Live Performance Benchmarking Comparison for rFactor 2

My opinion is that in this forum a mentality has crept that AMD must look worse no matter how, but if you look at the relation between the 980 and the 290 which is a non X and driven by a outdated base I think it actually looks really pretty bad for the 980, and that for a new gpu gen, apart from the price-performance ratio, and the possibility to cheat with the driver and tools.
..... all this is not helping anybody, either ISI to optimize nor the GPU manufacturer to improve there driver.

 

Speed1

I agree somewhat, AMD has caught up in the driver department. Not just in here, but in other programs as well. No need for synthetic test to prove that. Its pretty clear. Unless your on a dual core CPU any decent quad core CPU should have no issues runing rfactor 2.

What this way of benchmarking is proving though is I'm on a first gen I7 and the most CPU utilization I saw was 84 percent on one core. I get 10 extra frames if I overclock from stock. which is minor.

My GPUs where at a constant 97 percent for the first time since the early builds. No fluctuations or drops to 10% causing massive stuttering or frame pacing issues, thats a big plus especially for an AMD card.

This is showing that someone has done something in the right direction..............the question is who.

As for Rippers performance issues, I think its his own system or a bad card. Perhaps he might even have to RMA, I've seen this happen before as well as happen to mean. Considering another user of the same card was getting higher results.

 

As i said already homie, this is biased and something like underpining the personal interests of NV bootlickers, and this has nothing to do with the purpose to help anybody. This is not analyzing anything professionell, this is just d..k measurement without any solid logic, nothing else.

You know, i've seen a lot of idiots trouble shooting a defect on a car, which have a lot of time and money lost with there theory but had forgotten during process to check the fuses first.

 

I'm with you there somewhat speed1, I can see where you feel its boosting nvidia card users, I saw that originally in the first few threads, but what your also missing is that it actually showed how much AMD has caught up. Compared to that first benchmark chart we saw so its been counter productive if that where the case.

Your 290 at stock speeds is proving that. Your scoring better than I am as you should with a 290 vs a 7970 and our base system is roughly the same age, also during a time Intel CPUs gave better performance numbers than AMD CPUs.

You have to look at the glass half full bud.

 

I see and agree with you but i'm not a kid or a gamer homie, i sure was and am still to a part, but when I see no logic only children playing and people who believe the world is pink I can not keep up anymore.

You know my opinion is that exactly this and that sort of people destroying rf2 without to mean anybody personal, but.....................many people are just silent, just saying, and i know a lot of differnet opinions of experienced sim racer and real world professionells, believe me. It's not any silly server issues, it's the software itself and the userbase present which is loud and makes lot's of useless noise muting people, as this leads again nowhere until the last soldier is calm.

 

To make NVidia and AMD benchmarks as even as possible for benchmark comparison purposes, apply the following settings to NVidia Inspector and/or RadeonPro:
(Ignore all the settings in the following pictures as they are just random pics I found online. Just set your control panel settings to full default (click default, apply/save), and then just adjust what I have noted with red lines andf their corresponding notes, ignore everything else.)
[HR][/HR]
NVidia Inspector



[HR][/HR]
RadeonPro

 

My homie spinelli, crazy guy he is, thank you man, let me talk to you per pm bro from tha south, this leaves nowhere anymore. Ya know if 2pac was still alive he would be the president of the USA, is what i believe why he isn't alive anymore,...........fear !

 

I feel you brother believe me, your heart beats for racing and it's ok, but this topic is beyond every logic. Gn8 homie !

 

The fact that you believe there is some ulterior motive to all this (e.g. e-penis length) astounds me.

The whole purpose of this thread is to compare with a simple and short comparison test as its baseline. You are then more than welcome to compare outside of the stated settings with other users to test you theory further. As someone with an nvidia card to also do benchmarks with AA & AF completely off and see how the fps percentage gained compares with either card groups.

The thread is as much as means of comparing as it is of formulating new questions based on our results compared to other cards. One such example is with my own card not performing the same as other 970 card users.

 

Is one of the reasons why I say it leads nowhere, no insight, no understanding, no logic, only battling and personal interest. I leave the room and i wish you much success and fun because it is not worth the hassle, solely on the basis that it seems the responsible give a damn f...k on it as well, and i hence too. Maybe you can do to revive rfactor2 before it dies on this way.

 

Fine, you either don't believe my honest intentions with this thread or you simply don't see eye-to-eye with me on the benefits that can still be had in such a comparison.

Either way....you are more than welcome and capable of starting your own benchmarking with your own requirements. No one has tied your hands behind your back so please, stop making useless complaints and instead be pro-active about it if you really don't like what i've done.

I'm sorry that i had to speak to you like that but i fear it's the only way to get my point through to you.

 

This is unexpected and doesn't sound right at first glance. Would you be willing to repeat the tests and show the results please (including cpu utilisation graphs please)? Also, were you using "Max pre-rendered frames = 1"? Is your i7 in Hyper-threading mode? Whats the clock speeds?

