New AMD driver to reduce frame latency issues.

While better than the previous version, the 13.4 drivers in rF2 are....meh.

HD7970, sad to see such a fast card fall on its face with rF2. Hard.

 

It's a shame. This game should be a breeze for my overclocked 7950 but the results are quite different.

 

that's the MAIN reason I have not upgraded from my card either. Right now the game runs at 60 vsync locked to video. 80-120 with vsync off. All maxed. I don't know what else I am looking for, well other than the annoying tick tick with vsync off, being gone. I would HATE to get a monster card and it runs Worse than my current one. I am in hopes for summer, and if they can get it together with the promises of better memory management, and such

I guess I am looking for too much? I am curious, to the other members that have my card, 6950, or newer AMD cards. What are your results while in race, and Quali? I know in Quali you can have amazing frames without the AI n such. A full grid brings frames down to 20's 40's til the field thins out. I assume that part is optimizations on ISI's end, so in sticking with qualifying for the moment. All tracks run pretty much fine. As I wrote, 60 with vsync(environmental reflection drops it to 54-575fps) and up to whatever frames I want, with vsync off. My issue is the ticking with vsync off.

With this latest certified driver, the ticking has reduced considerably, a huge improvement for my system actually with this Sim. So I wanted to have other people's thoughts about that, with their cards. I hear the newer cards are having a harder time with this sim then the older ones like mine are. Right now the Sim runs great, I don't know what else I am looking for.. lol.

Just wanting thoughts about other peoples cards I guess. If AMD fix their drivers as they "promise" they are going to, things look good for AMD and rFactor2

Sorry for the ramble. Also was curious if there was any information about ISI working on improving relations with driver support with AMD, or they are just waiting for AMD to fix it? Enjoy our day

 

I run the game on full settings with 4xmsaa forced through Catalyst. I drop shadows to high and optimal but otherwise everything is maxed.

On an empty track I can get a rock solid 60fps but when AI is involved the fps can take a big hit. Even sitting on the grid the fps is down to 25-30 before the race even starts. Seems odd to me.

 

I'm currently running the 13.5b2 drivers. Seem to be the best for me so far, (HD7970 as well) however rFactor2 has a rough time performing with triples in multi view, even with just a dozen or 15 AI. Add HDR to the mix and it's really not playable IMO. This is with turning most options down from Full to High, med shadows, low special affects, etc.

Of course we're still in beta here and it would seem an update is immanent, so perhaps some of this will improve. I'm also comparing this to iRacing where I run pretty much anywhere with 25-30 cars, FPS locked at 119, most options turned on high. I typically only run shadow maps when I do want them, which is mostly on road courses. Not necessarily trying to compare the two sims, just my experience of them.

How ya doin' Bob?

 

Iracing calculates much less than RFactor 2. Shouldn't compare the frame-rates of the 2.

1. RFactor 2 calculates many more realtime shadows than iracing, Iracing uses many more fake/static/permanently drawn as part of the texture/object itself kind of shadows. These aren't shadows at all, but just an area of an object that has been permanently "painted"/"colored" in shades of grey to look like it has some sort of shadow on it.

2. Not only does RFactor 2 calculate heat, humidity and all that, but there are calculations being done based on not only where the track is located in the world, but also what season it is. So the location of the track, plus the time of the year, are all being calculated in RFactor 2's "real world/weather" system, and they affect the physics AND the graphics.

Iracing doesn't have/calculate ANY of that.

3. Continuing on with RF2's "real world/weather" system, it also calculates a real sky with real clouds. This, just like everything else in RF2, is being calculated real time, it's not a fake/pre-drawn artist's painting of a sky like in Iracing and many other sims/games.

Iracing doesn't have/calculate ANY of that.

4. Further continuing on with RF2's "real world/weather" system, it has full realtime 24/7 lighting, as the sun goes up, moon goes down, etc

Iracing doesn't have/calculate ANY of that.

