[Official] DX9 vs DX11 performance difference

Status
Not open for further replies.
BlackGoku said:
Hello, after the updates of 5 May in the DX11 I thought that I was going to get more FPS but my game is still unplayable and has ridiculous FPS compared with DX9.

GPU: R9 280 3 GB VRAM
CPU: A6-3650 overclocked to 3,9 ghz
RAM: 12 GB
OS: Windows 10 Pro x64
Controls: Logitech G27
Click to expand...

Your CPU is a bit unusual, it's an AMD APU (Accelerated Processing Unit), which has an embedded GPU. This type of CPU is usually run without a dedicated GPU in a HTPC or media streaming system. My guess is that the game is running on the APU instead of your dedicated R9 280 GPU, which would explain the huge drop in FPS. Probably none in testing has this type of CPU, so it might have been overlooked.
 
stonec said:
Your CPU is a bit unusual, it's an AMD APU (Accelerated Processing Unit), which has an embedded GPU. This type of CPU is usually run without a dedicated GPU in a HTPC or media streaming system. My guess is that the game is running on the APU instead of your dedicated R9 280 GPU, which would explain the huge drop in FPS. Probably none in testing has this type of CPU, so it might have been overlooked.
Click to expand...

You are wrong... The game is just using the R9 280 and I went from 60 fps with v-sync Full graphics with DX9 to 15-20 fps with DX11 without v-sync. And if we take care that DX9 without v-sync is getting me more than 100 fps... then we can say that the game has lost about 80 fps from DX9 to DX11.
Thats all.
 
Those guys benchmarking with Nvidia cards, you might want to try setting "threaded optimization" from CP/Inspector to off from the default auto. I got quite significant performance boost and more stable frame times that way. Performance isn't still at DX9 level, but much improved and I can hit quite steadily 80-90 fps on Nola Indy race with 25 AI Nissans when TO is set to off. This is most settings turned max which are the same ingame settings I used with the DX9 branch (shadows high, blur optimal and new post processing effects medium). If using DX11 compatible stuff, game is very playable in this way, for me at least.

This wouldn't be the first time threaded optimization can cause poor performance and/or stuttering, granted games affected are mostly older titles. I still think that at least one of the reasons for rF2 lack of performance is due to threading of rendererer data.

I have i7-3930K@4.5GHz, 16GBDDR3@1600, GTX980ti (boosting 1510MHz), 3440x1440@100Hz G-sync.
 
TheGrunt said:
Those guys benchmarking with Nvidia cards, you might want to try setting "threaded optimization" from CP/Inspector to off from the default auto. I got quite significant performance boost and more stable frame times that way. Performance isn't still at DX9 level, but much improved and I can hit quite steadily 80-90 fps on Nola Indy race with 25 AI Nissans when TO is set to off. This is most settings turned max which are the same ingame settings I used with the DX9 branch (shadows high, blur optimal and new post processing effects medium). If using DX11 compatible stuff, game is very playable in this way, for me at least.

This wouldn't be the first time threaded optimization can cause poor performance and/or stuttering, granted games affected are mostly older titles. I still think that at least one of the reasons for rF2 lack of performance is due to threading of rendererer data.

I have i7-3930K@4.5GHz, 16GBDDR3@1600, GTX980ti (boosting 1510MHz), 3440x1440@100Hz G-sync.
Click to expand...
I just tested the threaded optimization before reading your comment but for me personally i saw no difference. But it could be something to test for sure since it can benefit older titles when off.
 
TheGrunt said:
Those guys benchmarking with Nvidia cards, you might want to try setting "threaded optimization" from CP/Inspector to off from the default auto. I got quite significant performance boost and more stable frame times that way. Performance isn't still at DX9 level, but much improved and I can hit quite steadily 80-90 fps on Nola Indy race with 25 AI Nissans when TO is set to off. This is most settings turned max which are the same ingame settings I used with the DX9 branch (shadows high, blur optimal and new post processing effects medium). If using DX11 compatible stuff, game is very playable in this way, for me at least.

This wouldn't be the first time threaded optimization can cause poor performance and/or stuttering, granted games affected are mostly older titles. I still think that at least one of the reasons for rF2 lack of performance is due to threading of rendererer data.

I have i7-3930K@4.5GHz, 16GBDDR3@1600, GTX980ti (boosting 1510MHz), 3440x1440@100Hz G-sync.
Click to expand...

http://i.imgur.com/FwcfwlO.png My results.
 
upload_2017-5-6_12-6-44.png

Improvement with latest Buildenabling aghgressive threading. Full Screen issues persist.

My comments on how this transition from DX9 to DX11 is being handled
This Open Beta seems to be changing very fast. It is almost impossible to keep up to track.
If one intermediate version that you miss would have solved something you never know.

Me and others have already reported that the way the graphics card is being recognized is very strange and should be addresed prior to trying to fix any other thing as it could be causing all the rest of the problems.
I feel absolutely ignored regarding this item to which I haven't heard an official statement yet.
 

Attachments

SPASKIS said:
View attachment 7403
Improvement with latest Buildenabling aghgressive threading. Full Screen issues persist.

