Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
What is your point exactly?
That because not many games are running with Metal a year after release it's dead in the water?
One and a half years after release. But yes, that's pretty much my point. But to specify: that there aren't many games using Metal is a result of it being dead in the water, not the other way round – at least until Apple improves/fixes Metal so far that the developers can release their Metal based games without having to fear a complete disaster.

That Feral had to delay two ports indefinitely due to apparent problems with Metal (and I have yet to see any indication that macOS 10.12.2 fixed the problems that held back the release of Deus Ex Mankind Divided, Total War: Warhammer and Obduction), and that Aspyr pretty much went on record stating that supporting Metal is currently not worth the effort, does nothing to improve the impression.
 
Last edited:
WoW was already included in the number of Metal supporting games you can count "on the fingers of a single hand" I mentioned. The other two are Headlander and Refunct (see below).


…which I knew about (see above), but I wasn't aware they are using the UE4. Okay, point taken. But still: that precisely one single game (and a fairly obscure one nonetheless). This hardly equates to an impressive adoption rate. Unless you have other examples, of course…


Well, fact is also that several developers are sitting on Metal versions of their Unity/UE4 games, waiting for bug fixes by Apple and possibly the Second Coming of Christ (whichever might come first), with Obduction (UE4) and Firewatch (Unity) probably being the most prominent examples. So just having the engines capable of using Metal is apparently not sufficient.

One and a half years after release. But yes, that's pretty much my point. But to specify: that there aren't many games using Metal is a result of it being dead in the water, not the other way round – at least until Apple improves/fixes Metal so far that the developers can release their Metal based games without having to fear a complete disaster.

That Feral had to delay two ports indefinitely due to apparent problems with Metal (and I have yet to see any indication that macOS 10.12.2 fixed the problems that held back the release of Deus Ex Mankind Divided, Total War: Warhammer and Obduction), and that Aspyr pretty much went on record stating that supporting Metal is currently not worth the effort, does nothing to improve the impression.

I generally don't find your points false, so don't perceive the following as a counter, but rather an addition with a few elements of counters.

1) How many DX12 titles can you count that have been released? Or Vulkan? I can name a few, but it isn't like that requires many more fingers. Adopting a new API can be really time consuming and game production is a process that takes years.
2) I've seen confirmation that the Metal bug fixes are in 12.2; Though I can't remember where, so I release it's anecdotal evidence. But I believe it came from Feral or Aspyr themselves.
3) Yes, I've had email correspondence with Aspyr, and they said the gains from switching to Metal over OpenGL didn't lead to significant enough improvements that it would be worth it. However, with WoW there were significant improvements, and the Unity team seems happy about Metal too, so it goes both ways.
4) My biggest gripe with Metal though, isn't performance benefits or whatnot. It's features. If Apple is abandoning OpenGL and not going to also allow Vulkan on their platforms, we only have Metal, and as of right now, Metal doesn't have as much in terms of features as its counterparts. It's also easier to use (coming from the Unity blog), but lacking features is a huge bummer.
 
1) How many DX12 titles can you count that have been released? Or Vulkan? I can name a few, but it isn't like that requires many more fingers. Adopting a new API can be really time consuming and game production is a process that takes years.
About 20 games supporting DirectX 12, currently 4 supporting Vulkan on Windows and/or Linux (plus roughly the same number on Android) – and that in much shorter time. Admittedly, in some of these games, the support is still labeled as "beta", but it's still available in publicly available release versions.
 
About 20 games supporting DirectX 12, currently 4 supporting Vulkan on Windows and/or Linux (plus roughly the same number on Android) – and that in much shorter time. Admittedly, in some of these games, the support is still labeled as "beta", but it's still available in publicly available release versions.

If you include Android, we should include iOS Metal titles too, so let's just leave Android out of this, yeah?

So Vulkan is roughly like Metal, although yes, admittedly a shorter time period.
DX12 - well, a bit higher a number than I expected, but if you then look at how many games Windows gets compared to Mac, you divide it down, Metal has pretty decent adoption rate I'd say
 
  • Like
Reactions: MrUNIMOG
I don't think that this is a matter of analogies (PCs divided with PC Games, vs Macs divided with Mac games) rather than if Metal actually brought the Mac gaming one step forward in general; truth is it hasn't. Of course Feral and Aspyr will support Metal - they support Mac as a platform (with or without Metal) and Mac gaming would be near zero without them.

