Overclock.net › Forums › Graphics Cards › NVIDIA › Gpu usage not pegging in bf3 680 sli setup
New Posts  All Forums:Forum Nav:

Gpu usage not pegging in bf3 680 sli setup - Page 2

post #11 of 16
Thread Starter 
Quote:
Originally Posted by ViTosS View Post

Quote:
Originally Posted by wutang61 View Post

The whole thing about sli never going to 90+ is a load of crap guy. My 480's pegged no problem. And 101 fps avg is a joke for dual 680's. it's clearly a driver issue or a game patch problem.
Not maxing usage is ether a driver/game patch issue or a bottleneck. And a sandy bridge isn't a a bottleneck on two 680's.
And you can keep ati I had 2 7970's and the drivers were beyond wonderful..... I'm not a spend a grand and oh wait till they fix it kinda person.
Sent from my toaster with the 4g's and the wifis

When you had two GTX 480 was getting 99% usage on both EVERY game? I say that because a friend of mine with two 570 has the same problem as me, and in Battlefield 3 I doubt it you get more than 90% usage, multiplayer only

Yup. Every single game.

Gpus should run full bore until you hit a bottleneck. And 100fps is not the limit of my cpu.


Sent from my toaster with the 4g's and the wifis
post #12 of 16
Quote:
Originally Posted by wutang61 View Post

Yup. Every single game.
Gpus should run full bore until you hit a bottleneck. And 100fps is not the limit of my cpu.
Sent from my toaster with the 4g's and the wifis

What sorts of testing have you done on this particular game, at these exact settings, to determine that '100fps is not the limit of (my) CPU'?

BF3 in multi-player requires a significant amount of CPU power. And pumping out in excess of 100fps on ultra requires a *very* fast CPU.

My best guess based on the symptoms you describe is that you've reached the limit of how fast your CPU can do it's work in this particular gaming scenario.

Of course that doesn't rule out the possibility that better drivers or better game code might alleviate the issue (since the way the code of either driver or game is written can very much affect level of cpu-dependency), but I'd bet that if you did benches (using this exact gaming scenario) w/your cpu at various speeds you'd find that your FPS is quite sensitive to those changes ... and hence ... you're looking at some measure of CPU BN thumb.gif
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
post #13 of 16
I have the same problem with my 660 ti's. Every other game i play the usage will go above 85% the whole time, but bf3 only uses about 60%. Turning down different settings dont affect it and my cpu usage is only about 75%
    
CPUMotherboardGraphicsRAM
i5 3570k 4.8 Ghz ASUS SABERTOOTH Z77 atx MSI 660ti SLI Corsair Vengeance 1866mHz 16Gb 
Hard DriveCoolingOSPower
Mx200 250 GB, 1TB WD Black Corsair H80  Win7 Home Premium Corsair 850HX 
Case
Carbide 400R 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5 3570k 4.8 Ghz ASUS SABERTOOTH Z77 atx MSI 660ti SLI Corsair Vengeance 1866mHz 16Gb 
Hard DriveCoolingOSPower
Mx200 250 GB, 1TB WD Black Corsair H80  Win7 Home Premium Corsair 850HX 
Case
Carbide 400R 
  hide details  
Reply
post #14 of 16
i also have this issue but only with BF3, it doesnt max out my GPU or CPU its really quite bizarre; only thing i can think of is the game is reserving resources / threads for other operations like AI or direct compute operations. ive noticed this for a while but i have not been able to find a fix for it yet, frame rates are high enough tho so its not too bothersome considering it only happens in this game and no others.
Project EVA
(20 items)
 
  
CPUMotherboardGraphicsGraphics
I7 2600K Asus z77 sabertooth Gigabyte GTX980 Gigabyte GTX980 
RAMHard DriveHard DriveHard Drive
16gb patriot Viper 3 (8gbx2) Kingston Hyper X SSD Western Digital Black Western Digital Black 
CoolingOSMonitorMonitor
Noctua NH-D14 Windows 8.1 24" LG WP2442-BF BenQ XL2420Z 
KeyboardPowerCaseMouse
Razer Black Widow Ultimate Seasonic X1050 Silverstone FT02 Mionix naos  
Mouse PadAudioAudioAudio
Razer Destructor II Astro A40  Asus Xonar STX Swan M50w 
  hide details  
Reply
Project EVA
(20 items)
 
  
CPUMotherboardGraphicsGraphics
I7 2600K Asus z77 sabertooth Gigabyte GTX980 Gigabyte GTX980 
RAMHard DriveHard DriveHard Drive
16gb patriot Viper 3 (8gbx2) Kingston Hyper X SSD Western Digital Black Western Digital Black 
CoolingOSMonitorMonitor
Noctua NH-D14 Windows 8.1 24" LG WP2442-BF BenQ XL2420Z 
KeyboardPowerCaseMouse
Razer Black Widow Ultimate Seasonic X1050 Silverstone FT02 Mionix naos  
Mouse PadAudioAudioAudio
Razer Destructor II Astro A40  Asus Xonar STX Swan M50w 
  hide details  
Reply
post #15 of 16
Look at this review. They are pushing closer to 130 fps with a SLi system.--> http://benchmarkreviews.com/index.php?option=com_content&task=view&id=971&Itemid=72&limit=1&limitstart=6

Replicate that article and see if you get the same numbers. Set up your system with the same monitor resolution, same FOV, and in the Secure Parking Lot' stage of Operation Swordbreaker.

Set up your core clock and memory clock for the same settings.

If you can replicate their results (124 fps), then there is nothing wrong with your card or system. End of argument.

