Not all PC user can max out the game. My point is that majority are on DX11 ( Even AMD users) , and the game is only available for PC ,therefore, they have to make sure that it is enjoyable for majority of PC users day 1. As i told again and again that DX12 is a overhead at the moment will be treated like a second citizen until the majority shifts to DX12.TIL 80% of users are using 1060 or greater GPUs...
Its ridiculous that you guys are defending these results. You are making these forums go downhill
Not all PC user can max out the game. My point is that majority are on DX11 ( Even AMD users) , and the game is only available for PC ,therefore, they have to make sure that it is enjoyable for majority of PC users day 1. As i told again and again that DX12 is a overhead at the moment will be treated like a second citizen until the majority shifts to DX12.
I didn't say anything about DX12. I said the current version is running terribly. Look at the actual graphics there is nothing special going on here and massive frametime spikes even on the 980 ti / 1070 / 1080 they showed in their benchmark video.
Civ V was even between AMD and Nvidia, this engine has recessed.
Actually, it's always favored Nvidia. Likely due to the game being very CPU intensive when zoomed out, which is the only time performance is an issue. All you have to do is play like normal people do, more zoomed in, and in the late game, either switch a couple settings, or use a different view mode.
You also seem to be out of your mind if you think you can compare the graphics in this genre to just about any other. Just because it seems less extraordinary to you, does not mean it's not demanding. The way the graphics are presented will not lend itself to what you deem great graphics, yet it still takes a lot of power to do.
And I have to ask, do you play Civ games, or similar games in the genre?
I'm not sure what I'm supposed to take from that. It's using a GTX 650, at the highest settings, including 4k x 4K resolution overlay. Of course it won't do great.Game is extremely unoptimized,or at least the benchmark is,around and over 20% for the driver is ridiculous for an AAA title (for example metal gear 5 only uses around 5%) ,the heavy stutters (well almost lock ups for me) are accompanied by complete drop of gpu load so my guess is that new data is being loaded into the gpu,for a new segment of the bench maybe?
Anyway,very weak showing for a Civ game and the fact that nvidia cards can brute force through it definitely is no excuse.
I don't see the graphs of AMD doing as good or better in the thread. You only showed a CPU benchmark. The only time I've ever seen AMD do as good or better was when Mantle was used, which isn't an apples to apples comparison.I've shown benchmarks proving otherwise
Yes I play all types of games, FPS, MMO, RPG, RTS, 4x and Sim as well.
There is no excuse for the poor performance in this title, it is purely unoptimized.
Also, by definition shouldn't we say nvidia drivers have higher overhead? the main reason they get around the dx11 limitation that hits AMD is because their driver does more on the CPU cores to circumvent it. CPU usage seems to be regularly higher for nvidia GPUs than AMD GPUs.
Overall CPU usage on NVidia may be higher yes, but it's spread across more cores/threads due to DX11 multithreading, a driver feature that AMD does not support. So with AMD, only a single thread is being used for rendering with DX11, hence why the Radeons are more CPU limited than NVidia.
That said, there is still obviously some CPU limitations going on with NVidia as well, despite the use of DX11 multithreading. This goes to show that DX11 multithreading had limitations when it came to scaling across CPU threads.
Luckily, DX12 addressed all of these limitations though..
Quite sure this is incorrect....free upgrade finished end of July 2016
The Fury ran out of memory. 4GB is laughable, hence the laughable sales of the Fury range.
TIL 80% of users are using 1060 or greater GPUs...Its ridiculous that you guys are defending these results. You are making these forums go downhill
Not all PC user can max out the game. My point is that majority are on DX11 ( Even AMD users) , and the game is only available for PC ,therefore, they have to make sure that it is enjoyable for majority of PC users day 1. As i told again and again that DX12 is a overhead at the moment will be treated like a second citizen until the majority shifts to DX12.
It might be all fine and dandy that AMD might get better performance with DX12 and is probably true,but ultimately you can only run the game in DX11 now and all the reviews show Nvidia is doing better,so really it is a loss for AMD. They really needed to try and get performance up to scratch as quickly as possible. They cannot afford lapses like this if they wish to gain marketshare from Nvidia.
Actually, it's always favored Nvidia. Likely due to the game being very CPU intensive when zoomed out, which is the only time performance is an issue. All you have to do is play like normal people do, more zoomed in, and in the late game, either switch a couple settings, or use a different view mode.