The real question is; how many triple-A titles available to PCs have they been released for Mac (that they most probably wouldn't be released without Metal) ? Optimistic or not, truth is Mac gaming is at no better state now than it was before. Regarding the mention to blizzard and wow (that is hardly a new game), let's not forget that for the first time in their history they kept a game from released on Mac.
 
Last edited:
If you include Android, we should include iOS Metal titles too, so let's just leave Android out of this, yeah?
I'm just mentioning this because you can find lists with Vulkan supporting games where these Android apps are counted indiscriminately, thus inflating the number somewhat. I agree that such mobile apps shouldn't play a role in this discussion.
 
  • Like
Reactions: MrUNIMOG
BTW. Since UE4 has Mac support, how come we so rarely see Mac versions of UE games, and when we do, they usually still go through someone like Feral? How come the original developers so rarely release Mac versions themselves?

I'm not sure there have been any UE4 games for the Mac yet, have there? Obduction that I know of, which isn't released yet. Older UE games needed to be ported, especially since some like XCom2 are heavily modified and barely resemble the original engine.

Unless you have other examples, of course…

Why, since you'd just dismiss those as well. ;) Might be a good idea to read Mark's post again.

--Eric
 
BTW. Since UE4 has Mac support, how come we so rarely see Mac versions of UE games, and when we do, they usually still go through someone like Feral? How come the original developers so rarely release Mac versions themselves?

I'm not sure there have been any UE4 games for the Mac yet, have there? Obduction that I know of, which isn't released yet. Older UE games needed to be ported, especially since some like XCom2 are heavily modified and barely resemble the original engine.

Eric5h5 is pretty much correct but it really comes down to market size & money. When developers like Cyan (Obduction) or Frontier (Elite) take a risk on developing a Mac version they need to see Mac gamers buy it to invest in more.

I've only been at Epic working on UE4, the older Mac native port of UE3 was developed & shipped while I was at Feral. It was used on a few games (e.g. Spec Ops: The Line) but most UE3 games that made it to Mac were licensed by Feral or Aspyr and as they have their own porting libraries that's obviously what they'd choose to reuse. Most original developers & publishers back then were barely interested in PC versions, much less a Mac release, as the PS3 and Xbox 360 were so dominant.

Times are a little bit different now but its still the reality that not every developer has the necessary infrastructure to deliver releases on every platform under the sun, so for them licensing to an Aspyr or Feral just makes more sense. You even see this with Square Enix PC releases, which are typically developed by their dedicated PC studio, Nixxes and other publishers have similar arrangements these days. WB's Arkham Knight shows what can happen when this goes awry - but more often than not gamers aren't aware of the separation of development duties. I'm sure that at some point Feral and/or Aspyr will take on a UE4 game and I'll be most intrigued how they get on - there are a number of high-level engine changes for Mac that can't be easily hidden away in a porting library...
 
  • Like
Reactions: MrUNIMOG
e.g. Spec Ops: The Line) but most UE3 games that made it to Mac were licensed by Feral or Aspyr and as they have their own porting libraries that's obviously what they'd choose to reuse

Thanks for the information!
What just puzzles me endlessly now however, is how Spec Ops then ran so terribly, if it were using a native UE3 implementation and not a ported library. Holy crap the glitches! Crashes, textures missing entirely, frame rate issues (less a problem compared to other issues) and controller support that was very varying at best. A Steam update broke Dualshock 3 compatibility for the game, whilst other games kept on working fine.
 
Yeah :/. Sounds like progress, but also sounds like Metal really isn't quite there yet for Mac usage. To me, it seems like it's ready for use with Imagination GPUs </snip>

It would be fair to say that Mac Metal is a work in progress with iOS Metal being understandably ahead given that it has had longer in development.

and, maybe, Intel GPUs. Not quite there for dGPUs

This seems to be a commonly accepted myth but it is false. There are problems with Mac Metal but they don't cut that cleanly. If anything Metal suits one of the dGPU vendors very well, given the API design choices of DX12/Mantle/Vulkan...
[doublepost=1482185353][/doublepost]
Brad Oliver (Feral programmer) posted this on Twitter.

https://twitter.com/hoserama99/status/808650977536397312

And it being software development it introduced new ones as well. :)

Yeah, that's our lot in life!

