Overclock.net › Forums › Graphics Cards › NVIDIA › Erratic GPU Usage in Metro 2033
New Posts  All Forums:Forum Nav:

Erratic GPU Usage in Metro 2033 - Page 2

post #11 of 20
Thread Starter 
Okay:
Unplugged my secondary monitor, only have the primary plugged in now
removed the line "r_vsync on" from user.cfg files (all of them, metro has multiple)
restored nvidia control panel settings to default (triple buffering and vsync application controlled, with no special profile for metro)
metro profile removed from d3doverride, force vysnc and triple buffering options unchecked, shut the program down.

Metro 2033 benchmark run "Frontline":
settings - dx11, 1920x1200, very high, AAA, AF 16x filtering, 3 runs, Tesselation on, Advanced DOF off (I never played with it on anyways)
Results

I didn't get a screen of GPU usage in afterburner, but I was watching it through the benchmark. When I had smooth frames (50+) usage was in the high 80s, peaked at exactly 90%. When fps dropped off into the low-mid 30s, usage was usually between 45% and 65%. Never got super jerky, below 20fps like I've had happen during actual gameplay, but without any vsync or triple buffering options on, I saw a lot of horizontal tearing

@brettjv: I have ADoF off, and like I said, frames are only dropping when GPU usage drops, so my system has the capacity, it's just not using it. I didn't look at VRAM during the benchmark, but I've looked during gameplay and it's usually slightly above or below 1gb, never approaching 1.2 or even 1.1. Stupid question, what's the difference between AAA and 4xaa, and which is more demanding? I don't see a visual difference so I'd be happy going with the lighter weight option...other than that, I play at your exact same settings and same resolution, and same gpus according to your sig, but a higher clock, and I get choppy all the time, so something's gotta be wrong with what I'm doing.

I'll test based on the thread you linked

Edit: in the benchmark thread, Jam3s got this:
SLI GTX 470's (800/837/1600) ---- 59.72 FPS
at 1920x1080. I ran the benchmark with the exact settings specified, and my clock is slightly higher, and I got between 45 and 48 fps for the three runs. I wouldn't expect that kind of difference...

Edited by tommy1993 - 5/10/11 at 6:45pm
Gaming Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k @ 5.2ghz 1.43v Asus P8P67 Deluxe Asus + MSI GTX470 SLI - 885/1770/1800@1.1v 3x2gb Corsair Dominator DDR3 1600 
Hard DriveOptical DriveOSMonitor
1TB + 500GB Seagates, 80gb Intel X25-M SSD LG 22x CD/DVD Reader/Writer Win7 Home Premium x64 24" 1920x1200 Samsung 
KeyboardPowerCaseMouse
Logitech G15 XFX 850w Black Edition Coolermaster HAF932 Logitech G500 
Mouse Pad
Razer Goliathus 
  hide details  
Reply
Gaming Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k @ 5.2ghz 1.43v Asus P8P67 Deluxe Asus + MSI GTX470 SLI - 885/1770/1800@1.1v 3x2gb Corsair Dominator DDR3 1600 
Hard DriveOptical DriveOSMonitor
1TB + 500GB Seagates, 80gb Intel X25-M SSD LG 22x CD/DVD Reader/Writer Win7 Home Premium x64 24" 1920x1200 Samsung 
KeyboardPowerCaseMouse
Logitech G15 XFX 850w Black Edition Coolermaster HAF932 Logitech G500 
Mouse Pad
Razer Goliathus 
  hide details  
Reply
post #12 of 20
AAA is lighter-weight

Something to understand about GPU usage is that it's 'reactive', not 'proactive'.

IOW, your gpu's will run at 99% UNLESS there is something stopping them from doing so. So basically, your lower FPS isn't being CAUSED by the lower GPU usage ... rather, it's the other way around.

So, there's SOME kind of bottleneck that's keeping the cards from running at full capacity. Sometimes it's just a driver problem, pure and simple. Or it's a framerate cap. Or it's v-sync. Or it's a CPU BN. Or it's physX, because physX usage is invisible unless you have it running on a dedicated card.

OR ... and here's somethign that doesn't often get mentioned ... it's a bottleneck within a subsystem of the card itself. Your card is not a monolithic entity, it's 'usage' is being calculated based on how much load there is on a specific subsystem (I *believe* it's at the ROP domain, but I'm not certain).

However, your card has many OTHER subsystems (besides the one where GPU Usage is calculated) that COULD act as a bottlenecks.

Like I mentioned above ... if you run out of vram, your FPS will plummet as you start to use system ram. This will act as a bottleneck, and GPU usage will go down. Your memory bandwidth being saturated could do the same thing. So could running out of shader power, or your texture fill rate being too low for the particular scene.

So ... sometimes lower-than-max GPU usage is just reflective of a BN within the card itself

EDIT: I just spoke w/a graphics programmer (a total code ninja), and she says it's most likely being caused by texture swapping of some kind ... either within the card itself (moving them around in memory, or pulling them into the GPU caches), or due to textures swapping between vram and system main memory.