Also you said
"1 680 = 77fps
2 680's = 101fps
Something is wrong here"

Nothing seems terribly wrong. Did you expect SLi to double your framrate? That's silly, because SLi will never double your framerate. It's diminishing returns. Each video card isn't rendering half of the screen. It's a painfully complicated process of dividing up the work among the video cards, and it's entirely software dependent and terribly inefficient.
Deus Ex Machina
(19 items)
 
   
CPUMotherboardGraphicsRAM
i5 580M Dell 0R1203 NVIDIA Quadro FX 2800M Samsung 4Gb 
Hard DriveHard DriveOptical DriveOS
TOSHIBA MK3256GSYF  TOSHIBA MK3256GSYF  DVD-Rom HL-DT-ST DVD+-RW GU40N Windows 7 Professional 
Audio
Dell Computer 5 Series/3400 Series Chipset Fami... 
CPUMotherboardGraphicsGraphics
Athlon 64 X2 6400+  Asus M2N-E Nforce 500 SLi 8800GT  9800 GT 
RAMHard DriveOptical DriveOS
OCZ ReaperX WD LG lightscribe Multilayer DVD burner Windows 7 
PowerCase
Antec 500W Smartpower 2.0 Aspire X-Navigator 
  hide details  
Reply
Deus Ex Machina
(19 items)
 
   
CPUMotherboardGraphicsRAM
i5 580M Dell 0R1203 NVIDIA Quadro FX 2800M Samsung 4Gb 
Hard DriveHard DriveOptical DriveOS
TOSHIBA MK3256GSYF  TOSHIBA MK3256GSYF  DVD-Rom HL-DT-ST DVD+-RW GU40N Windows 7 Professional 
Audio
Dell Computer 5 Series/3400 Series Chipset Fami... 
CPUMotherboardGraphicsGraphics
Athlon 64 X2 6400+  Asus M2N-E Nforce 500 SLi 8800GT  9800 GT 
RAMHard DriveOptical DriveOS
OCZ ReaperX WD LG lightscribe Multilayer DVD burner Windows 7 
PowerCase
Antec 500W Smartpower 2.0 Aspire X-Navigator 
  hide details  
Reply
post #16 of 16
Quote:
Originally Posted by crimsontears809739 View Post

Look at this review. They are pushing closer to 130 fps with a SLi system.--> http://benchmarkreviews.com/index.php?option=com_content&task=view&id=971&Itemid=72&limit=1&limitstart=6

Replicate that article and see if you get the same numbers. Set up your system with the same monitor resolution, same FOV, and in the Secure Parking Lot' stage of Operation Swordbreaker.

Set up your core clock and memory clock for the same settings.

If you can replicate their results (124 fps), then there is nothing wrong with your card or system. End of argument.

Also you said
"1 680 = 77fps
2 680's = 101fps
Something is wrong here"

Nothing seems terribly wrong. Did you expect SLi to double your framrate? That's silly, because SLi will never double your framerate. It's diminishing returns. Each video card isn't rendering half of the screen. It's a painfully complicated process of dividing up the work among the video cards, and it's entirely software dependent and terribly inefficient.

It's really not *that* painfully complicated ... with SLI (aka AFR), the two cards just take turns rendering the image. Arguably the biggest 'complication' is in keeping the frame output properly metered. You want a pattern like this: F*****F*****F*****F*****F as opposed to FF**************FF***********FF. The latter denotes the output pattern associated with both 'microstutter' and 'very high scaling'. Because the load is constantly changing, without deployment of fairly complex algorithms to do the metering, it's easy for two cards to fall into that latter frame output pattern.

In the absence of BN's, SLI will normally provide 'scaling' on the order of 75-95%. The scaling in question here is only 31%. That is definitely not the results one would 'expect'. However, as I stated above, this particular gaming scenario requires a relatively high level of CPU power (BF3 multi-player with lots of players I mean), so I highly suspect that the reason for the abnormally low scaling number is ... CPU BN.

I don't know how many times I have to remind folks on the boards here that you cannot rule out CPU BN just by using the CPU 'usage' measurement in Task Manager and similar programs and observing that you're not at 100% usage (or near it). Those graphs are an ESTIMATE only. They are not a real-time, bare-metal reading of precisely what your CPU load is. A more accurate description of what you're looking at would be "This is what Windows would optimally like to be doing with this load'. But the CPU itself can have 'other ideas', and you won't know it from looking at the graph in TM or any other similar CPU usage graphing program (that I'm aware of).

Interesting fact ... back when Skyrim came out, there were certain areas of the game that were heavily CPU BN'd. The famous 'Steps of Whiterun' locale was where most people did their testing of the phenomenon. Well, I did extensive tests of the issue, and you know what I discovered? At the times when the game was being demonstrably CPU-bottlenecked, the CPU usage in Task Manager was *LOWER* than it was when the game was NOT being CPU-bottlenecked. And this was true whether you looked at combined 'core' usage, or looked at individual cores ... there was simply an across the board roughly 20% reduction in CPU activity when the bottleneck was in effect vs. when it was not.

The proper way to test of for CPU BN is to benchmark at your current CPU clock, and then bench again with a HIGHER CPU clock. You then divide the % FPS delta by the % Clock delta to come up with a ratio between 0 and 1. The closer that ratio is to 1, the more CPU BN'd you are at the lower of the two clocks (although this method can't differentiate between 'frequency' vs. 'severity', unless you don't move around during your test ... I can explain more what I mean by that if someone wants).
Edited by brettjv - 2/7/13 at 10:44am
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › Gpu usage not pegging in bf3 680 sli setup