5. I also remember, I believe, Tim pointing out that humidity will be higher in areas with more trees and such, like certain sections of the Monza track, down the long straights through the forest of the old Hockenheim, throughout certain sections of Spa, etc. Now I'm not sure if this difference in humidity (affecting the amount of rain, drying time, visibility, and more) is calculated dynamically aka in realtime, or if it's pre-determined. However, even if the areas where this difference of humidity occurs was pre-determined by the track makers, I'm willing to bet that the actual way the humidity difference occurs, relative to the "normal" areas of the track, is still being calculated in real time since everything else is being calculated in real time, and the 2 areas (extra humid and "normal") have a direct relationship to one another.

Iracing doesn't have/calculate ANY of that.

6. I haven't even mentioned RF2's "real road" system yet, and how it is calculated in real time and affects physics and graphics. Grip level changes, rubber build-up, drying times, wet weather lines, different areas of track all being affected differently by water, and how all of this is connected to the entire RF2 "real world/weather" system.

Iracing doesn't have/calculate ANY of that.

7. Also, I'm willing to bet, but can't confirm, that RFactor 2's tyre modeling engine pushes hardware more than any other consumer based video-game/sim.

Conclusion: RFactor 2 has, by far, much more of a dynamically calculated, and connected to one another, world than any other sim/racing game. Not only does RF2 calculate way more "stuff", but all this stuff is dynamically in realtime connected to and affecting other "stuff". RFactor 2 calculates the sim/game as one complete world where everything in the world is calculated, connected to, and therefore affected by everything else in the world, much, much, more than any other consumer sim/video-game, PERIOD.

I am not an Iracing hater nor an RF2 fanboy. I play iracing sometimes, and have purchased, and still play, most sims on the market.


P.S. Feel free to copy and paste this post, especially points 1-7 & conclusion, on any forum/website/article you'd like.

 

I really don't want to look like someone superior or what but as a programmer which had some experience in computer graphic i would say :


1. Yes, that would improve shadow performance but that's called optimization and no one can deny that the final result would be much more than acceptable, specially when you are inside a car at 90 mph. If you talk about replay, well older games used to improve replay definition separately from the real time game and trust me it is not that hard to do it.

2. CPU are extremely fast and completely separated from GPU, that kind of thing are calculated by the cpu, unless you don't force it to be calculated by the GPU which is not productive.

3. Again, there is a split between the process where the calculation of the density, position, direction and others stuff are done by the CPU and the RENDERING are done by the GPU, and again the computer are fast enough to do it pretty easily.

4. Same as the last two point. Light position, light direction, light density, light type and others stuff are calculated by the CPU and the render phase are done by the GPU, and in terms of light is not the 80's anymore where for further improved light (in terms of visual effect) had to be done manually by code because of SDK's lack. From 20 years SDK like directx and opengl offer lights features out of the box which can be easily used and are highly optimized, the programmer can improve the effect but it is done by tweaking the code and adding some effect.


2,3 and 4 For these the amount of calculation are very small if confronted to others stuff. In case that the calculation are excessive than it is a lack of optimization and simplification. Every stuff can be improved and simplified but still giving a result close a complex path. The question is complex or simplified path? Well it is all about trading and most of time it is not worth the complex path for many reason (bugs, result, etc...). If you want me to give you an example i'll give you flight simulator x and x-plane 9, the first one use statistic and probability based on an amount of historical data, the second one calculate everything in real time....well the result gives flight simulator a better final result with no doubt.



5. Almost the same about those above.

6. Again about the same, obviously there is a lot of trick around that. If you say about you car feeling then i would yes there are a lot of real time calculation but but if you are taking in consideration the IA then no there is not many real time calculation but rather a lot of script, it seems that the only real time calculation is the line, radius and speed. Not only it is scripted but also bad scripted if confronted to the old grand prix 4. I saw many time the IA loosing their rear wing and not having a tiny grip problem. The same for the suspension.

7. Same of the point 6, again the IA is extremely scripted.

Conclusion: CPU does a lot of dynamic calculation for the LOCAL player, the IA instead is extremely scripted. Weather stuff are not that heavy in fact grand prix had a complex weather engine and still provided good frames. If the bottleneck was due to the CPU you would not notify any change when you lower the graphic's detail, because CPU are not hit by rendering. The problem is purely due to the rendering engine, where Iracing did a PERFECT job. Rfactor is still in beta so it is still ok but i guess that they must revise they engine for a final release because like someone said it is not acceptable to have a high end video card and being hit that hard.
In terms of realism i prefer rFactor and i add that it is better than iRacing, in term of performance i'm sorry but iRacing is far ahead.