Regarding the mention to blizzard and wow (that is hardly a new game), let's not forget that for the first time in their history they kept a game from released on Mac.

Aspyr's blog post explains that the problems with bringing high-end AAA games to Mac extends well beyond the graphics API right now - the preponderance of low-RAM, Intel-GPU-only configurations in the Mac install base makes the *business* case for a Mac version pretty hard to make even when it is technically possible.

I don't think that this is a matter of analogies (PCs divided with PC Games, vs Macs divided with Mac games) rather than if Metal actually brought the Mac gaming one step forward in general; truth is it hasn't.

</snip>

Optimistic or not, truth is Mac gaming is at no better state now than it was before.

That's a pretty reasonable sentiment. Metal is a good API that adds a lot of features (integrated compute, multi-threaded dispatch, offline byte-code compiled shaders and with 10.12 fragment shader explicit texture writes & tessellation) that bring Mac graphics support closer to D3D11. Unfortunately that means us Mac developers are still playing catch-up so we've not really seen the benefits yet. The positive for me is that the rate of development for Metal year-on-year is much higher than it was with OpenGL, so there is hope.

About 20 games supporting DirectX 12, currently 4 supporting Vulkan on Windows and/or Linux (plus roughly the same number on Android) – and that in much shorter time. Admittedly, in some of these games, the support is still labeled as "beta", but it's still available in publicly available release versions.
So Vulkan is roughly like Metal, although yes, admittedly a shorter time period.
DX12 - well, a bit higher a number than I expected, but if you then look at how many games Windows gets compared to Mac, you divide it down, Metal has pretty decent adoption rate I'd say

For contrast: Vulkan support in UE4 for Desktop is currently limited to D3D 10/GL 3 "Shader Model 4" equivalence, i.e. the same as Mac OpenGL. We can't run the editor under Vulkan yet either - that still has to run in OpenGL or D3D. Mac Metal support is currently at D3D 11 "Shader Model 5" equivalence and has run the Editor from the first iteration in UE4 4.11 (which was SM4 only).

Had Apple waited for Vulkan it couldn't have arrived prior to 10.12 and we'd now be even further behind than we are with Metal. While the progress hasn't yet translated into shipping games as I discussed earlier - it is still pretty significant and puts us in a much better position going forward. There are some small features that I really do want Apple to bring to Metal and I'm sure others have their own lists but Vulkan would not be a panacea to the problem.

I'd also like to stress that not all the problems are caused by Apple or the GPU vendors. UE4 4.15 required a pretty significant amount of refactoring of our own Metal code to address issues which only surfaced as we switched games to Metal during the year. No matter how good you think you are, or how experienced sometimes your solution turns out to not be as efficient or elegant as you think it is. Software development really is hard and the engineers at Apple (& GPU vendors) working on Metal are genuinely good people who are working diligently on improving graphics support on Apple's platforms. Don't be too hard on the poor beggars because they won't get everything right and neither do I.
 
This seems to be a commonly accepted myth but it is false. There are problems with Mac Metal but they don't cut that cleanly. If anything Metal suits one of the dGPU vendors very well, given the API design choices of DX12/Mantle/Vulkan...

Well, I'd never expect it to be a clean cut at all, and I've seen how AMD usually benefits a lot more from the low-level APIs than nVidia. It's just that, if I am not mistaken, Apple first enabled Metal only on the Intel GPUs for the desktop environment rendering, and then first later on dGPUs. That would indicate to me they felt more confident about the functionality under Intel GPUs.

Aspyr's blog post explains that the problems with bringing high-end AAA games to Mac extends well beyond the graphics API right now - the preponderance of low-RAM, Intel-GPU-only configurations in the Mac install base makes the *business* case for a Mac version pretty hard to make even when it is technically possible.

low-RAM? I mean, I get the other problems, especially the install base problem, which I wager has always been the greatest factor in the limitted number of Mac ports being made, but I don't quite follow the RAM thing. I'm yet to see a game that requires 32gigs of RAM and 16gigs seems to be the standard for relatively new Macs. I've 16 in pretty much everything, though my iMac will soon go the way of 32 for video editing. But now that you mention that point, how come Mac ports usually require ≈2gigs more RAM than the Windows version? And since Intel has recently made pretty good strides with GPU, especially the Iris line, is it really that massive a problem only to have an iGPU? I mean, I'd never expect Titan X or Fury performance from an Iris, but isn't the Iris somewhat competitive with the 750m? That's a pretty decent card for laptops. Not amazing at all, but decent.

