My Simple Benchmark

Right now I'm quite happy with my current basic setup. I'm testing this using 1960s vehicles at Malaysia Sepang Full Loop. I can get 35 FPS on start, and I was at the back of the grid, with 20 visible vehicle and 19 AIs.

Game Setting:
1680x1050, 32-bit, Full Screen, VSync Off
HDR Off, in-game anti-aliasing Off
Circuit Detail High, Texture Filter Bilinear, Shadows Medium, Shadows Blur off, Enviro Reflection Off, Wind Off, other details are at Max, Yes, or On position. Visible Vehicles 20, Steering Wheel On, Default View Cockpit. Audio 20.

ATI Catalyst Setting:
Anti Aliasing: 4x Box Filtered, Adaptive Multi-sample AA
Anisotropic: 2x
Mipmap Detail: High Quality
Catalyst AI: High Quality, with Enable Surface Format Optimization

FPS Min: 30, FPS Max: 55, FPS Avg: 43

 

thanks for the info

 

Thanks, good useful info. Using your settings I get very similar FPS @ Malaysia and 35 - 43 @ Spa. Those frame rates are very playable. So good so far. Small complaint about "flickering" on some objects, like the windows on buildings, but good for a start.

CPU: Intel i7 2600s @ 2.8GHz
Motherboard: Pegatron Carmel
Memory: 8GB DDR3
OS: 7 Home Premium 64bit
Video: ATI Radeon HD6570 1GB GDDR5

 

Here's another "basic" setup that works quite well too (in my system at least).

This time, I'm using ATI Tray Tools (ATT) instead of ATI Catalyst to control the display driver. Oh one more: No post processing tool such as FXAA or ENBseries were used.

The in-game settings are still the same as the above setup.

Game Screen Resolution:
1680x1050, 32-bit, Full Screen, VSync Off, cockpit view, steering animation

Driver setting:
2xAA, Box Filter, Super Sample, EQAA checked
16xAF, Quality with Trilinear, High Quality AF checked
Texture Preference High Quality
Mipmap Detail Level High Quality
Catalyst AI High Quality, Surface Format Optimization's enabled
Texture LOD Adjustment 10.0
Flip Queue Size 0
Leave other settings in ATT at their default value.

FPS Min: 24, FPS Max: 61, FPS Avg: 42.5, FPS at start: 31


@mel1c to reduce flickering, you may use higher AA or change the AA method to Super Sampling (FPS is reduced though). Glad that you like my simple benchmark, thanks!

 

Taufikp, tried super sampling already, reduced my frame rate by about half. I'll look at your recent settings

 

This time, I'm testing rF2 in Windowed mode. Some said that they got high FPS boost in windowed mode, some even claimed it doubles their frame rate.

Everything is the same as the more recent setup, the only difference is, I ran this test in Windowed mode.

So, this is my test result:
FPS Min: 24, FPS Max: 62, FPS Avg: 43, FPS at start: 32.

Just a reminder, here's the full screen test result:
FPS Min: 24, FPS Max: 61, FPS Avg: 42.5, FPS at start: 31.

Well, in my system at least, the difference is very small and insignificant, in terms of frame rate/performance.

However, I see a huge difference visually.

When running rFactor 2 in windowed mode, the image clarity is superb. No more blurry engine lines at the back of those 1960s Formula cars. Engine details are rendered with such a clear and fine lines. Very high fidelity. In full screen mode, the engine lines are blurry and its fine details are lost, as if rFactor 2 being rendered at resolution lower than 1680x1050.

This is weird, because I don't get this phenomenon with iRacing, netKar Pro, GTR Evolution, and FIFA 12. And all of them are running at an even lower resolution: 1280x720 full screen (running rFactor 2 in 1280x720 full screen is, uh, an unpleasant experience to the eyes).

So why rFactor 2 give me this blurry images when running full screen, while the other games mentioned above are perfectly sharp and clear at even lower screen resolution?

Let's hope ISI can sort this out in the final release.

