I don't think you'll see crysis-level games in terms graphical focus unless someone was trying to develop the next benchmark game. (like bf3 for example)
My current project is actually extremely CPU-dependent, and even splitting it into 4 even threads with optimizations up the ass it can run slow on my min spec cpu (Q6600). I don't know if any other devs have similar projects under covers, but it wouldn't surprise me if i wasn't the only person who was developing a game that uses a ridiculous amount of resources.
I don't think you'll see crysis-level games in terms graphical focus unless someone was trying to develop the next benchmark game. (like bf3 for example)
BF3 runs at the max setting on my laptop. That game was mostly a console port. Crysis 1 on the other hand was a game that had its max settings configured for the future which is nice and lets the game age well when people play it a few years later.
It would be nice if we started to see that again where low settings is the default then medium and high are made for future computers. Actually supporting DX11 for instance. There is a cost involved so that may be why we stopped seeing it.
BF3 runs at the max setting on my laptop. That game was mostly a console port.
I believe the guys at DICE would disagree strongly with you there. The PC graphics are much nicer than console, although designed to run "today" rather than "the future". A number of major games today are supporting and leveraging DX11 as well.
SlimDX | Ventspace Blog | Twitter | Diverse teams make better games. I am currently hiring capable C++ engine developers in Baltimore, MD.
Most games are developed to run on both consoles and PC. A few games do use current hardware but they are rare since using older hardware allows their games to run on more PCs. Once the Xbox 720 and other next generation consoles are released I'm sure we will see many more games which take full advantage of modern hardware.
Firstly console games sell better in the majority of cases than the PC version. Secondly there is target market; take a look at the Steam hardware survey which, for the most part, is a 'gamer' market and you'll see the problem. The hardware of most game machines isn't that advanced which is a chicken and egg problem; no one has the hardware so no one takes advantage of it so no one buys the hardware.
Vista dying under a storm of FUD didn't help things and probably delayed forward motion by a year or more (taking DX10 with it), it's only within the last maybe 6 months that targetting Win7/DX11 API instead of XP/DX9 was viable but with multi-million dollar games with development cycles of a few years most people aren't going to bet on a Win7 style uptake, not after Vista so most people played it safe. Hell it's only this month that we are offically killing DX9 support in our engine at work (even if it has been left to rot since around October of last year, however at that point we were actively working on it still to keep it up to date feature wise).
Asset creation also enters into the fray; creating assets for large DX11 feature games is going to take more time and effort, as pointed out most games sell better on console where lower quality assets could be used. Asset creation is also the most expensive part of any game development cycle (outside of marketing of course) so 'wasting' money on assets very few are going to see isn't within a company's best intrests.
When it comes to the PC side of things development wise if you target a game on the consoles then most PCs will have the grunt to run it so the PC version requires less effort as it tends to Just Work. (Well, if you ignore driver issues and hardware compatibility problems with older cards... *flashbacks*)
The shift towards DX11 now is easy to explain imo; practise.
Aside from any home projects the majority of people in the industry have been working with the DX9 API/feature level day in, day out for the past 10 years or so, figuring out deployment and development of a DX11 engine isn't an easy thing and with the open secret that the next generation is going to be at least DX11 feature level hardware it makes sense to get a bit of a jump start on things before the hardware hits to figure out the quirks as best you can. Granted this also causes some cool tech and a nice looking game to turn up to but from an economics point of view the ROI is just as much tech learning as it is game development.
Of course this has also brought up the Elephant In The Room of PC gaming which is the overhead on the PC and driver issues which mean that under heavy usage even rendering from multiple threads to command lists (which AMD still doesn't properly support even on this 7970 series) doesn't gain you a huge amount as you end up bumping into driver limits/interface problems (DICE talk about this a bit). Basically your top of the line PC, running the latest OS and graphics card with the best CPU you can buy is good for ~15,000 draw calls per frame.
(We've an insane test case which draws 50,000 flat shaded cubes across N threads with one draw call per cube; X360 and PS3 can run at 60fps, no problem, a Win7 64bit i7 running 8 threads@ 2.6Ghz with a NV 470GTX card runs the same test at ~28fps having crashed into the driver/CPU bottleneck. Drop the draw calls to 15,000 and you hit 60fps, increase them and you'll start losing more and more time in the buffer swap in the driver. Yes, it's a totally over the top CPU bound problem but it's still a problem. Only thing which saves the PC right now is that console GPUs are so old.)
If you look at the sales chart for any big, AAA game, they sell around 90% of all the copies they'll ever sell in the first 3 months, and most of those are in the first few weeks -- If you launch with too high requirements, you'll essentially blow off the top of your peak sales period. You'd want to aim to look really good on the higher-end mainstream GPUs -- the ones that cost around $200 in the 3-6 months before your launch window.
In other words, why go through all the trouble of making an insane game just so that all of 12 people will be able to play it the way it was meant to be played at launch.
^ This. It is poor economics not to try and maximise the market into which you are selling your product i.e. the vast majority of players are unlikely to have the absolute top end system out at any time.
In other words, why go through all the trouble of making an insane game just so that all of 12 people will be able to play it the way it was meant to be played at launch.
I think the problem I have is they release with effectively one graphics setting in mind. It's painfully obvious when you see a console port and it has no graphics options in the menu. The nice thing about PC games back in the day is I'd buy them and play them on low settings. Then when I upgraded and installed the game again I'd notice I can play it on max settings and it breaths a breath of fresh air into the game. Current games have stopped doing that for the most part. Skyrim was mentioned in the original post. That game had very low system requirements. They didn't add SSAO in the core game. (They released a video a month after release showing off SSAO and some awesome graphical enhancements which was odd to see). Basically they released only what would run on consoles and medium end computers at the time which seems to be the norm now.
I remember playing Crysis 1 on low on my laptop and beating it. Awesome game. Then I went over to my friend's house and he had dual 8800 GTX cards and was playing at almost max graphics. Totally different game visually. It was like looking into the future of gaming.
Of course this has also brought up the Elephant In The Room of PC gaming which is the overhead on the PC and driver issues which mean that under heavy usage even rendering from multiple threads to command lists (which AMD still doesn't properly support even on this 7970 series) doesn't gain you a huge amount as you end up bumping into driver limits/interface problems (DICE talk about this a bit). Basically your top of the line PC, running the latest OS and graphics card with the best CPU you can buy is good for ~15,000 draw calls per frame.
Don't suppose you have a link to the article/presentation with the DICE talk on this?... Would be quite interesting to learn a thing or two about this sort of thing.
)
