-COH- and d3d 10 performance

[Gerard]

Anyone else slightly pee'd off by the performance hit this gives for very minimal (practically non existant in some cases) graphics boosts?


I done the performance test with 4xfsaa in d3d9 mode and got the following on a 3.2 core 2 quad and an evga 8800gtx ko edition:

d3d9
1920x1200 4xfsaa: Average 73.8| Max 153.6| Min 36.2 (everything maxed)


The same thing in d3d 10 with anything that had an ultra value slapped up to ultra

d3d10
1920x1200|Average 33.1|Max 68.0|Min 11.7




What exactly am i missing here in terms of performance? The only real graphical upgrade ive noticed was smoke and some explosions look a little better, yet its took away over half my average framerate, crippled my minimum framerate and took around 80fps of my max framerate.

Would making this game support multicore not have been a better idea before releasing a patch that does nothing but cripple performance for tiny visual increases?

 

[HangTime]

Possibly, but it wouldn't surprise me if MS have some kind of program whereby devs get a bit of support/finance to create DX10 enabled games.

It's also possible that maybe Relic wanted to use COH as a testing ground for DX10 tech, something to cut their teeth on without worrying too much about performance so that they could learn the ropes and then benefit from that experience when working on their next title.

Don't forget when we first starting seeing HDR in games like Far Cry, the performance hit on hardware of the time was pretty horrific too. Likewise in CoD2, DX7 mode was more than twice as fast as DX9 mode.

 

[Nullvoid]

Making the performance appear quite dismal on current dx10 cards seems like quite a good marketing strategy to me. All the enthusiasts will flock when the 2nd generation cards are released, handling everything oh so much better (for the next 6 months at least... ).

 

[Gerard]

The performance bump wouldn't annoy me if there was sopmething gob smackingly obvious with new improved enhanced graphics but really the difference is insignificant. Some textures are a little more detailed and a fe wother bits and bobs but at the cost of a massive fps drop.

This rush to half assedly tack on features so theyre able to claim their game is dx10 compliant is getting retarded.

 

[[PTG]shogun]

many of the DX10 fetures in CoH could not be handled by today's cards so you are seeing the less hungry ones


although if you look at lost planet from what i hear (and have seen benchmarks of) DX10 runs better than DX9 (although they aimed for better performance rather than looks however the DX10 version does look better )

 

[ergonomics]

many of the DX10 fetures in CoH could not be handled by today's cards so you are seeing the less hungry ones


although if you look at lost planet from what i hear (and have seen benchmarks of) DX10 runs better than DX9 (although they aimed for better performance rather than looks however the DX10 version does look better )

i personally feel the dx10 version is no different. there was a link with comparison shots, and trust me it looked no different. ill post it if i cba to find it

 

[Gerard]

i personally feel the dx10 version is no different. there was a link with comparison shots, and trust me it looked no different. ill post it if i cba to find it

Thers slight differences, some textures look somewhat sharper etc, but nothing worth the performance hit it gives.

 

[[PTG]shogun]

i saw the comparison shots and the DX10 one was better (yes it was manily lighting etc. a few subtle things and not much but just enough)

and im surprised at how it had better perfromance than DX9 (that was what they were aiming for no better looking, the devs i mean)

 

[knowledge123]

I echo the OP's comments, the DX10 version of COH looks practically no different, and the framerate is crippled by it.
Here is one of the sites that has done a DX10 comparison

and another

Benchies by extremetech

 

[LoadsaMoney]

I echo the OP's comments, the DX10 version of COH looks practically no different, and the framerate is crippled by it.

Its not supposed to, this was all said months and months ago, Dx10 just takes the load of your CPU and puts it onto your GPU so you can have more objects on screen, and run the new SM4.0 effects (which is all the graphical change is) and keep the frame-rate up as your CPU is not being used.

The reason why Dx10 performance is cack at present is because the drivers will take time to mature, they aint even got Dx9 performance sorted in Vista yet, and Dx9 games have been around for years, so how they are going to have non existant Dx10 games performance sorted in Vista when they finally appear god only knows, as it won't happen.

The cack Dx10 performance may also have something to do with the 8800's and 2900's not being powerfull enough, according to Anandtech

 

[Gerard]

Its not supposed to, this was all said months and months ago, Dx10 just takes the load of your CPU and puts it onto your GPU so you can have more objects on screen, and run the new SM4.0 effects (which is all the graphical change is) and keep the frame-rate up as your CPU is not being used.

The graphical change is not just sm4 effects, theres a lot of enchanced or new textures in the game along with more foliage at some points on the performance test.

 

[LoadsaMoney]

Its still not going to be all that different to Dx9 as everyone thinks.

 

[BlackDragon]

My latest edition of CustomPC went through a lot about DX9 and DX10, and also a big review of DX10 CoH. They believe the biggest performance hit comes from the extra light sources, and the shadows produced from these. Difficult to notice the difference unless you look closely, but its quite intensive. Try turning these down if the options are there and your performance should increase

 

[[PTG]shogun]

i saw the comparison shots and the DX10 one was better (yes it was manily lighting etc. a few subtle things and not much but just enough)

and im surprised at how it had better perfromance than DX9 (that was what they were aiming for no better looking, the devs i mean)

hmmm i was talking about lost planet there lol

 

[LoadsaMoney]

Yeah that Lost planet is a joke, i still don't get how that last comparison shot is 8fps slower in Dx10 than it is in Dx9 when its no sodding different.

 

[Gerard]

Yeah that Lost planet is a joke, i still don't get how that last comparison shot is 8fps slower in Dx10 than it is in Dx9 when its no sodding different.

Have you blown it up 1000% like most reviewers and noticed that 2 pixels are a different colour which is obviously the cause of the framerate drop?

 

[HangTime]

What's the word on Lost Planet then - in practice is DX10 faster than DX9 or vice-versa. I know 'in theory' that the DX10 mode is supposed to be faster, but is this actually the case (on Nvidia)?

 

[Margaret]

What's the word on Lost Planet then - in practice is DX10 faster than DX9 or vice-versa. I know 'in theory' that the DX10 mode is supposed to be faster, but is this actually the case (on Nvidia)?

Lost Planet is just terrible.

BioShock is said to run with more effects in DX10 while keeping the same frame rate as DX9 or something.

I look forward to seeing some benchmarks when BioShock is out.

 

[RavenUK]

What's the word on Lost Planet then - in practice is DX10 faster than DX9 or vice-versa. I know 'in theory' that the DX10 mode is supposed to be faster, but is this actually the case (on Nvidia)?

See post no9, there are benches on there for dx9 and 10.

 

[furnace]

Lost Planet is just terrible.

BioShock is said to run with more effects in DX10 while keeping the same frame rate as DX9 or something.

I look forward to seeing some benchmarks when BioShock is out.

Me too - I've always been a bit miffed as to why DX10 is slower when I thought it was supposed to be faster, hopefully Bioshock will show that DX10 has something to show for itself..!