For contrast: Vulkan support in UE4 for Desktop is currently limited to D3D 10/GL 3 "Shader Model 4" equivalence, i.e. the same as Mac OpenGL. We can't run the editor under Vulkan yet either - that still has to run in OpenGL or D3D. Mac Metal support is currently at D3D 11 "Shader Model 5" equivalence and has run the Editor from the first iteration in UE4 4.11 (which was SM4 only).

Had Apple waited for Vulkan it couldn't have arrived prior to 10.12 and we'd now be even further behind than we are with Metal. While the progress hasn't yet translated into shipping games as I discussed earlier - it is still pretty significant and puts us in a much better position going forward. There are some small features that I really do want Apple to bring to Metal and I'm sure others have their own lists but Vulkan would not be a panacea to the problem.

I'd also like to stress that not all the problems are caused by Apple or the GPU vendors. UE4 4.15 required a pretty significant amount of refactoring of our own Metal code to address issues which only surfaced as we switched games to Metal during the year. No matter how good you think you are, or how experienced sometimes your solution turns out to not be as efficient or elegant as you think it is. Software development really is hard and the engineers at Apple (& GPU vendors) working on Metal are genuinely good people who are working diligently on improving graphics support on Apple's platforms. Don't be too hard on the poor beggars because they won't get everything right and neither do I.

Sorry for including such a long quote but I kinda feel like I want to address pretty much all paragraphs.
Firstly; Really? The editor can run with Metal? I actually thought the UE editor only ran on Windows and other platform versions could be made from it, but running the editor was limited to Windows. Wicked!

Well, I wouldn't want Apple to replace Metal with Vulkan or anything. I'd just like Apple to maintain a cross-platform open source alternative alongside Metal. Windows having DX12 doesn't exclude Vulkan for instance. And I get that not having Vulkan forces stronger adoption of Metal since there isn't that alternative, but it'd make it easier to create cross platform stuff.

And lastly, my intent was never to bash developers. Neither at Epic, Apple, Aspyr or anywhere else. Whilst I've never done graphics programming, I've written enough code (hobby, nothing special, but still) to know it can be a killer sometimes. I mean, recently, an iOS update broke something I'd made, and I had no reason what had changed and why it'd screwed up the app. Fixed the crashing with help from the Debugger, but then some UI elements went wonky and so on. So yeah, point is that I get it. I really appreciate all the work done by every single person contributing to these projects. When you encounter a glitch, or software isn't available yet or whatnot, it's easy to get annoyed, especially if you lost work or whatnot, but you never say "hey, it's pretty great" when it works perfectly, you just use it, and expect it to work.

As always you bring lovely posts. Thanks for all the info
 
  • Like
Reactions: MrUNIMOG
What is your point exactly?
That because not many games are running with Metal a year after release it's dead in the water?

It hardly matters when Apple refuses to offer a single Macintosh with gaming level graphics chips in them. What's the difference? Apple is in the process of killing off the Mac. First it killed professional Macs and soon it will be hard to tell a Macbook from an iPad. 256GB of storage is for small lightly used computers, smart phones and tablets, not $2400 notebooks in 2016/17. And that's just one issue.
 
  • Like
Reactions: antonis
It's just that, if I am not mistaken, Apple first enabled Metal only on the Intel GPUs for the desktop environment rendering, and then first later on dGPUs.

The system UI and a game are very different beasts.


In terms of the typical Mac that makes up the install base, rather than us more demanding users, the skew is toward several year old laptops which often only shipped with <= 8GB RAM. PC gamers skew toward having lots of RAM, big-fat GPUs and CPUs. I'll point out that this matters much more for the Editor which consumes oodles of RAM. The flip-side of this is that it does suggest that Mac users tend to be happier with their H/W for longer...