You also seem to be out of your mind if you think you can compare the graphics in this genre to just about any other. Just because it seems less extraordinary to you, does not mean it's not demanding. The way the graphics are presented will not lend itself to what you deem great graphics, yet it still takes a lot of power to do.
Completely wrong
FX8350 + HD7950
I fail to see how Ryse has anything to do with what we're discussing I never said that AMD's drivers were restricted to a single thread for all games now did I? Also, CryEngine 3 uses IC manual multithreading much like the other big 3D engines. IC manual multithreading is much harder to implement than deferred context, which is done in the driver..
If AMD had done a better job with their DX11 driver, then deferred context multithreading would have been much more prevalent..
Most studios have said that DX11 deferred contexts don't work for their use cases, Dice even went so far as to say that it's broken. Civilization is one of the few exceptions where the devs have gotten benefits out of it.
I fail to see how Ryse has anything to do with what we're discussing I never said that AMD's drivers were restricted to a single thread for all games now did I?
Overall CPU usage on NVidia may be higher yes, but it's spread across more cores/threads due to DX11 multithreading, a driver feature that AMD does not support. So with AMD, only a single thread is being used for rendering with DX11, hence why the Radeons are more CPU limited than NVidia.
I don't see the graphs of AMD doing as good or better in the thread. You only showed a CPU benchmark. The only time I've ever seen AMD do as good or better was when Mantle was used, which isn't an apples to apples comparison.
Edit: I followed the link of the charts you showed, and you do realize you are comparing a 580 to a 7970, with the FPS being within 5%. Given they are different generations of video cards, that shows Nvidia's performance in a much better light than AMD.
Overall CPU usage on NVidia may be higher yes, but it's spread across more cores/threads due to DX11 multithreading, a driver feature that AMD does not support. So with AMD, only a single thread is being used for rendering with DX11, hence why the Radeons are more CPU limited than NVidia.
That said, there is still obviously some CPU limitations going on with NVidia as well, despite the use of DX11 multithreading. This goes to show that DX11 multithreading had limitations when it came to scaling across CPU threads.
Luckily, DX12 addressed all of these limitations though..
I dont see you specifically only talking about Civ, it seems to me you are talking in general.
Those two cards were within a few % of each other in general as well. The point is, they were equal in the game and Nvidia hasn't always been better. RS provided even more context for this.
AMD doesn't support driver command lists but it does support all of the other DX11 multithreading capabilities such as deferred command lists. The difference between the two is the first requires the drivers to handle turning single threaded calls into multithreaded ones where the second has the developers handling the multithreading capabilities.
Regardless we've seen that all other Civ titles are actually even. Anyone can see that a 1060 should never be beating a Fury and 480 but huge margins. Would you accept a game that showed Fury beating 1080 by 20% as well optimized? No, no one in their right mind not trying to push an agenda or paid to think so would.
Didn't Civilization B.E support Mantle? If so, then obviously Mantle would have removed the CPU bottleneck for Civ B.E..
Obviously AMD's drivers can do multithreading. I never suggested otherwise. I was talking specifically about deferred context rendering, which only NVidia supports. I'm not even sure that Intel's drivers support it either..
Optimization is a relative concept. It's not an on or off switch. It's possible for a game to be more optimized for a particular architecture or vendor, than another. My guess is that because Civ VI will support DX12, Firaxis didn't bother tuning the IC rendering path as much as they did in previous games, which put AMD at a disadvantage out the gate compared to NVidia which supports the DC rendering path..
When DX12 finally becomes available, both vendors should benefit greatly from using it though..
The benchmarks shown didn't use it
I was specifically talking about deferred command lists in my post, and AMD does support them
I know optimization isn't on/off, but its clear they didn't spend much time optimizing this engine as it is running terribly on all hardware (everyone ignoring the 980 ti / 1070 / 1080 in their benchmark video running terribly), but especially bad on AMDs.
I have played most the Civ games, I did not buy Civ BE, so forgive my lack of knowledge on that one. And if Carfax83 is right, then Mantle is the only reason for it. This game doesn't have Mantle.I am correct and hence why I posted it on purpose to remind PC gamers to upgrade while it's still free. I just installed a new system last week and upgraded from Windows 7 Ultimate 64-bit to Windows 10 Pro for free (including the Anniversary Upgrade). I plan on building 1-2 more systems this year and putting Windows 10 on them while the offer still stands. The only requirement is Windows 7 SP1 or Windows 8.1 (from 8) and certain versions are excluded:
Windows 10 free upgrade for customers who use assistive technologies
http://www.howtogeek.com/265409/you...-for-free-from-microsofts-accessibility-site/
This explanation does not make sense. First, 1060 3GB performs better. Second, had the Fury ran out of VRAM, it would have been slower than RX 480 8GB.
