How well does DTM Experience perform on your system?

[Billy Pilgrim]

Hi
I'd like to get some reports on how DTM Experience performs on your system.

Basically, I'm trying to figure out how well it'll perform on mine (dual GTX 780s, i7 4770K, triple screens). But I believe the game isn't optimised for SLI yet so I'm wondering what performance I might get if I use just one GPU.

What performance do you get? What settings? What FPS?

 

[Ernie]

Basically, I'm trying to figure out how well it'll perform on mine....

At the risk of repeating myself, but the best (and quickest) solution to see how it runs on your machine, is just to install R3E and try it by yourself. (DTM-E is using the same base)
Do some laps with the free content and you'll see how your system performs.

 

[Billy Pilgrim]

At the risk of repeating myself, but the best (and quickest) solution to see how it runs on your machine, is just to install R3E and try it by yourself. (DTM-E is using the same base)
Do some laps with the free content and you'll see how your system performs.

Thanks for the reply.
Yeah, you're quite right ("install R3E and try it"). It's just I haven't got my system yet, so I can't do that at the moment. I hope to have it up and running next week sometime.

 

[Bram Hengeveld]

I am using the 770 maxed out. Lovely graphics and performance so far

 

[Kurt Vanhee]

Depends on what resolution you want to race.
Had it running on Z68 mobo, 2600k proc, 16gb ram and 2 R9. 290 in crossfire on a triple screen Dell U2713hm, which gave me around 50/60 fps on low settings. Medium settings 30/35fps but for me not playable. Changed mobo and proc to a Z77 and i7 3770k giving me same result.
Like in other topic i started i am asking for improvement on crossfire mode.
Dunno if Sli is optimized.

 

[William Wester]

Running DTM at 5812 x 1080 (3 x 27"), i7 3.4/4.0 GHz, 16 GB, GTX 780 GHz OC Edition, looks and run great - with 11 AI (haven't tried more). Loving this game, sounds are sick!
I don't remember the FPS as I haven't been monitoring anymore, also not sure on graphic settings but definitely looks good (I know mostly high settings).

 

[Billy Pilgrim]

Depends on what resolution you want to race.
Had it running on Z68 mobo, 2600k proc, 16gb ram and 2 R9. 290 in crossfire on a triple screen Dell U2713hm, which gave me around 50/60 fps on low settings. Medium settings 30/35fps but for me not playable. Changed mobo and proc to a Z77 and i7 3770k giving me same result.
Like in other topic i started i am asking for improvement on crossfire mode.
Dunno if Sli is optimized.

Running DTM at 5812 x 1080 (3 x 27"), i7 3.4/4.0 GHz, 16 GB, GTX 780 GHz OC Edition, looks and run great - with 11 AI (haven't tried more). Loving this game, sounds are sick!
I don't remember the FPS as I haven't been monitoring anymore, also not sure on graphic settings but definitely looks good (I know mostly high settings).

Looking at Kurt Vanhee's system's performance and what FuzzyFassbender's getting, I can't help wondering if Kurt would be better off using just one GPU.
As far as I can tell, the 780 and the R9 290 are almost exactly as powerful as each other

Benchmark here:
http://us.hardware.info/reviews/495...0-bioshock-infinite---5760x1080-+-frame-times

 

[Kurt Vanhee]

Looking at Kurt Vanhee's system's performance and what FuzzyFassbender's getting, I can't help wondering if Kurt would be better off using just one GPU.
As far as I can tell, the 780 and the R9 290 are almost exactly as powerful as each other

Benchmark here:
http://us.hardware.info/reviews/4954/9/amd-radeon-r9-290-review-cheaper-hawaii-vs-gtx-780-bioshock-infinite---5760x1080- -frame-times

Can confirm this, no fps boost\more in crossfire.
Have a thread about crossfire.
Like said there the more memory your card has the more fps.
Therefore i am waiting for AMD to release the 8gb card

 

[Billy Pilgrim]

Can confirm this, no fps boost\more in crossfire.
Have a thread about crossfire.
Like said there the more memory your card has the more fps.
Therefore i am waiting for AMD to release the 8gb card

I'm not so sure that more VRAM will help.
Check this out - what do you think?
http://www.guru3d.com/articles_pages/geforce_gtx_780_sli_review,23.html
At the link, you've got GTX Titan sli vs GTX 780 sli (6Gb vs 3 Gb) on three monitors. And there's hardly any performance difference.