But now that you mention that point, how come Mac ports usually require ≈2gigs more RAM than the Windows version?

On the ports I've worked on there's often a memory requirement increase as a consequence of having to keep copies of data in system memory to optimise around weaknesses in OpenGL vs. D3D. Some Mac ports are also 64bit rather than the Windows version's 32bit which can increase memory usage. Plus there's usually some other odds and ends that are tricky to account for without going into the nitty-gritty of a particular project.

And since Intel has recently made pretty good strides with GPU, especially the Iris line, is it really that massive a problem only to have an iGPU? I mean, I'd never expect Titan X or Fury performance from an Iris, but isn't the Iris somewhat competitive with the 750m? That's a pretty decent card for laptops. Not amazing at all, but decent.

For PC games the NV 750M is now actually low-end and often unsupported! The Intel Iris Pro is generally slower still... not many AAA games support Intel on PC at all - those that do are outliers.

Firstly; Really? The editor can run with Metal?

Yep, the UE4 Editor has run on Mac since the first 4.0 release way back in 2014 - day and date with PC. Originally it used OpenGL but in 4.11, Metal became the default.

Well, I wouldn't want Apple to replace Metal with Vulkan or anything. I'd just like Apple to maintain a cross-platform open source alternative alongside Metal. Windows having DX12 doesn't exclude Vulkan for instance. And I get that not having Vulkan forces stronger adoption of Metal since there isn't that alternative, but it'd make it easier to create cross platform stuff.

Look at it from a business stance, why would you develop your own API and then go and support a competitors? That looks like unnecessarily duplicated effort and potentially negates any competitive advantage you might achieve with your own API if no-one adopts it. Apple have historically been much more controlling of their OS and what APIs appear on it than Microsoft, for good (yes, there is good in this) and ill - this is an example.

And lastly, my intent was never to bash developers.

I don't feel that you are, but there's an underlying sense to this thread that Apple aren't trying. That's not true. I was also trying to reinforce the point that not all bugs are their fault - I've been the most guilty party at Epic for Metal support and I've made my fair share of mistakes in the code that caused problems ;)

As always you bring lovely posts. Thanks for all the info

You're welcome.
 
Last edited:
  • Like
Reactions: MrUNIMOG
low-RAM? I mean, I get the other problems, especially the install base problem, which I wager has always been the greatest factor in the limited number of Mac ports being made, but I don't quite follow the RAM thing.
About 1/3rd of macs currently using Steam are equipped with 4GB or less. Most of the remainder are 8GB. With Macs, they're often the only Mac the user has. The actual Gaming PC market moved to 8GB minimums around the time DDR3 came out. There are still a good number of 4GB PCs about; but they're usually secondary devices (cheap laptops and so on).

Also keep in mind that this does not account for DRAM - or rather, that Macs seldom have it. That means that in practice, you have less memory to utilise on Macs - and a lot less memory bandwidth.
 
In terms of the typical Mac that makes up the install base, rather than us more demanding users, the skew is toward several year old laptops which often only shipped with <= 8GB RAM. PC gamers skew toward having lots of RAM, big-fat GPUs and CPUs. That was covered by Aspyr's blog post very nicely.

Sure, but that's comparing an average user to a gamer. Compare average Mac user to average PC user, and the RAM situation, I'd wager, leans towards Mac users having more RAM in general. But I have done no study to back this up. Of course this argument circles right back to the business argument with the install base problem

On the ports I've worked on there's often a memory requirement increase as a consequence of having to keep copies of data in system memory to optimise around weaknesses in OpenGL vs. D3D. Some Mac ports are also 64bit rather than the Windows version's 32bit which can increase memory usage. Plus there's usually some other odds and ends that are tricky to account for without going into the nitty-gritty of a particular project.

That makes sense. Keeping more stuff in memory generally seems like a good thing to do as well, as long as the memory is there of course. But loading it from disk or going to swap is pretty much the same, so no real downside the way I see it.

For PC games the NV 750M is now actually low-end and often unsupported! The Intel Iris Pro is generally slower still... not many AAA games support Intel on PC at all - those that do are outliers.

