[CPU/GPU] - Performance

Discussion in 'General Technical Support' started by Irathi, Dec 7, 2012.

  1. Hello

    This thread is intended for sharing and comparing your computers utilization of the CPU / GPU and the fps achieved. Please keep the thread civilized and clear of any fanboyism, nobody cares about your opinion of AMD/Intel/Nvidia. Let us try to make this a thread helpful to the devs where they can see actual performance for everyone out there, so do not troll this thread - you will be reported.

    All posts must include:
    • CPU, model name and settings (speed)
    • GPU, model name and settings (core clock, memory clock)
    • Ram, model name and settings (timings, speed, size)
    • Windows version
    • In game graphic settings.
    • Whether the in game counter was [GPU] or [CPU] limited. (enable it with alt+F)
    • Proof of the gaming sessions utilization and fps output, preferably a screenshot or movie. It is important that you do not simply post "my fps was xx and my cpu was xx". Picture or it didn't happen!
    • A description of the ongoing battle when the performance test was conducted. Preferably location and small/medium/large/extreme fight.
    If you are not able to supply the above, please refrain from posting in this thread as it will be regarded as noise. I would also like to remind once more, do not troll this thread and do not start with fanboyism. Also do not deal out harsh criticism to people trying to post their results, try rather to guide them on how to make better posts, not everyone knows what Aida64 is.

    Below is an example of a session i ran today. All settings can be viewed in the screenshot and supplied with my signature.

    Feel free to sticky this.
    • Up x 1
  2. Today i played for about 2 hours, logging more or less the same results all the way. The current screenshot is of a full assault on Eisa Tech Plant on Esamir. I would define it as Large, people everywhere, vehicles shooting at you all the time, couple of aircraft.

    Click to see a larger image, i use imgur to host the pictures.

    Settings in game: All high except shadows.

    [IMG]



    The game appeared to be [CPU] bound at all times during action, it would turn [GPU] bound indoors with few team members around. Outside it would vary from 35-65 frames per second - CPU bound. Indoors i could reach 80-100 and it would start to appear GPU bound.

    The huge spikes in FPS where it drops to zero is where I alt+tabbed into Windows to have a look at the log.

    [IMG]


    It appears that 2 cores are running at around 60%, one core is running around 40%. The rest are below 30%. This is old news, but yet again confirms that the game poorly utilize the available cpu power. As we can see the GPU rarely saw more than 50% utilization and the system ram never went above 3,66 Gb use.

    All my cores are so called "unparked", in case anyone was wondering, I saw no noticeable difference in performance by doing this. Which is kind of logical as the game barely mange to use the cores available before I unparked the idle ones.
  3. CPU: AMD FX-8150 (3.9GHz, disabled Turbo Boost, further OC to come, I just played with it for 10 minutes)
    GPU: EVGA GTX-680 (Stock)
    RAM: G.SKILL F3-2400C10Q-16GTX Trident X 16GB 4X4GB DDR3-2400 CL10 240PIN 1.65V Quad Chann
    OS: Windows 7 Home Premium x64
    Settings: High, Max FoV
    FPS:
    - In large fights: 15-20 FPS [CPU]
    - In medium fights: 20-25 FPS [CPU]
    - With nobody around: 50-60 FPS [CPU]

    I have a G15 keyboard with the EVGA utility running on it and it usually never go higher then 40% of GPU usage. It's clearly the CPU that is limiting my system.

    I'll edit with screenshot when I can.
    • Up x 1
  4. Everything you need is in the sig. There is one issue with what you are asking everyone to do however, it has been pointed out by dev many times, the resolution (in time) of the graphs we all use for CPU and GPU are not low enough to be accurate enough for the devs to use. The best way to help them is to simply give hardware and driver info with a format like Manhim used for FPS, different fight sizes with FPS and results of the in game alt+f CPU/GPU meter.
  5. My results are very identical to DonnyD using a Phenom II X6 1090T @ 3.6GHz but I'm using a 5770 which is about 3 generations behind his, but the 7850 is in the mail.

    I don't see [GPU] that often since the last patch, I used to see it all the time at warpgate and out int eh open but it doesn't even come up anymore.

