New AMD driver to reduce frame latency issues.

Discussion in 'General Discussion' started by Nand Gate, Jan 18, 2013.

  1. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Isarmann, I totally agree with every word you wrote, especially in your second last post about it not being as simple and clear as many seem to be making. I tried making this same point when I was saying it's not as simple as just a CPU doing it's own thing, and GPU another. Much more to it than that.
     
  2. Marc Collins

    Marc Collins Registered

    Joined:
    Jan 11, 2012
    Messages:
    3,159
    Likes Received:
    162
    This is a fascinating theory, but not one I have ever seen any evidence for. For example, I could swap an AMD card for a roughly equivalent nVIDIA in my own system and perhaps see a big boost in rF2 frame rates. But, if I switched-out the mainboard, the same AMD card might outperform the nVIDIA in rF2. My sense is that the mainboard might make-up 10% of the effect while the GPU is 90%. Anyone have evidence to the contrary?
     
  3. Nand Gate

    Nand Gate Banned

    Joined:
    Jan 21, 2012
    Messages:
    471
    Likes Received:
    12
    This is completely untrue, and this fact is born out by the thousands of benchmarks done using different motherboards. Graphics cards make ALL the difference. Motherboard chipset manufacturer is about 5th on the ladder of import to game smoothness and playability. Period.

    EDIT: Indeed anti-competitive laws in various countries prohibit exactly this. Intel and Nvidia have been subject to endless scrutiny over it. It is kind of like saying AMD branded "Performance" RAM works better with AMD than G.SKILL RAM. It doesn't. It is the same. They both conform to standards.

    What you need is the latest platforms (Nvidia GPU's are slightly ahead in rF2), cycles (hertz) and lanes. The more the merrier.
     
    Last edited by a moderator: May 2, 2013
  4. Cracheur

    Cracheur Registered

    Joined:
    Jan 3, 2012
    Messages:
    315
    Likes Received:
    8
    sorry but as mentioned before by other people, that is simply not true!

    There are boards that are slightly faster than others when it comes to GFX but not in relation to the gfx vendor choosen.
     
  5. Isarmann

    Isarmann Registered

    Joined:
    Mar 4, 2013
    Messages:
    77
    Likes Received:
    0
    "Simply not true..."

    So, this guy, here (http://hothardware.com/Reviews/QuadChannel-DDR3-Memory-RoundUp/), who said:

    Quite an idiot there, I should say...

    And this guy, here (http://www.hitechlegion.com/reviews/memory/16946-amd-performance-edition-8gb-ddr3-1600-memory-kit), who said:

    Another one! How very wrongheaded of him...

    And this guy, here (http://www.neoseeker.com/Articles/Hardware/Reviews/CMG4GX3M2B1600C7/), who said:

    What could he be talking about?

    And then this guy, here (http://www.tomshardware.com/reviews/bulldozer-ddr3-overclocking,3209-2.html), who said:


    Hard to believe that so many people can write so much about that which is so simple, so not true... and shill for these companies, with their illegal product offerings!

    Speaking of the lanes your cycles take their effectiveness from... have a look at these two images. How similar are those lanes looking to you?

    View attachment 7542
    View attachment 7543
     
  6. Marc Collins

    Marc Collins Registered

    Joined:
    Jan 11, 2012
    Messages:
    3,159
    Likes Received:
    162
    Thanks for the details. That satisfies my request for evidence quite nicely.

    I'd still ike a sense of proportion. I said 10:90 mainboard:video card was my historical experience. How far do you think it might have shifted?

    And lastly, it would make sense that AMD video cards might work best with AMD-based mainboards. Is the opposite true? That nVIDIA is "optimized" for Intel-based boards?
     
  7. Bob Simmerman

    Bob Simmerman Registered

    Joined:
    Mar 6, 2011
    Messages:
    16
    Likes Received:
    0
    I am doing just fine, Bruce, thanks for asking! And I hope the same for you :)

    I am noticing that the worst performance is at Brianza. I took out an F2 at Spa last night for a few test laps and starting at the back of a 14 car grid performance was noticeably better than Brianza, and the random stutters and slowdowns, though still present, seemed inconsequential in terms of affecting my stellar driving abilities. Ok, driving abilities...

