NEW Benchmark comparison between AMD & Nvidia - 2.0

PLEASE IGNORE THIS THREAD. THERE IS A PROBLEM WITH THE REPLAY BENCHMARK METHOD MAKING IT UNSUITABLE FOR ACCURATE AND CONSISTENT BENCHMARKING PURPOSES

There will be a new benchmarking thread called "Live Benchmark compoarison for rFactor 2" up soon...

[HR][/HR]
Unfortunately the replay file John made with the civic + silverstone for benchmarking was corrupt and giving wildly inconsistent results.

This is however a reboot of his well intentioned thread for having a new benchmark comparison so that we know how cards perform relative to one another.

For this benchmark "reboot", i've chosen to stick with the first replay benchmark made (by paarma) and using his max graphics settings. The two reasons for sticking to his replay benchmark....

1) So we can compare performance of our cards today vs 9-12 months ago in the exact same benchmark. Differences in performance will be due to either driver changes, build changes or some combination of both. It will be nice to see how (especially) AMD cards have improved over the last 9-12 months.

2) We know for sure that this replay works without a problem and yields consistent results each time it's repeated.

For reference, this is how the cards performed relative to one another 9-12 years ago (remember though these fps score do not reflect live performances, only use them to compare relative performance difference between cards...e.g. GTX 780Ti = 86 fps, GTX 770 = 66.4 fps....GTX Ti has 29.5% more fps. These relative performance differences apply to live performance as well):



Download: rF2 benchmark with Clio + Loch Drummond.zip

What is needed:
-rFactor 2, Renault Clio and Loch Drummond track
-Fraps-> http://www.fraps.com/

How to test:
-unzip the bench.Vcr file to "rFactor2\Userdata\Replays" folder
-use the rFactor2 settings shown in the three screenshots provided (please take extra care to ensure you've mirrored all settings perfectly!)
-load the replay
-press ENTER (for fullscreen)
-press INSERT x 3 (cockpit view)
-start FRAPS benchmark when buzzer sounds and lights go out
-stop FRAPS benchmark after two laps on the finish line.
-perform a total of 2 benchmarks (to ensure consistency)

Example results:

note:

-for nvidia users, please note you core speed in it's boost (load) state by using a program such as after burner. That way you can check what your actual core speed is during the benchmark by checking the after burner graph this way.

-"core boost" and "mem" refer to your core and memory clock speeds whilst in the load/benchmark state.

 

I guess i'm first

CPU: i5-2500K overlocked to 4.3GHz
rFactor 2 Build: 860
Graphics driver version: 344.11


GPU: GTX 770 Stock - 1085 core boost/1750 mem

1st Bench Run:
Frames: 12690 - Time: 199915ms - Min: 52 - Max: 79 - Avg: 63.477

2nd Bench Run:
Frames: 12668 - Time: 199484ms - Min: 52 - Max: 79 - Avg: 63.501

__________________________________________________


GPU: GTX 770 Overclocked - 1241 core boost/1852 mem (+156 core, +102 mem)

1st Bench Run:
Frames: 13460 - Time: 199759ms - Min: 55 - Max: 84 - Avg: 67.381

2nd Bench Run:
Frames: 13520 - Time: 199871ms - Min: 55 - Max: 84 - Avg: 67.414

 

CPU: i5-2500K overlocked to 4.3GHz
rFactor 2 Build: 860
Graphics driver version: 344.16


GPU: GTX 970 Stock - 1178 core boost/1750 mem

1st Bench Run:
Frames: 15520 - Time: 199525ms - Min: 63 - Max: 97 - Avg: 77.785

2nd Bench Run:
Frames: 15560 - Time: 199469ms - Min: 63 - Max: 97 - Avg: 77.823

__________________________________________________


GPU: MSI GTX 970 Gaming 4G - 1291 core boost/1753 mem (+113 core, +3 mem)

1st Bench Run:
Frames: 15957 - Time: 199806ms - Min: 64 - Max: 100 - Avg: 79.862

2nd Bench Run:
Frames: 15980 - Time: 199824ms - Min: 64 - Max: 100 - Avg: 79.893

__________________________________________________


GPU: MSI GTX 970 Gaming 4G Overclocked - 1489 core boost/1847 mem (+311 core, +95 mem)

1st Bench Run:
Frames: 16784 - Time: 199806ms - Min: 67 - Max: 106 - Avg: 84.001

2nd Bench Run:
Frames: 16796 - Time: 199845ms - Min: 67 - Max: 106 - Avg: 84.098

 

There's prob a ton in the last few pages of the thread that was just ended no?

 

With the Clio at Loch Drummond replay?

 

Why not using newer track and car from ISI and a new replay? Both Clio and Loch Drummond is old.

Anyway, i have zero inconsistency with the Civic + Silverstone replay.

 

And the Level 3 AA thing, what is that and how to explain the level of AA in the AMD drivers. I just can see 2X - 2XEQ - 4X - 4XEQ - 8X - 8EQ levels of AA in the AMD driver but rf2 offers 8 levels.

 

And what level is that for AMD than ?

 

It does seem like the FPS results achieved a year ago don't seem possible anymore.
I was one of the 780ti OC guys with an avg FPS over 100 while, with higher clocks, I'm now only able to achieve an avg of 84-85.
i5 4670k @ 4.4Ghz
780Ti core 1240/mem 1870
GFX driver 335.23

Code:

Frames,     Time (ms),   Min,   Max,     Avg
 17053,      199985,      66,     113,     85.271

Frames,     Time (ms),  Min,    Max,     Avg
 16934,      198347,     66,     112,     85.376

I did get 1 run where I suddenly had an average of 97fps but it only happened once in like 5 tries

(Don't include the following in the graph but it's somehow possible )
Frames, Time (ms), Min, Max, Avg
19524, 200268, 76, 131, 97.489

 

That's inconsistency right there, the replay was corrupted. We need a new thread. =P

 

This benchmark thing doesn't works well and the issues in my case still remains because as already said in the other thread, OC the 290 does nothing in rf2, and i'm still of the opinion either the CPU is to weak in general or to weak for rf2 or rf2 still doesn't utilizes AMD GPU's as it should.

Howsoever, only for the peace here again with the LochD. benchmark on Level 3 AA whatever it is for AMD cards.

CPU @3800

290 standard:

Frames, Time (ms), Min, Max, Avg
12433, 199946, 50, 79, 62.182

290 10% GPU OC:

Frames, Time (ms), Min, Max, Avg
12503, 200476, 49, 78, 62.367

As you can see something is off and i'm not going to test another three days for nothing. Awake me if anybody knows what's wrong, if it is the CPU or rf2 or both or something else. It's a mess and i'm weary.

 

No the max clock of the GPU is 1240 mhz, and yes it was done on a 4670k @ 4.4

 

CPU: i7-2600K overlocked to 3.8GHz
rFactor 2 Build: 860
Graphics driver version: 340.52


GPU: GTX 580 Stock - 772 core/2004 mem

1st Bench Run:
Frames: 10360 - Time: 200789ms - Min: 43 - Max: 64 - Avg: 51.596

2nd Bench Run:
Frames: 10343 - Time: 200258ms - Min: 43 - Max: 65 - Avg: 51.648

 

What about the settings in the Nvidia (or AMD) Control Panel? How should they be set?

1920x1080 doesn't show up as a choice for me.