radar before end of the year, I will be happy equally

I think that's waaaay over ambitious, but I like your optimism......

Not sure where to start, ...

Well said. Most of us want systems that most never learn how to use to their full extent and we cry about realism and then go 1v1 with an SA-10. I would love to get more systems, but I would also prefer that things like VOIP are implemented, so at least we can all use comms in MP. A lot of people don't bother cooperating or using comms today, which is one of the most immersion-breaking things out there.

As for IADS, I'm with you. I've tried to set up a "realistic" IADS network using triggers, with SAM traps and SAMs having a chance of going dark if a HARM is launched. It's my own mission, I know where everything is and how it'll react and even like that, it's impossible to successfully engage it by myself.

I think the Chinese, North Koreans, Russia, Iran, might disagree with your assessment of how many combat vehicles they have.

 

 

For the past 20 years Western airforces have mostly only been performing CAS mission in areas of the world with mostly perfect conditions for E/O sensors, in a fairly benign threat environment.

 

 

Plenty of countries have been caught out preparing for the last war they fought, only to find the next one is completely different.

 

 

Swap that for an area in Asia with a humid environment and a rainy season, an enemy who has a IADS and the ability to jam GPS, you might want to run interdiction missions with an aircraft with an a/g radar capability.

Exactly this. Not to mention TGP's don't see through weather very well, while radar is an all-weather weapon. Not a huge problem in places that have relatively benign weather most of the year. But a big tactical consideration in places that don't. One of the key drivers of JDAM development was in fact this exact problem, LGB's had serious drawbacks when the cloud base was low in Yugoslavia. Not to mention the rather famous sand storms and episodes of shitty weather in both the first and second Iraq war air campaigns.

Not sure where to start, but will try hard to be nice... As a general rule we are terrible pilots. We fly solo, employ atrocious intercept tactics, and rely on terrible AI to survive. I reckon 99% of players have no idea about what intercept quadrants and positional advantage are, how to plan an intercept for desired lateral separation, the relationship between aspect, ATA, cut, HCA, etc., much less how to plan an intercept to hit a desired employment point. No real pilot would commit against multiple bandits head on, allowing themselves to get within WEZ.

 

FWIW there are exceptions. If you take the time to learn section BFM/SEM tactics, know first/second fighter roles, comm brevity, know the difference in your responsibility as the engaged fighter vs. free fighter, know how to keep your lift vector out of phase to set up a forward quarter shot if the engaged fighter can't employ after the first turn, there are groups with ex-pilots that fly that way... but it's a small group.

 

To some extent you are correct, recent conflicts have involved unchallenged air dominance, strategic JDAM strikes and CAS missions. A-G is not the primary sensor in such missions but hardly irrelevant. Its hugely important for ship strikes, without radar the harpoon is basically useless. Even GT and GMT are likely more useful in DCS given the lack of civilian traffic. Ground radar is very good at picking up large metal things hidden in foliage that a tpod would miss. But with one exception, recent conflicts have not been large scale ground engagements with unprotected tank columns in the desert. That isn't really due to doctrine so much as the relative force capabilities, but i digress.

 

A-G radar has a few other uses besides picking up 30000 ground units though. DBS and SAR are good examples given the types of missions we fly. We love tpods around here, even the gimped versions we have are nice for finding that JSOW target from 10nm out :lol:. SAR produces a lower resolution image than optical sensor, but it is all weather, limited in range only by the curvature of the earth and your altitude. And, as I said before, foliage that obscures hardened targets is penetrated by radar, making it particularly useful for producing an image from standoff range that can be shared via datalink for coordinated attack with SLAM-ERs without ever coming within range of that SA-10.

 

Everyone seems to forget about AGR and CCIP accuracy, or TA for night low altitude. Neither of which suck.

 

But honestly dude I don't know where to start on the psare 6-8 cores for ground unit AI. I am not going to drone on about the limitations of DX11, high overhead drivers, or the fact that the majority of DCS players don't have a "spare 8 cores." To wit I will just say that the devs will soon have the ability to better utilize multi core processors, it is among the principal drivers for the move to vulcan api.