 

Max, the ticking as you call it is due to frame render times. It has been an issue for a while and that is what AMD's main focus is on at the moment which is why the latest driver has reduced the tick for you. I would look into the beta driver to see if you get any further improvements in that area as that would give you a general idea if they are still working towards eliminating that issue.

 

Time for a new card for me coming soon. Should I just ignore AMD and get an nVIDIA since both iRacing and rF2 seem to work so much better with them? My other main concern with AMD is the appalling looking graphics in rF2 unless you use SuperSampling. Adaptive Multi works wonders on everything except rF2. Kills the frame rate. But since this has been the case all the way back to rF1 (also looked like crap unless SuperSampling was used), I am not holding my breath that it will suddenly get fixed by ISI for rF2.

Comments?

 

Williang83, I will have to disagree on agreeing with you. You have no idea how complex the algorithms being used by RFactor 2 are for all of it's physics systems.

- I bet GP4's wet weather system was much more scripted than RF2.

Also, using fake, drawn, shadow looking objects is not simulation. It's 2013, the point of computer software is to dynamically simulate and calculate as much as possible, after all, computers are on a slow path to having everything being real-time dynamically calculated (like real life).

- I never discussed, or even mentioned A.I.

- CPU and GPU are very much connected, especially in scenarios where graphics and physics systems are not only real-time dynamically calculated, but are dynamically connected to each-other.

- You can't make the argument that the tyre model only calculates the player. You can have tyre models so complex that they take 20 supercomputers to run just a set of 4 tyres, and we have no idea how far ISI has pushed tyre modeling.

- Have you seen Battlefield 3's damage system? It might look good, but it's scripted aka fake aka pre-determined. On the other hand, Battlefield Bad Company 2's damage system, had much more calculation in it and was MUCH better and real technically speaking.

- Have you seen Nvidia's boat with real time calculated waves demo? It takes an Nvidia Titan GPU (and some CPU) to do all the calculations (graphics, physics, graphics to physics linking, everything) just to process 1 single boat with waves. It looks gorgeous, not close to perfect, but way better than any game. We have to push forward technology, with "faking" and "scripting" things as least as possible (especially in hardcore sims).

We should be using all our computers' power to be calculating and displaying this sort of stuff, rather than using it to display 9 billion polygons of a stupid car model so that people get brainwashed into thinking a game is better because of it's looks, or so that people get obsessed with taking pictures and videos of the ****tiest sim just because it looks nice, or simply wasting all the power of the computer and playing at 250 fps like IRacing does.

I do agree RF2 definitely can be optimized more, but the general direction and way of thinking of ISI is PURE SIM in my opinion, rather than a wanna be sim video game, like others.

However, RF2 still needs DX11, NOT for looks, but for the technical side of things (it will most likely get it later) and also more complex models for damage, transmission, drivetrain and sound are desperately needed. Along with further improvements to all other physics systems already in place (rain, tyre model, suspensions, etc). ISI seem to have the most "hardcore sim" way of thinking about things, that has it's positives but it's negatives as well (mote complex, lower frames, maybe not as good graphics, etc).

 

Well i only used my experience in hardware programming when i used to program in CPU instructions and had pretty knowledge of CPU capabilities. I also use to work on complex algorithm to deal with CPU instructions lacks.
Again as i learned if someone pick the complex path instead of the simplified one might be a big mistake due the same result. Surely GP4 weather was scripted but that's the point FINAL RESULT.
You can't compare iRacing and rFactor without taking in consideration IA because this is the main difference in term of what has to be calculated and in a racing game what would dramatically decrease performance are in fact real IA calculation which rFactor doesn't do.
CPU and GPU are not connected directly. Older games used to connect them (ex. first Hitman), when you do that you see when the performance is low the game frames are not skipped but the game goes in slow motion. Today no one do that because it is better to have missed frames than slow motion game specially for online games. If you want you can easy find game coding on google to see how it works but roughly the pipeline process goes like this :

- there is a main loop that goes until the process is killed.
- this loop call one or more method, but primary it calls the "game" method which calculate positions, damages, directions, collisions, etc... All done by the CPU even collision are done by CPU and not GPU even if this is the responsible for the rendering.
- Once the "game" method is done, the rendering method is called. The rendered method does nothing else than send all "coordinates" to the GPU.
- For last but not least, the GPU receive it as a pipeline independently, process it and draw it. Mean while since this is a pipeline some data might be lost creating a frame skip.