In the mean time, I'll run rFactor 2 in windowed mode only, maybe even at desktop resolution (1920x1080), because the average frame rate at this resolution is 37.5 FPS, which is still acceptable in my book. I really can't go back to the full screen mode after seeing the super clear rFactor 2 in windowed mode.



Update:
Turns out I don't need to run at 1920x1080 windowed mode. The frame rate of 1080p full screen in fact much better than windowed mode, averaging at 41 FPS. Image fidelity at 1080p full screen is very good, equals to 1050p windowed mode. Still, what troubled me the most is why at 720p full screen rFactor 2 loose all that clarity while other games (listed above) don't.

 

Very Nice- But There is really no need to Overclock your cpu for rfactor when you have a Quad core.

 

My recent setup (above) running at 1920x1080 full screen works very well. rFactor 2 now looking sharp and clear. To simulate the HDR, I'm using FXAA Level 1 (read my previous post -in this thread- about how I named my FXAA setup). The frame-rate (FPS) performance is also quite stable. As I said in my post above, averaging at 40-41 FPS.

To minimize stuttering, I set the Replay Fidelity to 'Minimum' (while I let the length at its default value: 20 seconds). I got this tip from days of F1 Challenge '99 modding in RSC Forum!

Now, with Replay Fidelity set to minimum, I still get stutter, but in very, very short time. Usually I'll get stutter at T1 and T2 in Sepang, but now the stutter only happen at T1, and it doesn't last as long as before.

All in all, I'm quite satisfied with my rFactor 2 setup now.

 

"FPS Min: xxxx, FPS Max: xxxx, FPS Avg: xxxx" I think you getting those numbers from mod mode... If using mod mode, well thats not the optimal case. The optimal case is the normal single/multiplayer exe and kinda why there are two versions of the executable.

In mod mode we do extra work that is not needed in the normal environment of single/multiplayer

The single/multiplayer exe does not show FPS Min/Max is why I assume you are using mod mode

 

Well, I haven't touch the mod mode at all.

I got the FPS min, max, and avg using Fraps 60 seconds benchmark, and I ran the same replay over and over again through all different setups listed here. After doing a benchmark of that replay with particular setup, I'll do a 15 laps race just to 'feel' the result in real time mode.

I fall in love with rF2 all over again now.

 

A user in NoGripRacing forum reported that changing his screen from 1920x1080 60Hz to 1920x1080 59Hz making his rFactor 2 graphics look sharper and clean.

I tested this and can't reproduce it. My rFactor 2 at 59Hz still as sharp as the 60Hz screen.

source post: http://www.nogripracing.com/forum/showthread.php?t=280634

 

Just installed DiRT 3. Love the graphics. Even at ULTRA setting preset (which will use DX11 with Tessellation activated), still managed to get a decent average FPS of 41.51 (using DiRT 3 built-in benchmark) at 1920x1080 with 2xMSAA.

Running DiRT3 at my favorite resolution, 1280x720 with 8xMSAA, using DX9 (all at high setting), the graphic is gorgeous and crisp, and, I just can't stop admiring it. The graphic made me going slowly instead of rallying, like some tourist doing sight seeing in Finland woods, driving a Mini Cooper .

And DiRT 3 physics? Actually it's surprisingly 'okay'. Honestly, I was hoping to get an arcade-ish physics from DiRT, but I was wrong. The force feedback however, is waay behind rFactor 2. I really miss all the inputs in the way I feel it in rFactor 2. Kudos to ISI for making such a great FFB.

 

I also tried jayblue's (NoGripRAcing) settings, which use in-game AA and saw little difference to my basic settings, based largely on taufikp's settings from 1-30-2012 above. Maybe jayblue's settings gave a minor hit to FPS, and equal quality. But my settings are all Adaptive Multi Sampling AA in AA Mode. I evidently don't have the processing power to run Super sampling. It gives me unplayable FPS. So for me, CCC AA or in-game are about equal.
CPU: Intel i7 2600s @ 2.8GHz
Motherboard: Pegatron Carmel
Memory: 8GB DDR3
OS: 7 Home Premium 64bit
Video: ATI Radeon HD6570 1GB GDDR5

 

Just install ATI driver 12.2 preview version, and I got an issue with its Catalyst Control Center (CCC).