 

Be careful what you wish for, however. I hate to drone on about how bad we are, but we are. Given the lack of coordination, knowledge and skill, my confidence in our ability to engage and suppress a modern integrated air defense is approximately 0%. Are you volunteering to be the guy who spends his afternoon getting a few dozen guys organized, taking off, lining up to top off, launching four TALD decoys, spending 20 min in the stack, landing, then maybe launching to help refuel inbound? We can't all lob aim-54's from 40 miles out, for every strike mission there's 10 dudes launching decoys, and 5 or 6 refueling other Hornets.

 

I'm not going to get into the development feasibility, I think is beyond the level of this thread. But your average DCS player would be pretty happy to have a 6700k. Good CPU, mine hit 4.9ghz and was damn near as good for gaming as my 9900k. Unfortunately, like most chips until very recently, it had only 4-cores... so that spare 6-8 might be an issue. Also, I seriously question your assertion that 1 core is enough for everything so use all the other cores for AI, but it's an irrelevant point. It also ignores the fact that you still have to draw those units, so its GPU resources are well. But again, irrelevant point.

 

AG radar is taking a while, I'll give you that. So is TWS, AZ/EL, scan centering, bump acquisition, SCAN RAID, EXP modes, L&S stepping, spotlight, among others. We'd all like it faster but it is what it is, bitching isn't going to help, especially when we don't have a clue what we're talking about :music_whistling:

Sad but true. Most of the "realism" in the sim is limited by solo players. One alternative solution, and not a perfect one by any means, is to have solo players be flight leads of AI flights. Then again I'm not really convinced it would lead to more realism.

Not sure where to start, but will try hard to be nice... As a general rule we are terrible pilots. We fly solo, employ atrocious intercept tactics, and rely on terrible AI to survive. I reckon 99% of players have no idea about what intercept quadrants and positional advantage are, how to plan an intercept for desired lateral separation, the relationship between aspect, ATA, cut, HCA, etc., much less how to plan an intercept to hit a desired employment point. No real pilot would commit against multiple bandits head on, allowing themselves to get within WEZ.

 

FWIW there are exceptions. If you take the time to learn section BFM/SEM tactics, know first/second fighter roles, comm brevity, know the difference in your responsibility as the engaged fighter vs. free fighter, know how to keep your lift vector out of phase to set up a forward quarter shot if the engaged fighter can't employ after the first turn, there are groups with ex-pilots that fly that way... but it's a small group.

 

To some extent you are correct, recent conflicts have involved unchallenged air dominance, strategic JDAM strikes and CAS missions. A-G is not the primary sensor in such missions but hardly irrelevant. Its hugely important for ship strikes, without radar the harpoon is basically useless. Even GT and GMT are likely more useful in DCS given the lack of civilian traffic. Ground radar is very good at picking up large metal things hidden in foliage that a tpod would miss. But with one exception, recent conflicts have not been large scale ground engagements with unprotected tank columns in the desert. That isn't really due to doctrine so much as the relative force capabilities, but i digress.

 

A-G radar has a few other uses besides picking up 30000 ground units though. DBS and SAR are good examples given the types of missions we fly. We love tpods around here, even the gimped versions we have are nice for finding that JSOW target from 10nm out :lol:. SAR produces a lower resolution image than optical sensor, but it is all weather, limited in range only by the curvature of the earth and your altitude. And, as I said before, foliage that obscures hardened targets is penetrated by radar, making it particularly useful for producing an image from standoff range that can be shared via datalink for coordinated attack with SLAM-ERs without ever coming within range of that SA-10.

 

Everyone seems to forget about AGR and CCIP accuracy, or TA for night low altitude. Neither of which suck.

 

But honestly dude I don't know where to start on the psare 6-8 cores for ground unit AI. I am not going to drone on about the limitations of DX11, high overhead drivers, or the fact that the majority of DCS players don't have a "spare 8 cores." To wit I will just say that the devs will soon have the ability to better utilize multi core processors, it is among the principal drivers for the move to vulcan api.

 

Be careful what you wish for, however. I hate to drone on about how bad we are, but we are. Given the lack of coordination, knowledge and skill, my confidence in our ability to engage and suppress a modern integrated air defense is approximately 0%. Are you volunteering to be the guy who spends his afternoon getting a few dozen guys organized, taking off, lining up to top off, launching four TALD decoys, spending 20 min in the stack, landing, then maybe launching to help refuel inbound? We can't all lob aim-54's from 40 miles out, for every strike mission there's 10 dudes launching decoys, and 5 or 6 refueling other Hornets.