Oh - that's actually a surprise to me. My MacBook Pro has the Iris Pro (I think it's 5300? Maybe it's 5200? The Crystal Well (Haswell) one) and I am actually very impressed with what Intel has done. Remember the days of GMA... Oh God GMA. I guess that the fact that even when I Bootcamp I only play older games helps in that I haven't noticed.

Yep, the UE4 Editor has run on Mac since the first 4.0 release way back in 2014 - day and date with PC. Originally it used OpenGL but in 4.11, Metal became the default.

Super bloody cool that is. Pretty unrelated, but is there some way of getting to mess about with it, without licensing for comercial use, if you just want to dick about with it for fun?

Look at it from a business stance, why would you develop your own API and then go and support a competitors? That looks like unnecessarily duplicated effort and potentially negates any competitive advantage you might achieve with your own API if no-one adopts it. Apple have historically been much more controlling of their OS and what APIs appear on it than Microsoft, for good (yes, there is good in this) and ill - this is an example.

Well, the way I see it, Khronos and Vulkan isn't as much a competitor really. I mean, Apple is a member of the Kronos Group afterall. And it's not like having Swift means Apple won't support compiling apps that've been written in C++, right? i get the business argument, but here's another one - Who's the end customer? Probably someone who doesn't care if it's Metal or Vulkan if it runs well, right? Would supporting both bring more interested developers and thereby more software to the platform and ecosystem? Probably, right? Going by that, it'd be smart to support both.

I don't feel that you are, but there's an underlying sense to this thread that Apple aren't trying. That's not true. I was also trying to reinforce the point that not all bugs are their fault - I've been the most guilty party at Epic for Metal support and I've made my fair share of mistakes in the code that caused problems ;)

Sure that's fair enough. I don't think the tone against Apple is entirely fair, but I also doubt you'll dispute that Apple doesn't prioritise games in the same way that MS does, right? And I'm not saying Apple isn't doing anything, but it's just not their field in the same way. I mean, Microsoft makes games and they've got the Xbox and all that, so it's a no brainer it'd be more of a concern for them. But for people who love the Mac and also love games, it can understandably be frustrating to see the favourite platform be less competitive in that area.

The system UI and a game are very different beasts.

Very true. Used to even be two different GPUs for 2D and 3D stuff, so yeah. But surely then a subset of Metal that related to the UI, Apple felt was more suited for the iGPUs? Or do you wager it had more to do with them needing the performance to run Retina displays more, and thereby getting prioritised for what you might call beta-testing?
[doublepost=1482196091][/doublepost]
Also keep in mind that this does not account for DRAM - or rather, that Macs seldom have it. That means that in practice, you have less memory to utilise on Macs - and a lot less memory bandwidth.

What on Earth do you mean? DDR is DRAM. Double Data Rate Dynamic Random Access Memory. Maybe you meant VRAM for the Macs without dGPUs?

About 1/3rd of macs currently using Steam are equipped with 4GB or less. Most of the remainder are 8GB. With Macs, they're often the only Mac the user has. The actual Gaming PC market moved to 8GB minimums around the time DDR3 came out.

How many Steam PC users have 4 gigs or less? As I said earlier, comparing gamers and average users I don't think is entirely fair. And the comment about 8gigs being minimum from when DDR3 came out, that I don't really buy at all. That's way back in 2007, and game developers did not aim for 8gigs as a minimum back then. To give some context, Crysis 1 was released 2007 and for Vista and newer it required 1.5gigs of RAM or more
 
  • Like
Reactions: MrUNIMOG
What on Earth do you mean? DDR is DRAM.
dRAM is sometimes used (if inappropriately) as an abbreviation for dedicated VRAM, to separate it from the case where regular RAM used as VRAM. The reason is that software usually does not differentiate between the two.

But yes to clarify; theres often a pool of dedicated VRAM installed with the GPU on PCs. Most Macs do not come with this. (As most macs are Airs and 13'' Pros)

That's way back in 2007, and game developers did not aim for 8gigs as a minimum back then.
No, but gamers did to allow other software such as browsers to run in parallel. You need to look beyond what just the games needed.
 
Last edited:
dRAM is sometimes used (if inappropriately) as an abbreviation for dedicated VRAM, to separate it from the case where regular RAM used as VRAM. The reason is that software usually does not differentiate between the two.

