Continuting performance issues in games on Sapphire Radeon HD 7970 OC

mesopotamian

Distinguished
Apr 10, 2011
72
0
18,630
I wrote to this fourm back in the end of November about issues I was having with the graphics card, a Sapphire Radeon HD 7970 OC Dual-X in a first time pc build. You can view my thread with the following link…
http://www.tomshardware.com/forum/forum2.php?config=tomshardwareus.inc&cat=33&post=379628&page=1&p=1&sondage=0&owntopic=1&trash=0&trash_post=0&print=0&numreponse=0&quote_only=0&new=0&nojs=0
At that time I realized that I was not getting the frame rates I should be getting in games such as Crysis 1 & 2. In addition certain programs such as CPU-Z and the Steam client displaying incomplete and/or incorrect information about my video card raised my suspicion of a possible hardware issue. A response I received from the Steam community fourm suggested that I should not have to be concerned about the issue regarding the graphics card information being displayed by Steam and CPU-Z. I had not yet used FRAPS or knew the developer console commands to display FPS when I wrote my original post about my performance woes. Doing so in Crysis 1, Crysis Warhead, Crysis 2, and Battlefield 3 yielded highly disappointing results. In the single player portions of all four games at max graphics settings 1920x1200, I was only getting within the range of 30 fps. In my original post, it was suggested that I get into contact with Sapphire from whom I received a replacement graphics card only to not see any difference in performance. If the graphics card is not bad, the other possibility that was brought up was a bad or failing PCI-E slot on my ASUS P8Z77-V Pro motherboard. Thankfully, I found out that I did not have to replace my motherboard, at least not yet. I went into the Catalyst Control Center and switched my Anti-Aliasing method from Supersampling to Multisampling resulting in a massive performance increase. The Battlefield 3 campaign (I have not yet tried the multiplayer), runs at a brisk and buttery smooth frame rate at maximum settings with 4x MSAA and 16x AF at 1920x1200. On the level “Operation Swordbreaker,” the frame rate usually hovers in the 60s or 70s, goes as high as the very low 90s and drops in one section into the 50s. Despite the performance I am getting in BF3 campaign, performance in the Crysis series remains disappointing. Right now Crysis 1 shows fps in the 40s most of the time, will sometimes dip into the 30s, and may go as high as 60 fps during gameplay. The level “Second Chance” in Crysis 2 at very beginning the runs buttery smooth with 60+ fps until I step outdoors where fps then hovers in the 40s most of the time. Both games are running at max settings with 8x MSAA and 16x AF at 1920x1200. Benchmarks from qualified online publications indicate better performance than what I am achieving. This is further compounded by youtube video from users claiming to be getting higher fps in the crysis games. I am not entirely sure of BF3 either. Turning Down MSAA to 4x did hardly anything to improve performance in the Crysis series. I have not tried multiplayer for the Crysis games or BF3 so I don’t know the fps I would be getting on a very large map with 64 players in BF3. If you read my previous thread I also took note of performance issues I was having in ancient Source engine titles like Half-Life 2 (One section in Ravenholm showed a significant performance drop). Again disabling supersampling alleviated the problems however it seems odd that a game as old as HL2 would have drops in frame rates on a modern high-end PC. With 8x SSAA fps in Left 4 Dead 2 at max settings and 1920x1200 dips into the 40s. However, fps in Left 4 Dead 2 is most likely right within the ballpark when I looked at benchmarks for the game in a Tom’s Hardware article on Anti-Alising modes. I have similar results with Portal 2. In the end, I am convinced that I am not getting the performance in games that I should be getting on my PC setup. This especially pertains to the crysis series but I am not completely sure about the others I have mentioned and others I have not. I need this community to help me determine whether I am getting the proper performance and if I am not, the root cause of the problems must be determined. BIOS and graphics card drivers are both up to date.

My specs are as follows

Intel Core i7 3770k
ASUS P8Z77-V Pro
Sapphire Radeon HD 7970 OC 3GB
16GB GB Corsair Vengeance LP
 

hero1

Distinguished
May 9, 2012
841
0
19,060



Please do feel free to use paragraphs next time, it makes it easier to read.

On to your issue, I think you are indeed, as you said, setting up your games too high. In some games a single GTX 680 or HD 7970 can't max out even at 1080p. You are gaming with a 1920x1200 resolution and you are expecting constant 60FPS when you have everything at Ultra or Very High? That's impossible. If you truly want to get the performance you think you deserve then you need to replace your monitor with a 1080p or get a second card. And if those are not good options then I suggest you don't get at ultra settings or very high settings because then you are going to see frames dip to mid 40s and mid 30s.
 

redeemer

Distinguished
Well what you need to do is reset your catalyst setting back to default, seems you have been enjoying a lot of 8X MSAA etc. Crysis 1 isn't very well optimized furthmore it is a DX9 title. Running 8X MSAA in Crysis 1 is just ridiculous you need to relax on your settings a little. Your BF3 fps seems about right, the 7970 is a very capable card no doubt but it has its limits
 

mesopotamian

Distinguished
Apr 10, 2011
72
0
18,630


I have tried running Crysis in only 1080p, and dropping down AA with a negligible boost in performance. The youtube videos I have linked below claim to be achieving better performance than what I am getting at the same graphics settings. Yes some of these users claimed to have overclocked their hardware. But I would think that I should be getting a similar range of fps as they are at 1080p.

http://www.youtube.com/watch?v=Df0lxJTVbFc

http://www.youtube.com/watch?v=B-XGYNaLsUY

http://www.youtube.com/watch?v=9XHjUWztgzs
 

hizodge

Honorable
Nov 22, 2012
752
0
11,160
To be honest, I leave everything to either 'Use application settings' or off on the Catalyst Control Center settings and use the ingame options instead. The only time I use it is when a game is missing anti-aliasing, which thankfully doesn't happen often these days. The only game I can really think of is Spec Ops: The Line, also using it on Tomb Raider at the moment as it makes you pick between FXAA (which is dirt cheap performance-wise, but blurs the entire image) and supersampling (which is very performance heavy, first renders the frames in 2X, 4X or 8X your native resolution and then downscales them). As there's no middle ground like MSAA or SMAA, I just typically go with standard non-morphological 4xEQ.

Another thing; 8X MSAA is a total waste of performance. In 99% of all the games I've played with MSAA to this date I still haven't seen a game where the 8X MSAA had any visual improvement against 4X, unless you zoom in 100x on some edges or something. Supersampling on the other hand is nice if you can afford it, again - 4X vs. 8X; no gain here, just pain. This setting is a setting used mainly for anti-aliasing still images and it will basically kill your frametimes when even enabled at 2X in newer games. You just stay away from Supersampling, you hear?
 

mesopotamian

Distinguished
Apr 10, 2011
72
0
18,630


I have for the most part have tried everything you have mentioned. I do not use supersampling anymore, and have been using in game settings in Crysis from the get go. With some clues found on the web such as the youtube videos I have linked in the previous post, I am skeptical that I am getting the performance I should be getting in Crysis at max graphic settings and the resolution afforded by my monitor. The users in all three videos claim to be getting performance above what I am achieving during gameplay. All three users play at 1080p but I do not think that performance at 1920x1200 would be much slower and probably not noticeable. That said, I have to try BF3 multiplayer for the very first time (especially a big map with 64 players) although performance in BF3 campaign seems to be right within the ballpark of where it should be.