  6. Hello and thank you for replying to the thread.

    Could you please include the following in any new posts:

    • In game graphic settings.
    • Whether the in game counter was [GPU] or [CPU] limited. (enable it with alt+F)
    • Proof of the gaming sessions utilization and fps output, preferably a screenshot or movie. It is important that you do not simply post "my fps was xx and my cpu was xx". Picture or it didn't happen!
    In case you are not happy with your current fps "5-15" I would also like to inform you that your current GPU is below the minimum requirements for the game. Source: http://help.station.sony.com/app/answers/detail/a_id/36590
    (ctrl+F and type in "540m" to find your card easily)
  7. Specs: sig

    Settings:
    [Display]
    FullscreenRefresh=0
    Maximized=0
    FullscreenWidth=1366
    FullscreenHeight=768
    WindowedWidth=1024
    WindowedHeight=768
    RenderQuality=0.750000
    Mode=Fullscreen
    FullscreenMode=Fullscreen

    [Rendering]
    GraphicsQuality=1
    TextureQuality=1
    ShadowQuality=0
    RenderDistance=1005.000000
    Gamma=0.000000
    MaximumFPS=120
    UseLod0a=1
    VSync=0
    OverallQuality=-1
    LightingQuality=1
    FogShadowsEnable=0
    EffectsQuality=1
    TerrainQuality=0
    FloraQuality=0
    ModelQuality=3
    ParticleLOD=0
    MotionBlur=0
    ParticleDistanceScale=0.650000

    RenderFlora=Off
    --------------------------------------------------------
    Game is CPU limited, http://imgur.com/qNgUk
    i can't find my screens from game :/
  8. Ok i saw a MAJOR fps increase today. In fact the in game meter would now show me that i am GPU bound more often than CPU bound, even in larger fights.

    Granted the fight was not AS big as the one in my previous post, however it was a good one. Most importantly, the gameplay felt a lot smoother.

    Note if you are comparing the two posts that the fps graph has been increased from max 100 to max 150. This was due to the fps being well up in the hundreds in small fights.

    I ran Fraps during the medium/large fight and the results where astonishing compared to earlier. With the same settings i recorded:
    Frames Time (ms) Min Max Avg
    7630 128981 44 83 59.156

    That is up from average 35-40ish! GREAT JOB SOE! I will send you'r code monkey that christmas banana.

    There still seem to be 1 core taking the major load, but notice the other ones are much more active now. Especially the bottom two in the center. The GPU usage is also a LOT higher, resting at around 70% and sometimes bumping up to 99-100%.
    [IMG]

    my settings where untouched from the last time.
    [IMG]
  9. CPU: i5 2500k @ 4.8Ghz (OC)
    GPU: Asus GTX 670 DC II TOP; GPU Clock: 1228Mhz, Mem Clock: 7182Mhz
    RAM: Make - I forgot. Speed: 1866Mhz, Timings: 9-10-9-28-2
    Windows 7 Ultimate 64 bit
    CPU Bound, almost always. So the massive OC isn't helping much apparently.
    I'll quote this post later when I have the proof you apparently want.
    Game runs on High except fog shadows.
    FPS is usually very high (90+ ish), except when in large/extremely large fights, amp/tech plant where it drops to 30 ish, which I can live with I suppose, but I'd rather be in the 50'ish range.

    Ps. OC is stable
  10. Ok, I'm currently not at home so I have no screenshots available.

    Recently people complained about worse fps after 13th of jan (was it 13th?), personally I have not noticed an fps drop for my own computer. I played the game 2 nights ago and found a good sized battle with enemies "everywhere", tanks, aircraft, infantry and all.

    It might have been a bit slower than before, I saw lower 30's like 33-34 fps a couple of times, but it was a decent battle after all.
    My settings where untouched and overall I felt I had more or less the same performance as before. However I did notice a higher latency than what I'm used to. Maybe people are mistaking lag for low fps.

    Idk, the game feels the same to me, awaiting 30th jan patch with anxiety :)
  11. Intresting, I have basically the same processor (8350). Yet a somewhat weaker graphics card a 650 GTX oc'ed to 1.2. And often dip int othe high 20'ies in the massive battles on miller.