The argument of the developer targeting 80% of PC users (aka NV owners) has at least 3 problems:
1) "Going by Mercury Research's information, AMD increased its share of total discrete GPU sales to 34.2% of the market (by unit volume) in the second quarter of 2016" --> That means NV's market share is closer to 75%, not 80% and AMD has been gaining market share every quarter. The 80% figure hasn't been true for a while now and yet it's still repeated as fact. At least if one is going to quote market share, at least have up-to-date data.
2) It assumes AMD and the developer will not optimize DX12 to perform vastly better than DX11. It could be a strategic choice by AMD to focus all of its efforts on DX12 performance for this title. It would be wrong to blame the developer for ignoring AMD's GPU optimization if it was AMD themselves who chose this route. Therefore, claiming that the developer just focused on 80% of the dGPU market share is misleading if we actually see vastly improved DX12 performance with AMD down the line.
3) It assumes that 75% of NV users have modern GPUs and most importantly are not CPU bottlenecked first. Civ 5 was ridiculously CPU dependent:
I highly doubt that these "75% of NV-owned GPU PCs" gaming market who will play Civ 6 have a CPU as fast as even an i7 920 @ 4.0Ghz. That means, there is actually a large chance that with slower CPUs, all of AMD/NV GPUs in TPU reviews could be CPU bottlenecked under DX11. This is why we need more professional review sites to test CPUs in this title. In other words, the GPU perfomrance per TPU could be vastly different from real world performance because TPU used an i7 6700K @ 4.5Ghz.
Steam shows that Windows 10 is at 48.69% (Windows 10 64-bit at 47.28% and Windows 10 at 1.41%). The remaining portion 95.32% - 48.69% = 46.73% are on Windows XP/7/8/8.1. That means, at least according to Steam, the majority of Steam gamers are using Windows 10, hence can benefit from DX12 in Civ 6. You also ignored my point above about potential CPU bottlenecks for this title for the majority of Civ 6 player base. Do you honestly think they are sporting i7 920 @ 4.0Ghz, i5 2500K @ 4.8Ghz/i7 3770K and so on? RX 480 is getting 49 fps at 1440p. Are you suggesting that's bad performance for a turn-based 3d board game now?
That's a fair criticism. I agree. As far as marketing is concerned, AMD has lost in supposedly an AMD GE title (!) until DX12 shows up and possibly addresses the CPU bottlenecks under DX11. I care a lot more about the future of DX12 than NV or AMD winning in Civ 6. It's not a good sign at all when almost all of the top DX12 games launched in 2016 have either delayed DX12 code-path, launched with DX12 performing poorly or at best as good as DX11, or had good FPS under DX12 but horrendous frame times which defeats the purpose of DX12. It's really a bad start to the DX12 era and hopefully it does not negatively shape the future focus of AMD's and NV's 2018 GPUs. The only way for developers to move to the next generation DX12 API is to start learning and adopting. Seems like the previous transitions from DX8/8.1 to DX9 or DX9 to DX11, this one is going to take a while.
Not entirely accurate. Per AT's testing, in Civ BE, R9 290X and 980 were performing similarly at 4K and at 1440p, R9 290X was still up there with GTX780Ti. 7970 925mhz reference and 680 were roughly equal.
GameGPU shows similar results. Civ BE ran just as good if not better on AMD, on top of having SLI and CF support!
R9 280X deliveried > 30 fps at 4K, R9 290X delivered almost 50 fps, HD7990 (7970 CF) delivered 60 fps at 4K and R9 295X2 achieved 92 fps, amazing scaling across single AMD GPUs and CF.
That's fair but would you honestly say that Civ 6's graphics justify Fury X and 980Ti performing at 47.8 and 56 fps @ 4K, respectively, when R9 290X reference card got 49 fps at 4K in Civ BE?
Minimum frame rate in Civ BE on GTX780Ti and R9 290X at 1440p was almost 60 per AT:
There is one other comment from TPU that recalls the days of Civ 5 where tessellation hammered HD5870 series:
"Civilization VI uses a newer version of the Firaxis engine, with higher-resolution textures, higher terrain detail taking advantage of tessellation, " This could be another reason for poor AMD performance, on top of the DX11 CPU bottleneck. But then it doesn't explain the excellent performance of AMD cards under Civ BE.