    As a comparison--I know, I know, there isn't all the fancy physics and what not--I can run F1 2012 at Ultra, with 8x MSAA, at 1920x1200, pegged at 60 FPS with VSYNC on. Falcon 4.0, Rise of Flight, NR 2003, all maxed, all frame rates pegged at 60 FPS with VSYNC on. So, while I don't think ISI is intentionally dropping the ball here, it is clearly evident that my NVIDIA 480 GTX card--or even my Quadro 4000 card for that matter--are superior at rendering the rF2 world faster than a card that is, on paper and in other programs, vastly superior in gaming land. It's frustrating, but, hopefully, things will get better as I refuse to yank out the card on this new PC for one game.

    For now...


    PS: I'll start playing iRacing again when they get the dirty windshields back in there.
     
  8. Bob Simmerman

    Bob Simmerman Registered

    Joined:
    Mar 6, 2011
    Messages:
    16
    Likes Received:
    0
    Awesome, but still doesn't explain why a card half as fast as the 7970 runs rF2 twice as fast. Wanna talk calculations? I bet Falcon 4.0 does more calculations loading data from the targeting pod than rFactor 2 does in a 24 day race at Sebring.
     
  9. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Highly doubt that. I'm sure just the tyre model of rf2 alone takes more processing power than the entire falcon 4.0 calculations, multiple times over.

    That other flight sim, the one with the a-10 and the chopper. Now THAT'S a real flight sim.

    You're right though, something is messed up in RF2 regarding the performance difference between Nvidia and AMD. However, I never mentioned anything about that in my post you quoted :).
     
  10. kaptainkremmen

    kaptainkremmen Registered

    Joined:
    Sep 25, 2012
    Messages:
    935
    Likes Received:
    17
    Quote from ISI technology page


    "In some cases the computations from our tools are too complicated to be run in real time on current home technology, but the attributes of vehicle dynamics are now fairly well understood, and anything that has a significant real world bearing must be simulated in some manner, even if it is using a simplified model (creating the same results) within retail software. As we go forward, and technology advances, those simplified models can become more and more advanced, and therein lies the benefit of ongoing software development over boxed products."

    http://imagespaceinc.com/technology/car-models-physics/
     
  11. Bob Simmerman

    Bob Simmerman Registered

    Joined:
    Mar 6, 2011
    Messages:
    16
    Likes Received:
    0
    Extremely doubtful, in fact, utter nonsense--if the tire model took more processing power than F4 BMS 4.32, you'd need a Cray to run it.

    And DCS is a single digit frame pig no matter what card you run, but that's another argument for some other forum.

    Sarcasm aside, I don't see how a lengthy diatribe of what rF2 is doing vs. what iRacing isn't doing has anything to do with THE FACT that NVIDIA cards work better than ATI cards.
     
  12. Jamezinho

    Jamezinho Registered

    Joined:
    Aug 27, 2012
    Messages:
    1,256
    Likes Received:
    0
    Exactly.
     
  13. Nand Gate

    Nand Gate Banned

    Joined:
    Jan 21, 2012
    Messages:
    471
    Likes Received:
    12

    You have spent a lot of time on finding irrelevant information! The facts are simple. I can link to thousands of motherboard benchmarks showing that the difference is minimal, but anyone can google that.

    Infact, much of what you link there (confusion through voluminousness?) agrees exactly with what I was saying....

    To coincide with the release of Intel's current flagship Sandy Bridge-E processor and companion X79 chipset, a number of Intel’s memory partners released new quad-channel memory kits optimized for the platform. Previous Intel platforms were designed to offer optimal performance with two or three-channel memory configurations; Sandy Bridge-E and the X79 Express, however, perform best with a quad-channel setup.

    What has this got to do with AMD/Nvidia/Intel optimisation...NOTHING?!!?!