I'm very intrigued by what you're reporting here because it doesn't fit with what i've been observing over the last week. Again, more than happy to be demonstrated wrong but i'd really appreciate seeing the data please.

I'm now 99.99% confident that it's not my system or a bad card and here's why. I compared my 970's performance with 970 reviewers and other users in Tomb Raider, Metro Last Light, 3D Mark - Firestrike, Unigine Heaven and Unigine Valley. Used the exact same settings and clock speeds and the results were identical.

What this means however is that the problem lies with rfactor 2, either on my installation (because 2 other 970 owners scored correctly to one another in the live benchmark) or there is some hidden difference in settings or the way rf2 is running between mine and theirs.

I also tested rf2 with a clean install of Windows on a completely different hard drive and results did not change.

More deductions are required to figure out what is causing the disparity in performance.

 

Yes, it's at 98+% and constant.

 

1x NVIDIA GeFore GTX 780 Ti @ 1163 MHz core - 1750 MHz (7000 MHz effective) memory
- Did not use the equal-between-NVidia-and-AMD driver settings posted in post #66 - just DrR1pper's settings specified in the OP
- Repeated benchmark 8 times, every time exact same results. Results are as follows:

Avg: 164 - Min: 131 - Max: 191
the seemingly low CPU demands of this particular benchmark means that there is no difference between 3.3 GHz and 4.5 GHz with my Intel i7-4930K CPU

[HR][/HR]
Due to the very high framerates and low demands of this particular benchmark, perhaps you should consider changing the benchmark settings to night time (let's say 12AM), 8xMSAA, and rain (or at least to just 12AM and 8xMSAA). We should then be set for a quite a while

Default ---------------- Avg: 164 - Min: 131 - Max: 191
12AM ------------------ Avg: 148 - Min: 115 - Max: 178
12AM, 8xMSAA -------- Avg. 137 - Min: 107 - Max: 163
12AM, 8xMSAA, Rain - Avg: 126 - Min: 99 - Max: 147

 

And that's a very good thing. Thanks for confirming!

I had thought about this as well but since we're already 4 pages in with results, maybe we should stay the course, or at least for now. We can also start over if need be but some of the results here have raised questions that we need to answer before me move on.

I know, this means there may be yet another NEW benchmarking thread but if it's needed, then we must. It's important we do these tests to find out if there are any things we have missed affecting performance so we can come up with a well formulated benchmarking process that tests the gpu only.

Can you please report back with your score at stock GTX 780 Ti clocks as well please. That way we can compare performance ratios between stock GTX780 Ti and stock GTX970 (using Sentri's results) and use online reviews to see how the performance ratio in rf2 stacks up against other games and benchmark.

Thanks!

 

When my GPU used to boost and all that crap I think it used to stay within 1090 and 1110, that's not too much lower than my 1163 so I don't think the results will be too much lower. I can't set it back to all that auto-boost, dynamic clock crap as I installed a custom bios specifically to get rid of all that boost crap. The boost cause a micro-stutter every time it changed clocks, even if it just changed by 5 MHz or so, and my GPU graphs shows it changing clocks tons and tons of times. It would usually stay at a high clock, but even if it's always switching between 1100 and 1125, every time it switched it would cause a stutter. In certain situations the boost crap just outright lowered framerates like when using VSync, the card thinks it doesn't need to have such high core clocks in order to maintain the framerate, so then it lowers it clocks and then when something all of a sudden more demanding shows up on screen then the clocks are too low and then the framerates dip and then you have to wait for the clocks to rise again, what a joke.

 

Oh ok, np then.

Regarding your dynamic clock changes, i have had no experience of this on my GTX 770 or 970 before. When in a game or benchmark of any kind, i've always seen the core and mem clock stay constant.

How much is you clock speed dynamically changing by under load? +/- 20mhz or all the way down to idle and back up to boost?

edit: i just realised you said "I think it used to stay within 1090 and 1110" so i'm guessing that's what you meant when saying it's was dynamically throttling?

If so, TechAde had the same problem as you with stock clock of his card as well. The solution is to add some core voltage (in gpu-z or afterburner, etc) but i guess it's too late now with your permanent bios change.

 

There are the low idle/2D/desktop clocks and voltages, and there are the 3D application clocks and voltages. Simple as that. Just like AMD does it, and just like NVidia used to do it before all this boost crap .

With regards to the second part of my previous post (post #76), A. I have edited the post slightly, and B. one thing to keep in mind is that your benchmark is generally one we can keep using for a while, so we may have quite a few results over the past few days but it's hardly anything compared to the results we'll have in 3 weeks - let alone 3 months - from now.

When you say your clocks don't change, have you used a monitoring graph (available in Inspector, Afterburner, Precision) and checked the entire graph over a good 5 minutes of gaming or so? Play a real low demanding game (like rFactor 1) with no opponents, but enable VSync @ 60 Hz, see what happens during a 3 lap stint or something. Then try it during an intensive Benchmark (no VSync or framerate limiters) and check again. Make sure you stretch the graph so that you can more easily see any small clock changes of 5, 10, 20, etc. MHz.