As you can see, the GPU was done with the idea to be 100% independent, some crazy programmers tried to force both to keep them directly related to each other but this crazy idea, luckily, has been dumped.
Please does not take in consideration really complex tires model because both rFactor and iRacing have complex tyre model, the difference between them are in some details.

Again i will end with if the CPU was the bottleneck then you would not notice any difference when lowering graphic details, this is just a pure and simple thing, it can be seen in others game which require a lot of CPU.

About the "You have no idea how complex the algorithms being used by RFactor 2 are for all of it's physics systems", well the answer is pretty easy.. When you are playing alone with no IA does you computer struggle? No? So the amount of calculation are far within you computer capabilities, everything related to IA are scripted so not heavy which is not the same for the GPU which calculation increases according to the amount of objects and details.

I'm a big ISI supporter but let's not give to them more than they really do (which is amazing), they have done a bad rendering engine and that's all. We have a plenty of game with better engine, but they are still in time to improve it since we are still in beta.

 

Actually, without A.I. people have much lower frames than they would expect, taking into consideration how not state of the art RF2 visually looks.

You cannot say that it's about even between RFactor 2's and IRacing's tyre model processing intensity because they are both "complex". You have no idea how complex and computer processing intensive one is relative to the other. Like I said, there are industrial tyre models that literally take many, many supercomputers to run. RFactor 2's tyre model may be 3% as close to these industrial supercomputer models, while IRacing's may only be 1.2% (example).

It's not always about the end result, but sometimes how you get there as variables are always dynamically changing. The end result of Battlefield 3's fake predetermined damage may be beautiful to some peoples eyes, but is crap to my, and many other players' brains compared to Battlefield Bad Company 2's much more real time calculated damage system (mostly for buildings and stuff like that).

P.S. I don't think I give ISI too much credit. As of public build 198 the transmission model, drivetrain model, damage model and the entire freaking sound model is too basic.

 

The bottleneck in rF2 is the GPU for sure. Yes there is a lot more going on in rF2 compared to other sims but it is mostly done in the CPU as williang83 said. Hasn't anyone looked at CPU usage in rF2? It doesn't come anywhere near 100% on my i5-2500K yet I still struggle to run 60fps without turning stuff down.

 

While this discussion is highly informative, I didn't intend with my post to get people too worked up. That's the reason I stated that I wasn't comparing iRacing to rFactor2, just my experience of them. I also stated that I realize we're still very much in beta with optimizations surely to come later.

There's no question that ISI is on a ground-breaking mission to build the most highly accurate sim racing experience possible. It would only seem logical that the end result will provide something that is comfortably playable, certainly for those with higher-end systems.

 

I understand why many seem to agree with the quotes above--- although most of the argument seems to rely on CPU utilization not being all so high during play... but the real crux of this argument depends on this "fact" that changing your video settings WILL radically change your framerate.

Only thing is, it doesn't--- not for everyone. If you read the comments of those who have ATI cards but don't complain about their framerates, you will see hints of what I have experienced, to wit: graphics settings, especially your big "framerate killers", do virtually nothing to mine.

I have seen others say the same. HDR on? 2-3 frames difference, for me, if detectable at all. Mirrors? 2-3 frames, if that. Reflections? Shadows? Detail levels? None of them change that much on my framerate.

I think that you'll find, as I have, that others who have ATI cards and run with all settings maxed out... regardless of their final framerate, the important point is: we run our settings on max because turning them down doesn't make any difference. At least, I think we can fairly say, not enough of a difference to turn them down. I know in my own calculus, losing HDR, reflections, shadows, and mirrors just to go from 60 fps to 72 fps is not a worthwhile tradeoff.