Had never heard that before. I'd say that's confusing as bunkers though, as DRAM already is an abbreviation, and there's no way of knowing if the use relates to VRAM or static vs. dynamic.

No, but gamers did to allow other software such as browsers to run in parallel. You need to look beyond what just the games needed.

8 gigs though? Crysis was really demanding and required 1.5 gigs on Vista or 7..... 1.5. With 4gigs that leaves 2,5 gigs, which back then was a decent bit. (OS usage is accounted for in the game's needs). If a game recommends 8 gigs nowadays, do you also think you need 42gigs to be safe (scaled by the same amount as in the Crysis example)? And if you play a game, you don't need that much running in the background, unless you like being distracted a lot
[doublepost=1482197350][/doublepost]
Had never heard that before. I'd say that's confusing as bunkers though, as DRAM already is an abbreviation, and there's no way of knowing if the use relates to VRAM or static vs. dynamic.



8 gigs though? Crysis was really demanding and required 1.5 gigs on Vista or 7..... 1.5. With 4gigs that leaves 2,5 gigs, which back then was a decent bit. (OS usage is accounted for in the game's needs). If a game recommends 8 gigs nowadays, do you also think you need 42gigs to be safe (scaled by the same amount as in the Crysis example)? And if you play a game, you don't need that much running in the background, unless you like being distracted a lot

Here's an excerpt from an article from 2007
It is amazing how the hardware market has changed over the last two years. Processor clock speed hasn't increased, yet performance has multiplied thanks to dual- or quad-processing cores per CPU. Graphics power traditionally doubles with every new graphics processor generation; 2 GB of RAM have become fairly affordable and hard drives have reached the terabyte-capacity level. With the exception of specific applications and workloads in the area of high-definition content, audio/video transcoding, biometry or scientific workloads, sufficient performance is accessible for all mainstream users today - even for gaming.


The article says 2 gigs of Ram has just gotten affordable. 8 is mentally much at this point.
 
  • Like
Reactions: MrUNIMOG
Sure, but that's comparing an average user to a gamer. Compare average Mac user to average PC user, and the RAM situation, I'd wager, leans towards Mac users having more RAM in general.

Quite possibly - we're really discussing the types of Macs being used to play games vs. the types of PCs used for games and for that there is a difference. As noted by others Apple don't really have an equivalent to a gaming PC for Mac, which then feeds into the installed base.

Unrelated, but is there some way of getting to mess about with it, without licensing for comercial use, if you just want to dick about with it for fun?

Funnily enough, yep, it's free to use:
https://www.unrealengine.com/what-is-unreal-engine-4

Well, the way I see it, Khronos and Vulkan isn't as much a competitor really. I mean, Apple is a member of the Kronos Group afterall.

Vulkan absolutely is a competitor to Metal as it has been adopted as the next-gen 3D API for Android, so a proprietary API is a potential competitive advantage. Supporting Vulkan could undermine that as it may reduce adoption. Also, Apple aren't a member of the Vulkan board, their Khronos membership is a legacy of their support of OpenGL.

Sure that's fair enough. I don't think the tone against Apple is entirely fair, but I also doubt you'll dispute that Apple doesn't prioritise games in the same way that MS does, right?

That's absolutely fine and I would agree. I disagree with sentiments such as "Metal sucks" - I think that's a bit unfair and dismissive of the efforts that have gone into it. If it were easy to design a 3D graphics API then OpenGL would not have become such a cruel mistress...
 
  • Like
Reactions: MrUNIMOG
Quite possibly - we're really discussing the types of Macs being used to play games vs. the types of PCs used for games and for that there is a difference. As noted by others Apple don't really have an equivalent to a gaming PC for Mac, which then feeds into the installed base.

Sure that's fair enough. I think for most people who play games on Macs, the Mac was purchased for other things (in my case general compute and video editing) and gaming was an "oh well, I have a nice computer" sort of thing - unlike with PCs where some people have PCs made specifically with gaming in mind. The only way of doing that on the Mac side is pretty much Hacintosh. Aside from that there are compromises all around.

Funnily enough, yep, it's free to use:

You do realise you just ruined my exams, right? Hehe

Vulkan absolutely is a competitor to Metal as it has been adopted as the next-gen 3D API for Android, so a proprietary API is a potential competitive advantage. Supporting Vulkan could undermine that as it may reduce adoption. Also, Apple aren't a member of the Vulkan board, their Khronos membership is a legacy of their support of OpenGL.