 

[William Wester]

Ok, took a little time today to evaluate various graphical settings to get the best triple screen performance with my setup. I didn't exhaustively try every setting combination but in general I found what works very well for me.

Hardware: i7-3770 3.4-4.0 GHz, Gigabyte GTX780 factory Core Clock: 1019 MHz (Base) 1071 MHz (Boost) - Memory Clock: 6008 MHz (GV-N780GHZ-3GD), 16 GM RAM, 5812 x 1080

With the default "HIGH" graphics setting option, I was getting 56-78 FPS, GPU 90-98%, 1905 MB GPU memory consumed. Not bad but I wanted to tune for 60 FPS at all times.

I then switched to the "CUSTOM" graphics setting option and tried various adjustments from the HIGH settings before coming up with the below (best bang for resource consumption).

Ultimately, the only thing I needed to change from the default HIGH settings to make a large FPS improvement was "Reflection Type: Simple Cube Map" (instead of Reduced). This provided 80-94 FPS, GPU 90-98%, 1874 MB GPU memory consumed. And to me I couldn't tell the difference - looks great! After finishing with all testing I switched VSYNC on, this smoothed things out quite a bit. With VSYNC the GPU was running at 65-75 capacity - a steady 60 FPS.

All testing done on Red Bull with 11 AI.

I have very happy with the performance of DTM with my single GTX 780 and triple screen.

 

[Blkout]

Ok, took a little time today to evaluate various graphical settings to get the best triple screen performance with my setup. I didn't exhaustively try every setting combination but in general I found what works very well for me.

Hardware: i7-3770 3.4-4.0 GHz, Gigabyte GTX780 factory Core Clock: 1019 MHz (Base) 1071 MHz (Boost) - Memory Clock: 6008 MHz (GV-N780GHZ-3GD), 16 GM RAM, 5812 x 1080

With the default "HIGH" graphics setting option, I was getting 56-78 FPS, GPU 90-98%, 1905 MB GPU memory consumed. Not bad but I wanted to tune for 60 FPS at all times.

I then switched to the "CUSTOM" graphics setting option and tried various adjustments from the HIGH settings before coming up with the below (best bang for resource consumption).

Ultimately, the only thing I needed to change from the default HIGH settings to make a large FPS improvement was "Reflection Type: Simple Cube Map" (instead of Reduced). This provided 80-94 FPS, GPU 90-98%, 1874 MB GPU memory consumed. And to me I couldn't tell the difference - looks great! After finishing with all testing I switched VSYNC on, this smoothed things out quite a bit. With VSYNC the GPU was running at 65-75 capacity - a steady 60 FPS.

All testing done on Red Bull with 11 AI.

I have very happy with the performance of DTM with my single GTX 780 and triple screen.

Agreed, the one setting that makes the most impact is reflection type. Simple Cube Map will save a good bit of performance from reduced. I run everything else at the max setting and get anywhere from 60-80fps at 2560x1600 resolution with a grid of 11 depending on the track and using the cockpit view. Using the hood view, I get a boost in framerates of about 10-15fps above the cockpit view. This is using an Intel I5 2500k at 4.4GHz and dual GTX 780's in SLI, however SLI doesn't work in this game so I'm using a single 780 at 1228MHz.

 

[Travis]

I'd posted this info once before but in another topic. The following is a breakdown of the reflection settings from SimBin's art director:

• Simple reflections: show the sky and blends image based lighting data as reflections. This offers very good on performance.
• Reduced: reflection are in real time however the sim renders objects that are selected for being visible in the mirrors and 512 resolution for each cube side.
• Full: Realtime reflections that reflects everything except cars.
• Reduced 1k: Reduced: reflection are in real time however the sim renders objects that are selected for being visible in the mirrors. 1K resolution for each cube side.
• Full 1k: Realtime reflections and reflects everything except cars at 1K resoultion. Performance takes a hit (I think it's around 20% more taxing than the 512 resolution) at this resolution as it's meant for high end PC builds (e.g. multi-GPU setups).

As for how the sim runs; v-sync'd at 60fps with a full grid using a EVGA GTX 680 at 1920x1080 - not having a mutli-screen setup helps.

I have custom graphics settings, pretty much everything is set to its highest level except for the reflections which remain at their default setting, anti-aliasing is set at 4xMSAA and I have turned off motion blur & FXAA. One can gain back some fps by turning off split shadows too.