    I've been experimenting around with the game settings (everythign on medium-ultra) and the only real difference seems to be changing the maxium FPS. IT all runs about high 30-40'ies in most battle situations. Soon as zergs get involved though, its dipping down to 27-28 on a fairly regular basis.

    The two variables that stand out then, I wonder which is the more important?
  12. How fast is your CPU running? I keep mine at 4.53Ghz when I play, sometimes I OC up to 4.7 and 4.8, but all the results here where done at 4,53ghz

    Btw we have the exact same processor if they run at the same speed. AMD only produce 1 chip. They increase the speed on the FX-8350 and they cut of 1 module on the FX-6300 and 2 modules on the FX-4300.
  13. I'm only on stock cooling, so at the moment 4.3 is as far as I dare push it.

    Yeah, thus the 'basically'. I mostly got it as it was on sale and not that ambitious an overclocker.
  14. I can really recommend the Noctua NH-D14 if you ever consider trying it or simply want a silent computer. I can not hear my CPU fan while playing BF3, but I don't OC for that game. In PS2 it becomes audible at 4.53Ghz, but just a nice swoosh. Nothing like the stock cpu cooler.
  15. Thank you for the tip, i'm looking into it, my case is not the biggest though (mirco atx board) so that might be a bit big. But a more reliable cooling solution is a must.

    Though sying that i've been playing around with it since, and there is a discrepancy of allmost 10c (i.e massive) at times between what AMD overdrive reports are the CPU temps and what Razergamebooster and Asus AI suite report.

    Odd....

    If not for the fact that at 4.3 I'd thought I was touching 60c when running PS2 i'd overclock more.
  16. Im at a meeting atm (yes, taking job vey seriously... zZzZz) so dont have screenshots with performance curves.

    But my setup is as follows:

    i7 920 c0 @ 3,8 ghz 1.2v (using coolermaster v8 cooler, 50-55c during gaming)
    2x stock Msi gtx 680 cards sli
    Ocz platinum tripple channel 6gb @ 1444 mhz, 7-7-7-21 timings
    Asus p6t deluxe rev1
    Ocz agility 4, 256 gb (on which the game is installed)
    WD cavier black 1 tb
    Creative soundblaster pro soundcard
    Thermaltake 1200w psu
    Thermaltake armor+ case, highly modified, with extra fans, and improved airflow
    Asus PB278Q screen

    Windows 7 64 bit
    Using latest nvidia drivers 307.90

    Using max ingame settings, with 2560x1440 resolution and then textures set to 0 and graphics to 4 in userconfig.

    Fps (with v-sync disabled - im usually using adaptive v-sync) is as high as 120 when just standing in unpopulated areas, and low as 30 fps in the huge battles... but im cpu limited all the time.
    At 60 fps (what id have at the warpgate, with tons of people there) the gpu load is about 50 %. At 30 fps the gpu load is roughly 40%.

    I should probably add that in all dx11 titles, varying from bf3 to shogun 2, my computer is actually gpu bottlenecked... coincidence? No, because in all dx11 titles, the workload is spread evenly across all threads on my cpu, so when my gpu's are maxed out, the cpu is at roughly 80% load on each thread. In planetside 2 however, the load is 100% in thread 1, but only 20% on the 3 other cores. I blame it on the poor choice of making ps2 dx9 exclusively.
    • Up x 1
  17. You forgot to mention the amount of ram.

    Btw, i find that the speed of core 1 is the only thing that really matters for this game - so if you could make a stabile OC at above 5 ghz with core 3 and 4 disabled, i think youd see a performance increase.
  18. Yes, 100% on thread 1, 80% on DirectX thread I think, and also other thread like input thread, sound thread etc.
  19. I remember when Battlefield: Bad Company 2 came out, core #4 (on quad cores) always hit 100% and we found out that it was the audio thread maxing out that one core as turning the audio volume to 0 got rid of the 100% CPU load. DICE optimized the Frostbite 1.5 engine and the game runs better now, Frostbite 2.0 for BF3 is one of the best out there now.

Share This Page