After quitting rFactor 2, I check the task manager and found that 12.2 CCC is using 120+ MB of RAM! CCC that come with 12.1 didn't use that much RAM if I remember correctly. Aside from this issue, I found no change in terms of performance.

So I disable CCC, and use ATI Tray Tools (ATT) instead.

Also, I found these in my short, lazy-ass mode experiments:
- Set the video driver AF to 'Application' and now using in-game trilinear. Didn't measure the performance, but the graphic quality is better than bilinear setting (obviously).
- FXAA's Bloom module is an FPS killer. Using this module alone can drop the frame-rate as much as 16 FPS.

UPDATE:

Using in-game 16x anisotropic filter doesn't hurt the performance at all. However, the graphic quality differences between 16xAF and trilinear can not be seen clearly. Oh well, I chose 16xAF, who knows.

After further testing, the version 12.2 driver actually gives quite an improvement over version 12.1.
Using the same setup from post #24, I got this result:

FPS Min: 27, FPS Max: 69, FPS Avg: 48, FPS at start: 36

A 7-points jump on average FPS. Nice driver update by AMD.

Another ATI feature (found in ATT only) that is worth to mention (I just did a simple test):

Block Write
Enabling Block Write boost the frame-rate by 4 FPS. However, since ATI disable this option by default, I'm guessing it's because not all card will benefit from it. You can try this yourself to find whether your card supports it. My card is ATI HD5770 PCI-e 16x.

Multi-Threading
Enabling multi-threading does help the GPU performance (although not much, only 1-2 FPS). I recommend to use Min=0 and Max=2. Using Max more than 2 doesn't improve GPU performance. Once again, this is also disabled by default. So not all ATI card have support for multi-threading. Note: only enable this feature if you're using multi CPU. In single CPU this feature won't work at all.

 

SMAA - another D3D9 Injector

I just tried rFactor 2 with another d3d9 injection.

This time it is SMAA, the result of joint work between Crytek and Universidad de Zaragoza Spain. SMAA is also said to be the successor of FXAA (made by nVidia) because the result is much better anti-aliasing and less blurring in textures. Here's their description of SMAA: "Our method shows for the first time how to combine morphological antialiasing (MLAA) with additional multi/supersampling strategies (MSAA, SSAA) for accurate subpixel features, and how to couple it with temporal reprojection; always preserving the sharpness of the image."

The video below should be watched in High Definition to get better look at how SMAA's doing compared to other anti-aliasing technique:

Based on my test, at 1920x1080, SMAA is reducing the average frame rate by 3 FPS only. However, the notorious shimmer effect on distant fences and jagged white line on the edge of the track still appears (both of these problems can be eliminated almost completely using 8xSSAA [Super Sampled Anti-Aliasing], but at the cost of massive FPS drop).

You can download the injector from MrHaandi's Crypto Corner blog.

 

I think it's time to do another benchmark since build 60 is out.

However, my first impression of build 60: LEGENDARY!

 

Ah aye sir.

It's 3 FPS less from 48 FPS (average) from my previous post (ATI 12.2 driver). Using SMAA I had 45 FPS average. Visually, the end result is better than FXAA, I'm hoping ISI could implement SMAA as well. They already have FXAA implemented in build 60, I guess it should be easy to implement SMAA next, isn't it?

 

Some feedback is greatly apreciated.
I have an amd phenom x4 20 at 2.8Gh(unlocked the 4th core), 6Gb ddr3, and a GeForce 550 ti I run everything on the Highest setting, no rain drops, no wind, 16 visible cars and the rest on or yes also 4x AAF . I use 1280x768 32 bit on my vizio 26' tv 720p. I get 59 to 61 frames but I am not very satisfied.
You think it is because I am using my TV instead?

Ps: By the way LFS runs great on high with 100 frames.

Thank you again for your time

GreekSchumi