 

[Blkout]

I'd posted this info once before but in another topic. The following is a breakdown of the reflection settings from SimBin's art director:

• Simple reflections: show the sky and blends image based lighting data as reflections. This offers very good on performance.
• Reduced: reflection are in real time however the sim renders objects that are selected for being visible in the mirrors and 512 resolution for each cube side.
• Full: Realtime reflections that reflects everything except cars.
• Reduced 1k: Reduced: reflection are in real time however the sim renders objects that are selected for being visible in the mirrors. 1K resolution for each cube side.
• Full 1k: Realtime reflections and reflects everything except cars at 1K resoultion. Performance takes a hit (I think it's around 20% more taxing than the 512 resolution) at this resolution as it's meant for high end PC builds (e.g. multi-GPU setups).

As for how the sim runs; v-sync'd at 60fps with a full grid using a EVGA GTX 680 at 1920x1080 - not having a mutli-screen setup helps.

I have custom graphics settings, pretty much everything is set to its highest level except for the reflections which remain at their default setting, anti-aliasing is set at 4xMSAA and I have turned off motion blur & FXAA. One can gain back some fps by turning off split shadows too.

Any idea what split shadows are?

 

[Travis]

Any idea what split shadows are?

Hi,

When selected, Shadow Split results in shadows being rendered in two passes. This means high quality shadows are being rendered near the camera. When set to off, shadows near the camera won't be of high quality hence the frame rate being better.

 

[Blkout]

Hi,

When selected, Shadow Split results in shadows being rendered in two passes. This means high quality shadows are being rendered near the camera. When set to off, shadows near the camera won't be of high quality hence the frame rate being better.

Ahh OK, that's the same as the two pass shadow option in iRacing.

 

[Blkout]

For anyone interested in improving their performance in DTME and R3E, the biggest performance offenders is Reflection Type, changing this one to cube map gives a big fps boost from the reduced setting. This one does affect image quality however, if you need more performance, it may be something you just have to live with. The remaining two setting are not as bad but certainly can be tweaked for better performance. Change Shadow Split to no and Motion Blur Type to fast.
Between these settings, its possible to double your frame rate.

 

[Peter]

Those are not the 2 biggest performance offenders. There is also shadow split and fxaa that take away lots of fps.
http://www.racedepartment.com/forum/threads/hw-requirements-cpu-intensive.80150/#post-1610429

 

[Blkout]

Those are not the 2 biggest performance offenders. There is also shadow split and fxaa that take away lots of fps.
http://www.racedepartment.com/forum/threads/hw-requirements-cpu-intensive.80150/#post-1610429

Based on my own testing of every single setting in the game, those are the worst, shadow split and complex motion blur are the next worst behind both of those.

Load a single race, choose Moscow Raceway and monitor your fps, so far that has been the most demanding track graphically that I've tested. Start at the medium performance setting in the graphics menu and slowly increase one setting at a time and watch your performance, you will see the same as what I stated. Who uses FXAA if you can get away with using MSAA? FXAA was designed to be a low performance impact form of AA when MSAA is too demanding.

 

[Travis]

FXAA should make little to no difference to the frame rate as it was designed to be a high performance, low resource cost anti-aliasing solution. As ATI state on their website:

"Fast approXimated Anti-Aliasing or FXAA is a post-processing technique that removes jaggies using a very memory-efficient, fast and light-weight pixel shader developed by Timothy Lottes"

The problem I have with FXAA is that is blurs the image way too much. I've tried injecting SMAA (http://mrhaandi.blogspot.com.au/p/injectsmaa.html) into R3E without much success - it affects the tyre smoke in a negative way.

 

[Billy Pilgrim]

Well, I can now answer my (the OP's) question as I've at last got my PC.

I've got two GTX 780s, and an i7 4770k, on triples.

If I put up the settings to max/ultra, I get poor framerates. Strictly speaking I think I haven't actually tried at 'ultra'. I knew it wouldn't get good results. I suppose at ultra I'd get around the 40s.

Anyway, I've got mostly everything on high, except I've turned down reflections to 'simple cube', and shadows are on medium, split shadows are off (AA - I don't remember). The game's looking pretty good at those settings though (but not as good as GSC's looking on my system) and I'm getting 58 to 75 fps, which is nicely playable.

I guess it'd be performing way better if it was sli optimised. One of the developers, on an R3E thread, did say that it would be sli optimised although he didn't say when.

Well, here's to the R3E/DTM developers - great job, lovely game(s)! Keep up the great work. And please put that sli optimisation on the priority to-do list.