Now, I'm not claiming I know what the "real secret" to all this is. I don't know. But, your set of "facts" about this game does not fit everyone's experience of it. So either some of us are under collective delusion, or the answer is not quite so simple as "the game is GPU bound."

Here's the other point I would make: in upgrading my hardware, I jumped back into the hardware scene... Read any gaming-related reviews lately? Look at the benchmark results... look at the Titan reviews... look at how confounded, how flummoxed the reviewers are to explain what goes on. Remember how memory didn't matter? Post Sandy Bridge, memory appears to matter more than any other component in an Intel system... Powerful cards brought to their knees by visually amazing modern games... or not. Suprising results all over the place. Read the motherboard reviews. Processor reviews. They test with about 12 games, on setups that control for the reviewed products... Look at the number of times you'll see results that confound the reviewers expectations, or explanations for them.

In other words, guys who do what is going on in this thread for a living--- comparing hardware and the performance difference it makes in various computer games--- do not find it anywhere as simple or cut-and-dried as you guys make it out to be. And the theories we're hearing here do not explain all sets of results, only some of them.

 

I do but it depends what you're running, of course I don't usually run with the in game settings maxed out because my GPU isn't fast enough, i5-3570K@3.6, GTX470@stock.

 

And here is my i5-2500K at stock clock:


So why the big difference? Maybe because I use "Max Framerate=60". Even rF1 is nearly CPU bound if you run with your framerate unrestricted.

Just realized you can't see the column header on the Taskman. For those not used to the taskman in Win8, the first column is the CPU usage. Mine is using 20% in this screenshot.

 

Agreed my screenshot isn't showing my typical usage it's just an indication of what's possible, it was one of the reasons I ditched my i3, I thought if the cpu usage can reach those levels I want it on proper cores.

Usage will be lower if you use lower settings and cap the fps.

 

Brilliant post and reflects my thoughts and my experience with my ATI/AMD card. Hence why I was asking for my upcoming new card purchase should I just give-up trying to understand or wait for magical solutions and gor for an nVIDIA that for, mostly inexplicable reasons, seems to run a lot better on rF2 and iRacing, which are really the only two I care about until AC comes out?

 

Personally, I suspect that unified architecture matters more now than anyone is giving it credit for... Time was, all x86 systems had broadly similar architecture--- say with the data path from an ISA slot to the bridges and memory. It is not so simple now. Look at the differences; broad, sweeping differences in motherboards and how they do everything, based on differing chipsets. Your board and mine might have 4 DIMM slots that take the same memory; PCIe slots that take the same cards; and SATA6 interfaces that take the same drives... but if you have Intel's Sandy Bridge architecture, and I have AMD's Dragon platform, there is NOTHING, literally NOTHING about how data moves, how memory is controlled, how the buses are arranged--- none of it is the same.

ATI has had a unified architecture plan ever since the Dragon/Leo platform... Meaning, the motherboard NB/SB chipset, the CPU itself, the motherboard architecture itself, and the graphics card were all designed to work together. When you contemplate how complicated just your video card is; basically as complex as the rest of the computer, but on a card--- how could that ever work equally well with system architectures as different as Sandy Bridge and Leo are?

rFactor 2 isn't "ATI Optimized", isn't it? Well, if you're using an ATI card in an Intel/Nvidia system, tell me--- how hard do you think Intel and Nvidia worked to optimize their hardware for your ATI card? Is the rest of your system "ATI optimized"?


My personal opinion is that motherboards matter way more (and have, for about ten years) than the individual components that go into them. After all, what ties the components together? What makes them work together? What is moving the data? Notice how many things on your motherboard have "controller" in their name?

So I feel that one should (of any component) buy the one that best matches what you have already--- this is the place to be looking for "hidden performance gains" and secret optimizations. Make the system match itself, in other words, before you expect the sim to match your hardware in some next-level way. Or, perhaps it could be put, "if your system isn't internally optimized first, how could ISI possibly handle optimizing it externally, through software?"

Wouldn't that be analogous to taking a car with weight problems, too much engine, too little tires, poor gear ratios for the track... and then trying to "optimize" it with the "right" gasoline, spark plugs, paint, and tire pressures? If the "internal" isn't quite unified, what possible difference could the "external" make?