My comments on how this transition from DX9 to DX11 is being handled
This Open Beta seems to be changing very fast. It is almost impossible to keep up to track.
If one intermediate version that you miss would have solved something you never know.

Me and others have already reported that the way the graphics card is being recognized is very strange and should be addresed prior to trying to fix any other thing as it could be causing all the rest of the problems.
I feel absolutely ignored regarding this item to which I haven't heard an official statement yet.
Click to expand...

Thanks for your help SPASKIS, for sure the team is aware on what you're posting, maybe they're too busy to answer everybody ;)
 
Clouds definitely hurting fps in build 1803126. One core of CPU maxes out and GPU utilization drops to about 50%-60%.

settings: Level 4 AA, PPE=off, resolution 5760x1080 multiview, no sync, aggressive threads=off (on was about 4% worse)
system: single GTX 1080, 8 GB VRAM; i7-4770 @3.4 GHz, 32 GB RAM
circuit: NOLA-A
car: GTR 500
view: cockpit camera, center virtual mirror

DX9 sunny, Avg 114 fps, Min 105 fps, Max 122 fps
DX11 sunny, Avg 114 fps, Min 99 fps (-6%), Max 127 fps (+4%)
note: higher variance for DX11 despite same average fps as DX9

DX9 overcast, Avg 106 fps, Min 98 fps, Max 114 fps
DX11 overcast, Avg 53 fps (-50%), Min 49 fps (-49%), Max 58 fps (-50%)

DX9 rainy, Avg 102 fps, Min 95 fps, Max 109 fps
DX11 rainy, Avg 48 fps (-53%), Min 44 fps (-54%), Max 53 fps (-51%)

Since rainy appears to include the same clouds as overcast, I'll only attach files for sunny & overcast.
 

Attachments

Last edited:
Emery said:
Clouds definitely hurting fps in build 1803126. One core of CPU maxes out and GPU utilization drops to about 50%-60%.

settings: Level 4 AA, PPE=off, resolution 5760x1080 multiview, no sync, aggressive threads=off (on was about 4% worse)
system: single GTX 1080, 8 GB VRAM; i7-4770 @3.4 GHz, 32 GB RAM
circuit: NOLA-A
car: GTR 500
view: cockpit camera, center virtual mirror

DX9 sunny, Avg 114 fps, Min 105 fps, Max 122 fps
DX11 sunny, Avg 114 fps, Min 99 fps (-6%), Max 127 fps (+4%)
note: higher variance for DX11 despite same average fps as DX9

DX9 overcast, Avg 106 fps, Min 98 fps, Max 114 fps
DX11 overcast, Avg 53 fps (-50%), Min 49 fps (-49%), Max 58 fps (-50%)

DX9 rainy, Avg 102 fps, Min 95 fps, Max 109 fps
DX11 rainy, Avg 48 fps (-53%), Min 44 fps (-54%), Max 53 fps (-51%)

Since rainy appears to include the same clouds as overcast, I'll only attach files for sunny & overcast.
Click to expand...
This agrees with my findings too. I think the team are well aware and working on it
 
Anyone else toggled through the various camera options (cockpit, trackside etc.)?
The massive variation in FPS depending on which camera you're viewing is enormous.
Trackside cams are way down on FPS, while certain cams when pressing the 'Home' button
have the best rates???
 
When disabling crimson settings and relying only on the game settings major performance loss.
Track: Silverstone
Car: Nissan GT500
Resolution: 1920 * 1080, All to Max, default FOV
AA: Level 8 in DX9, Level 5 in DX11 (no PP)
DX11 50% less than DX9
DX9
upload_2017-5-7_16-57-6.png

DX11
upload_2017-5-7_16-57-58.png
 

Attachments

Marcel Offermans said:
We are reading all the feedback, but like others have mentioned, cannot respond to every post individually.
Click to expand...
I understand that Marcel. However, it is not me individually who is having this problem. I have already read several similar cases. That's the reason why I was expecting some type of statement confirming the problem.

Reading about vsync being next problem to solve without even mentioning this problem as a known issue was quite frustrating.
 
SPASKIS said:
I understand that Marcel. However, it is not me individually who is having this problem. I have already read several similar cases. That's the reason why I was expecting some type of statement confirming the problem.

Reading about vsync being next problem to solve without even mentioning this problem as a known issue was quite frustrating.
Click to expand...

I think they have classed the issue with clouds under 'performance' which they have said they are working on as the priority and vsync will be later.
 
SPASKIS said:
Reading about vsync being next problem to solve without even mentioning this problem as a known issue was quite frustrating.
Click to expand...

The simple fact they read your report and the simple fact they will investigate all the reports should be a great result for you, no need to be frustrated in my opinion. Give them the time to check all the reported bugs and if your find is correct they will solve it.
 
DaVeX said:
The simple fact they read your report and the simple fact they will investigate all the reports should be a great result for you, no need to be frustrated in my opinion.
Click to expand...
I disagree on that.

That is the minimum I would expect after dedicating many hours already to freely betatesting the product that they are making business with.
 
Last edited:
Status
Not open for further replies.
Back
Top