    There have always been differences in architecture between Intel and AMD, but now we are seeing pronounced differences within the manufacturers, based on the chipsets used. One of the examples that stands out in my mind was many manufacturers adding a “hey, really, this RAM will work with the P55 chipset…guaranteed!”. What was the need for this? Well, basically it was because so much RAM wouldn’t work properly with the chipset. Now we are seeing a good percentage of the RAM released being optimized for a specific platform or chipset. This is not only a compatibility issue, this is about optimizing the RAM for the way it specifically interacts with a given platform.

    Utter BS! Which RAM "does not work" with WHICH chipset? He doesn't say. Pretty daft of RAM manufacturer, huh...? Too many people put POST problems down to magic. Anyhow, this is not a motherboard benchmark?


    Since the release of the Nehalem architecture from Intel, the market is saturated with memory kits tuned specifically for this platform. The 1.65V requirement forced memory manufacturers to release DIMMS that would operate at this voltage - specifically for the Bloomfield processors. The Lynnfield processors have that same requirement. When one looks at dual-channel memory kits, many if not most of them are labeled as "Designed/Engineered/Tuned specifically for the P55 platform". `We did not see such things as often before the i7 era; there were a few targeted at the NVIDIA chipsets, but that's about it.

    So what happens to the other platform which also has a dual-channel memory sub-system - the AMD AM3 platform? There are also some kits designed specifically for that plaform, but let's just say they are much less in number...

    Did you even LOOK at the benchmarks in this article? They are almost identical - just like I said....and this ALSO is not a motherboard benchmark.


    AMD has enjoyed the performance benefits of an integrated memory controller for more than twice as long as Intel. And yet, it seems that Intel sets today's standard for acceptability. When Core i7 (the Bloomfield version) launched with 1.65 V as its recommended maximum voltage, manufacturers divided their DDR3 DRAM across two product lines: higher-voltage AMD and lower-voltage Intel products.

    Both RAM work equally well, with IDENTICAL benchmarks. And this also does not include a motherboard benchmark showing your claim.

    So all you have done is link to a heap of irrelevant information, and one outright false piece of info.

    None of it which supports your initial claim:

    "My personal opinion is that motherboards matter way more (and have, for about ten years) than the individual components that go into them." Which is still utter bollocks.

    This is completely untrue, and this fact is born out by the thousands of benchmarks done using different motherboards (and IDENTICAL COMPONENTS). Nice try at baffling with BS, though? Sorry you are incorrect. You have "informed" some, though....

    EDIT: Please show me the benchmark review where they prove that motherboard choice is the most important selection for smooth gaming. Because I know it doesn't exist. The graphics card is - obviously. Wow. I can't believe how much effort people can put into supporting non-factual statements - then the equally tortuous process of debunking it.

     
    Last edited by a moderator: May 3, 2013
  14. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    .There are tyre models that take, literally, 10s (notice the "s" at the end) of "supercomputers". :)

    Further proved my point. DCS, which is, by far, the most superior consumer flight sim, is a "single digit frame pig no matter what card you run".

    That's because it had NOTHING TO DO WITH Nvidia cards vs AMD cards. Yes, I know, that is what this thread is supposed to be about, but It was meant as a direct response to the poster that was comparing RF2 and IRacing frames.

    Before talking the way you do, I believe that you should read and research more.
     