Sure, but being used by Android doesn't necessarily make it a competitor. Whilst it's not the sole way of doing Android apps anymore, Java isn't a competitor to apps either right? It's a technology used to create a product, not a product in an of itself. Made as an open standard. But I didn't realise Apple wasn't with the Khronos group aside from a remnant from the OpenGL usage. And yes it would slow adoption of Metal, but it would also increase availability of software on the platform. Furthermore, the things that will be made with Vulkan on Android won't just be 1. party stuff. It'll be 3. party stuff too, so Apple could get some of those apps over on iOS easier with Vulkan support (same deal with Windows/Linux and Mac). I can see both arguments, but personally, I find it more sensible to have both technologies for developers to use.

That's absolutely fine and I would agree. I disagree with sentiments such as "Metal sucks" - I think that's a bit unfair and dismissive of the efforts that have gone into it. If it were easy to design a 3D graphics API then OpenGL would not have become such a cruel mistress...

Agreed.
 
  • Like
Reactions: MrUNIMOG
Sure, but being used by Android doesn't necessarily make it a competitor.

Pardon? Apple iOS devices and Andriod devices are most definitely direct competitors. The situation is in the grand tradition of the olden days of 8bit micros and 16bit consoles where you had to choose between one or the other so the HW companies fought tooth and nail for business (e.g. Sega Mega Drive vs Super Nintendo) and desperately sought any competitive advantage they could find. That includes SDKs and APIs.

I can see both arguments, but personally, I find it more sensible to have both technologies for developers to use.

Both strategies do have merit as well as drawbacks and I'm curious as to how it will play out between companies adopting different approaches. It definitely isn't as simple as this API being good and that one being less so.
 
  • Like
Reactions: MrUNIMOG
Pardon? Apple iOS devices and Andriod devices are most definitely direct competitors. The situation is in the grand tradition of the olden days of 8bit micros and 16bit consoles where you had to choose between one or the other so the HW companies fought tooth and nail for business (e.g. Sega Mega Drive vs Super Nintendo) and desperately sought any competitive advantage they could find. That includes SDKs and APIs.

Android and iOS, most definitely. Vulkan being used by Android doesn't make it a competitor to Apple though. I mean, isn't that sort of like saying that because Android phones use screens, iOS devices can't have screens at all? It's a technology used in the development of third party software for the device that could potentially also be used for third party software development on iOS, leading to more software on iOS, and should Vulkan turn out to give more benefits than Metal, there could be software that might actually make people jump ship and switch from iOS to Android for Vulkan, let's say that a desktop Vulkan game is ported to Android or something. Now I realise the same possibility is there with Metal and that a Metal game could surface that they can't port as well with Vulkan and so on, so the argument goes both ways, but! Apple can have Vulkan, Android can't just have Metal.
 
  • Like
Reactions: MrUNIMOG
Android and iOS, most definitely. Vulkan being used by Android doesn't make it a competitor to Apple though.

In general:

Exclusive or superior games on one platform using a proprietary API that is not available on the major rivals is a significant competitive advantage and should result in more units sold. An especially strong motivator if the platform holder makes its money from *hardware* sales.

Plus a graphics stack is hard and costly to write - as we see on Mac - so supporting by two or more means diluting effort and focus. No company chooses to do this unless they are already heavily invested in multiple solutions as it makes no business sense otherwise.

Finally for developers if they can ship a common port across multiple platforms, even if it is compromised on some of them, will be attractive versus supporting multiple proprietary APIs to eke out everything on each. To drive support of a platform specific API usually requires the platform holder to give developers no choice.
 
  • Like
Reactions: MrUNIMOG
Finally for developers if they can ship a common port across multiple platforms, even if it is compromised on some of them, will be attractive versus supporting multiple proprietary APIs to eke out everything on each. To drive support of a platform specific API usually requires the platform holder to give developers no choice.

I understand the logic behind everything you say, but it is with this last paragraph here that I think the business sense in supporting Vulkan comes into play, since supporting Vulkan would mean it'd be more likely for developers to also make an iOS version of something, if they're already working on a Vulkan version for another platform.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.