 

I'm not going to get into the development feasibility, I think is beyond the level of this thread. But your average DCS player would be pretty happy to have a 6700k. Good CPU, mine hit 4.9ghz and was damn near as good for gaming as my 9900k. Unfortunately, like most chips until very recently, it had only 4-cores... so that spare 6-8 might be an issue. Also, I seriously question your assertion that 1 core is enough for everything so use all the other cores for AI, but it's an irrelevant point. It also ignores the fact that you still have to draw those units, so its GPU resources are well. But again, irrelevant point.

 

AG radar is taking a while, I'll give you that. So is TWS, AZ/EL, scan centering, bump acquisition, SCAN RAID, EXP modes, L&S stepping, spotlight, among others. We'd all like it faster but it is what it is, bitching isn't going to help, especially when we don't have a clue what we're talking about :music_whistling:

So well put. Everyone should read that. One of the first things I do is remind myself, when I get in the cockpit in VR is, I have no idea what I am doing really but I'm gonna try my hardest within this limited framework to do it right.

Thanks for that, it needed to be said.

So well put. Everyone should read that. One of the first things I do is remind myself, when I get in the cockpit in VR is, I have no idea what I am doing really but I'm gonna try my hardest within this limited framework to do it right.

 

Thanks for that, it needed to be said.

For all our complaints, DCS must be a pretty deep game given the size of the manuals that come with each module. That said, for every word in the Hornet game manual, the US Navy has produced thousands. If you are truly interested in study, send a PM or something and I will be happy to point you in the right direction or fly with you.

I would love to read some evidence that Vulkan would give you anything that a properly leveraged DirectX 12 won't provide. Not because I care about one over the other, but because I am a graphics nerd and would love to read about it. So if you have any good links, share them here and maybe we can distract people from the A/G radar whine fest by saying Vulkan over and over.

There's no peer review scientific study that is going to say a particular piece of software runs better on one api over the other. There's too many variables, particularly with the new lightweight APIs as so much comes down to implementation. They are actually very comparable technically, and arguments can be made for both. I don't pretend to know what ED's internal deliberations were but running a proprietary engine as DCS does tends to incentivize devs to prefer open source solutions. Many devs, especially non-AAA studios that don't get custom driver updates for their launches, are eager to get out from under MS lock-in, less driver surprises, and for games like DCS where, relative to the FPS flavor of the month, much more time is spent on the engine and graphics relative to gave development, an open source API where you can build what you want (rather than beg MS) would be pretty attractive. Again, I am not privy to their internal discussion and am just guessing here. It's worth noting that by and large most sims (and other open world games) are going vulkan. Some of the reasons explained below...

First though, I am not sure you're asking the right question. What we should be comparing is Vulkan to DX11, and how those APIs perform using physically based rendering techniques. I don't know my audience here, so I am not sure how technical to be. I am shooting for middle of the road but apologies if I miss the mark. Rather than DSX12, what we are really interested in the difference between what we have now, and what we will have next - right? So not DX12 vs Vulkan? But rather, Vulkan vs. DX11.

Again, not knowing my audience here... GPU's have matured from dedicated shader calculators to programmable devices - first CUDA, parallel processing, now tensor and RT cores... At the same time, drivers have become enormously bloated. Every time a new AAA is released there's a driver update. That's a lot of CPU overhead just to tell the GPU what to do, which is made worse by the poor multi-thread support. This is obviously an issue for DCS.

So to start with, by requiring devs to clearly define how operations will be executed, CPU overhead is decreased by a substantially lower overhead driver and much improved multitcore utilization. Almost everything in Vulkan is premised on the concept of queues. Everything from drawing a unit, compute functions like AI, weather, and flight models, even memory operations. For many graphics operations assigning a queue family to a logical device (CPU/GPU) means the CPU doesn't have to do it, and as operations are sent to the command buffer, or some variation of primary/secondary buffer depending on how the developer chooses, but either way the command buffer can be built across theoretically (I think?) infinite cores.

I don't pretend to be an expert on the topic, and I don't fully understand semaphores/fences aspect - basically devs handle app level threading and synchronization issues, but allowing multiple threads to create and submit commands in parallel is much more efficient in multi-core CPU. I will leave it at that because, again, I'm not read up on the tools devs have to manage the asynchronous computation issues, but I think we all agree that better multi-core utilization is good.

