Crysis 2 PC edition OT

They're still a hell of a lot better than they would have been as a Microsoft dev deliberately cancelling the PC version to help xbox sales.

Besides, the latest patch puts them back up with the better PC devs. Poor timing not withstanding.

 

Yeah, they're not buggy as such, just limited by the format The picture I posted was really just an ideal situation - it doesn't work nearly as well in reality.

They don't, but only because it's not possible to do that in a post process. They can diffuse the reflection using the bumpmaps because AFAIK they're able to retrieve that information from the image

 

Wouldn't simple pre-filtering do the trick?
Basically you would have case of cone tracing a mip mapped heightfield.

 

Crysis 2 DX11, performances et tessellation à la loupe

Great article. Great work, Crytek: http://www.hardware.fr/articles/838-6/tessellation-loupe.html

 

tess factor 8 seems sweetspot..

EDIT1:
waste of triangles lol



Dx11 Extreme - Ultra -Wireframe
beauty comes from POM


EDIT2:

 

Even the 560Ti is beating the 6970

 

I don't understand what they're doing with Tessellation. Did they really apply stupid amounts of triangles on that path curb for no benefit? What is that scene with the load of triangles seemingly floating in air? is this a case of "apply massive tessellation in whatever we want with no regard to actual benefit or performance"?

 

Sometimes it seems they've used it when there's any kind of groove on a surface..can sometimes look pretty crappy:

http://i.imgur.com/F20Gk.jpg

It is used amazingly for the water though.

 

Can we influence the amount of tesselation in the game config?

 

On ATI cards you can do it easily via the CAT control Panel.

 

That's NVIDIA's: The Way It's Meant To Be Coded

 

EA clarifies, confuses stance on platform exclusives

http://techreport.com/discussions.x/21252

 

Whoah. They believe in absolute freedom of choice for customers...except for when they choose not to? What a crock of shit!

Yeah, typical EA asshattery. Screw them and the horse they rode in on.

 

And is Valve better? Any of their games dont appear on others DD services, talk about freedom of choice.

BTW why Crysis 2 was removed from Steam when Dragon Age 1/2 and Mass Effect 2 are still there? Both have DLC's that arent sold on steam and You cant buy Bioware Points either.

 

Just finished the game. What a ride. Loved the suit and the presentation was frequently out of this world (all cranked up to Ultra of course). Definitely one of the best games I played this year.

 

Apologies for being off-topic but I installed the 990X and this is what I got

 

Yeah. Windows doesn't rate anything above 7.9, and it takes the lowest scored item as your final score. It also gets things a bit wrong (it will for instance score RAID1 at a lower speed than single disk, even though in real world use RAID1 is faster).

There was some article a few months back about a guy attempting to build the highest MS scored machine, and he was using things like RAID5 SSD and quad, hex-cored CPUs and GPUs and other nonsense to do it.

So really, nothing to worry about, your scores are great except for the disk, which is just a standard disk score. You'll have to go RAID or SSD to improve much beyond the single disk score.

 

Well that's a fair bit faster than me! My 2500K only gives me 7.5 while I get 7.6 on memory with 4GB DDR3 1600Mhz and and 7.6 in both graphics with an HD4890 1GB. My HDD drags the whole thing down though with a puny 5.7.

Back on topic:

I was thinking about the pre-set profile settings for Crysis 2 and developers general worry about labelling them in such a way that PC gamers don't get up themselves when their PC can't handle the highest details.

Specifically, if Crytek had released C2 with the Ultra settings enabled from the get go, I can imagine some major kick back with regards to the performance, especially on AMD cards. So what could they do in terms of labelling settings that would overcome this?

The way I see it, a simple description of each preset level is fine as long as it's done correctly. Something tongue in cheek like the difficulty settings from Doom would be good. In particular they should highlight where the consoles sit on those pre-sets since half of the anger about poor performance down to how the users PC and/or the graphics compare to consoles. Something along these lines:

Gamer: Console graphics settings, your PC may be steam or coal powered
Advanced: For PC's with a little extra magic gaming sauce
Extreme: If you don't have a high end gaming rig, you're not invited to this party
Ultra: Only nuclear powered uber rigs that make all other PC's tremble with fear need apply

Something along those lines should placate the majority of gamers who may complain about poor performance at whatever setting, especially if comparisons to consoles are part of the motivation for that complaint. Hell if they'd have done something like that they might have even got away with not having the more detailed graphics options in the first place.

 

Which is the harbour benchmark? I can't seem to find that one. I did start the GPU benchmark in the bin32 folder though and was able to stay above 60fps 95% of the time with lowered graphics settings (while keeping everything that might effect CPU performance at Very High. That's on a stock 2500K. I'm betting the bits where it dropped below 60fps were GPU limited too.

I tried that one and lowering the graphics options had a hige effect for me. Again, keeping physics, sound, particles etc.. at very high as they may imapct CPU performance. There is one small section where the benchmark drops into the teens but for the majority is between 40 and 50 fps.

Interestingly though, resource monitor shows my CPU at around 50% usage throughout the benchmark but core 4 is near 100% for the duration. It seems most likely that Crytek have physics constrained to a single core while other things run on the other cores (my other 3 cores where all around 25% - 35% utilised. Better splitting of that thread, whether its' physics or something else should allow the CPU to push this game event further.

 

Yeah I don't really care for the windows scores all that much other than for bragging rights haha. I am running 2 x velociraptors in raid 0 and I get a 6.2 for that. SSD is definitely something I am considering but that will be a Q4 purchase at least.

Back on topic, if a 4 core is maxed out 100% for the duration of the benchmark are there results showing how a 6 core cpu fairs in those benchmarks? I wonder if Crytek's new engine is able to max out all cores or not by dividing up the workload in as many threads as there are cores and what would happen if Hyperthreading were to be turned on!