    Last edited by a moderator: May 3, 2013
  15. williang83

    williang83 Registered

    Joined:
    Sep 28, 2012
    Messages:
    287
    Likes Received:
    153
    Spinelli, you keep pushing about the tyre models and keep saying that there are tyre models that take 10s to be calculated on super computers and bla bla bla. Well sorry but this phrase is so wrong in many levels.
    When i was finishing my university i had this amazing opportunity to see with my eyes one of their external project which was a project done in the 90's for pirelli, which was the manufacturer for gran turismo class. The software had the duty to simulate and in the end, by crossing the result with all historical data and statistic, the software had to suggest the best chemical formula for a given race weekend.
    Let me tell you this, i saw with my eyes even some important algorithm exactly in that area so i know how complex they are and trust me rfactor or iracing or whatever don't come even close to it, so it is not worth to compare and keep saying like you are doing.
    Specially comparing a flight sim with a racing sim just to emphases your point , they are completely different in all point of view and to be honest in term of IA and core engine well, that 90's game kick ass of most recent game (rfactor 2 included). The same goes for grand prix 4, it is incredible that someone cannot even replicate some old games behavior.
    You can't say that oh my god since a tyre algorithm can be so complex then rfactor 2 tyre algorithm is automatically complex (this is how you sound like). I could say the same since a combat flight simulation algorithm can be even more but much more complex then falcon must be complex, but i won't because i know that i would sound not so smart.
    By the way, about dcs and falcon the first one has a better collision and others stuff that must be better due to 10+ years of difference but still falcon have a better core engine and a better weapon behavior fidelity (which is what really matter for most ppl). So if something is newer it doesn't mean better in all sectors. The same goes for rfactor 2, it is better at some levels but at others it still have a lot to improve and it doesn't matter how evolved is the tyre engine, it won't never be that complex not for general purpose computers, it is pretty simple that the rendering engine sucks.
    For last but not least there is nothing to be proud about when you compare a 90's game with a modern game and this last one is not superior in ALL aspect specially with 10+ years of difference.
     
    Last edited by a moderator: May 3, 2013
  16. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Exactly. That's the point I was trying to make. Tyre models, like industrial ones, formula 1 teams' models, etc etc are wayyyyyyyyyy more complex than our sims, they have many supercomputers powering their physics systems. That's exactly my point. I was saying that a person can't say something so generalized like saying that RFactor 2's and IRacing's tyre model are both very, very similar complexity wise, computer processing intensive wise etc, because tyre models, like those of industrial grade (f1 teams, tyre companies, etc) are so unbelievably complex and computer intensive, and are sooo much more advanced than our "primitive" video game simulations that there is sooo much further that our video game sims still have to go. Therefore, just as an example, IRacing's model may only be 1.3% as complex and computer intensive as an industrial grade model, while RFactor's may be 3% as complex and computer intensive as an industrial model. That difference of 1.3% and 3% would be pretty big in this case, and would mean the RF2 model is much more complex/computer processing intensive than the iracing model, therefore lower framerates, relative to IRacing should be expected.

    Again, just an example. Not saying this is the case, but you can't just go on and say it's not the case either. After all, physics systems are the primary objective for most sims (but unfortunately not for all), and our home consumer ones are sooo far off from industrial ones, so it's impossible to know how much further/complex each video game sim's physics systems are relative to eachother.

    I wasn't comparing flight sims to racecar sims, just mentioned that DCS is much more technically complex and demanding than Falcon 4.0, that's all.
     
    Last edited by a moderator: May 4, 2013
  17. Minibull

    Minibull Member

    Joined:
    Sep 28, 2012
    Messages:
    1,556
    Likes Received:
    18
    It's all rather complex XD

    What I am looking forward to is when ISI release their proper details about rF2. Maybe when it is a bit more finished and finalised. Where they talk about their tyre model, what it does, how everything interacts with the temp, humidity, weather, engine, etc.
    I remember someone asking about glitchy looking tyre deformation in replays, and one of the ISI programmers (cant remember his name, sorry) said that their tyre model is calculating SO much data in real time that the replay files would be monstrous in size. Still, hopefully they will produce some form of details regarding their systems at some point :)
     
  18. Nand Gate

    Nand Gate Banned

    Joined:
    Jan 21, 2012
    Messages:
    471
    Likes Received:
    12
    New BETA driver and Application Profiles out. See OP
     
  19. Nand Gate

    Nand Gate Banned

    Joined:
    Jan 21, 2012
    Messages:
    471
    Likes Received:
    12
    New BETA2 I missed. See OP.
     
  20. TIG_green

    TIG_green Registered

    Joined:
    Jul 21, 2012
    Messages:
    3,038
    Likes Received:
    44
    Have anybody tested this yet? Any performance improvements?
     

Share This Page