While the biggest advantage is CPU utilization, for flight sims it may/should be better GPU efficiency as well. Flight sims tend to draw a lot of very similar objects, where loads consists of a large number of objects that do not require complex shading. Vulkan allows devs to adjust the number of worker threads and move rendering off the main thread for what hopefully will be a big performance boost for flight sims.

If interested, here is a demo on vulkan rendering dozens of high poly helicopters with very low CPU overhead (hit v to disable v-sync). https://nvidia.app.box.com/s/rj9nzxkpz21qu34h8zew301kray9urbf The threaded fish demo allows you to change the number of schools displayed, and the number of worker threads to give a more interactive feel of the difference: http://developer.download.nvidia.com/mobile/shield/assets/ThreadedRenderingVk/ThreadedRenderingVk_20160707.zip

I haven't heard ED say anything on point, but ray tracing is almost certainly in the long term plan. All ray tracing is, fundamentally, is simulating light exactly as it behaves. Landscapes are all about light, and a given DCS draw may have multiple tens of thousands of light sources. Not all light is visible, either. I've wondered if ray tracing wouldn't solve some of the challenges related to air to ground radar implementation - it is, essentially, built in real beam, and evidently pretty easy to add in Vulcan (supported down to 1070 cards now, I believe), and can be toggled on and off for those that do not have compatible cards. As they are meshing everything already, I suspect its in the plan, though probably not right away. We can hope though. Mesh demo here for those unfamiliar:

My last point is VR. In addition to the above, games must move to either DX12 or Vulkan if they want to support variable rate shading. Unfortunately it is RTX only, so it to is probably a couple years aways, but VRS is the foundation for next gen VR. VRS allows developers to increase processing power/image quality applied to different areas of an image very quickly. This tech is a prerequisite of foveated rendering, which is simply eye tracking combined with VRS, maximizing image quality where you are looking, and decreasing quality where you are not.

Why is it that I always miss these messages, and only come back to the forum after they've been deleted and posting rights revoked? Damn!!! :lol:

 

Perhaps slightly off-topic, and I apologise up-front for it, but any news on Vulkan? Do we really think it is going to make a huge difference? Will it change much with regards to visuals, or will it be more of a CPU/RAM/Maximum number of units/potentially larger engagements thing? Imagine, servers with hundreds of people online at the same time, with no hitches, hiccups or micro-freezes!!!!!

Reasonable expectations are key, we won't see double fps overnight. Its a change they will have to eventually make to keep up with the industry and be ready for next gen VR, and there is good reason to expect improved performance particularly for systems bottlenecked at the CPU. It should also free up resources to add more complexity elsewhere given the additional computation resources available.

Last I heard the engine was done, or mostly done, and they were redoing textures and shaders. I don't know if they can just re-export objects or if everything must be rebuilt. Note also how all the cockpits/maps are being upgraded to mesh. It's been a while since I heard an ETA, so I hesitate to even say anything, but last i heard was early 2020.

That said, not hating on ED, but the big downside to Vulkan is that with all the control it gives devs, its offers a ton of opportunity to screw things up. Development has been slow for everyone, and the No Man's Sky launch performance issues are probably going to be pretty common. So while it will be a good change, particularly in long run, I'd expect some rocky betas. Still, if you've had the chance to play NMS on a good system, you can extrapolate what's coming, which I think is worth being excited for.

...

 

First though, I am not sure you're asking the right question. What we should be comparing is Vulkan to DX11, and how those APIs perform using physically based rendering techniques. I don't know my audience here, so I am not sure how technical to be. I am shooting for middle of the road but apologies if I miss the mark. Rather than DSX12, what we are really interested in the difference between what we have now, and what we will have next - right? So not DX12 vs Vulkan? But rather, Vulkan vs. DX11.

 

...

Thanks, but no. I did ask the exact question I meant to ask :). The differences between DX11 and Vulkan (or DX12) are obvious to any graphics nerd I think? I was wondering why everyone is making such a big deal out of going to Vulkan over going to DX12. You mentioned untangling from MS and that several (or most?) big open world engines would go to Vulkan, which is interesting. So that part kinda makes sense to me. If domain experts believe the support from MS and perhaps larger pool of developers (?) are worth less than having more agency and control over their own destiny, I think I buy that.

Sorry you wrote so much stuff on the other topic, but I am sure someone who doesn't know what Vulkan is probably will learn a lot from your post.