Apparently the latter DOES happen quite a bit, even if your vram usage on the card is NOT being exceeded. Something I didn't know until just now.
Edited by brettjv - 5/10/11 at 7:23pm
    
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 20
Also check the slot speeds, you've got 3 pci-e lanes filled up with a non nf200 board.
Crazy things are bound to happen.
Mighty-iTX
(12 items)
 
One foot in..
(17 items)
 
mATX
(12 items)
 
CPUMotherboardGraphicsRAM
i7-4770K M7-Impact Strix GTX980 Samsung 
Hard DriveCoolingOSMonitor
Crucial M4 256 Glacer 240L W8.1 K272HUL 
KeyboardPowerCaseMouse
Saitek Seasonic X-650 Prodigy MX518 
CPUMotherboardGraphicsGraphics
4770K/45/43 Maximus VI Extreme R290X R290X 
RAMHard DriveHard DriveOptical Drive
Samsung Crucial M4 64GB RAIDR LSSWM 
CoolingOSMonitorKeyboard
Cooler Master Glacer 7 Ultimate 64 SyncMaster Eclipse 
PowerCaseMouseMouse Pad
XFX 1050 BE 600T Silver MX518 Desk 
CPUMotherboardGraphicsRAM
i5-4330 Z87 Gryphon GTX690 G.Skill RipJawsX 
Hard DriveCoolingOSMonitor
WD Blacks Hyper 212 W8.1 ACER 
KeyboardPowerCaseMouse
Logitech Ultra X3 1000w Corsair 230T Orange Razer 
  hide details  
Reply
Mighty-iTX
(12 items)
 
One foot in..
(17 items)
 
mATX
(12 items)
 
CPUMotherboardGraphicsRAM
i7-4770K M7-Impact Strix GTX980 Samsung 
Hard DriveCoolingOSMonitor
Crucial M4 256 Glacer 240L W8.1 K272HUL 
KeyboardPowerCaseMouse
Saitek Seasonic X-650 Prodigy MX518 
CPUMotherboardGraphicsGraphics
4770K/45/43 Maximus VI Extreme R290X R290X 
RAMHard DriveHard DriveOptical Drive
Samsung Crucial M4 64GB RAIDR LSSWM 
CoolingOSMonitorKeyboard
Cooler Master Glacer 7 Ultimate 64 SyncMaster Eclipse 
PowerCaseMouseMouse Pad
XFX 1050 BE 600T Silver MX518 Desk 
CPUMotherboardGraphicsRAM
i5-4330 Z87 Gryphon GTX690 G.Skill RipJawsX 
Hard DriveCoolingOSMonitor
WD Blacks Hyper 212 W8.1 ACER 
KeyboardPowerCaseMouse
Logitech Ultra X3 1000w Corsair 230T Orange Razer 
  hide details  
Reply
post #14 of 20
Thread Starter 
Quote:
Originally Posted by grunion View Post
Also check the slot speeds, you've got 3 pci-e lanes filled up with a non nf200 board.
Crazy things are bound to happen.
hmm, forgot about that...but sli benchmarks I looked at with this same board or a lower end one with a 2500k/2600k tended to show essentially identical performance with first gen i7 rigs that didn't have pci e lane issues. do you think taking my sound card out is worth a shot? I'm hesitant to do it because I've had creative drivers working happily for a while, and whenever I mess with the card or the drivers at all it takes me like a full day to get it working properly again. I hate creative drivers
Gaming Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k @ 5.2ghz 1.43v Asus P8P67 Deluxe Asus + MSI GTX470 SLI - 885/1770/1800@1.1v 3x2gb Corsair Dominator DDR3 1600 
Hard DriveOptical DriveOSMonitor
1TB + 500GB Seagates, 80gb Intel X25-M SSD LG 22x CD/DVD Reader/Writer Win7 Home Premium x64 24" 1920x1200 Samsung 
KeyboardPowerCaseMouse
Logitech G15 XFX 850w Black Edition Coolermaster HAF932 Logitech G500 
Mouse Pad
Razer Goliathus 
  hide details  
Reply
Gaming Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k @ 5.2ghz 1.43v Asus P8P67 Deluxe Asus + MSI GTX470 SLI - 885/1770/1800@1.1v 3x2gb Corsair Dominator DDR3 1600 
Hard DriveOptical DriveOSMonitor
1TB + 500GB Seagates, 80gb Intel X25-M SSD LG 22x CD/DVD Reader/Writer Win7 Home Premium x64 24" 1920x1200 Samsung 
KeyboardPowerCaseMouse
Logitech G15 XFX 850w Black Edition Coolermaster HAF932 Logitech G500 
Mouse Pad
Razer Goliathus 
  hide details  
Reply
post #15 of 20
Don't take it out just yet, check the mainboard tab of cpu-z, also check gpu-z.
Mighty-iTX
(12 items)
 
One foot in..
(17 items)
 
mATX
(12 items)
 
CPUMotherboardGraphicsRAM
i7-4770K M7-Impact Strix GTX980 Samsung 
Hard DriveCoolingOSMonitor
Crucial M4 256 Glacer 240L W8.1 K272HUL 
KeyboardPowerCaseMouse
Saitek Seasonic X-650 Prodigy MX518 
CPUMotherboardGraphicsGraphics
4770K/45/43 Maximus VI Extreme R290X R290X 
RAMHard DriveHard DriveOptical Drive
Samsung Crucial M4 64GB RAIDR LSSWM 
CoolingOSMonitorKeyboard
Cooler Master Glacer 7 Ultimate 64 SyncMaster Eclipse 
PowerCaseMouseMouse Pad
XFX 1050 BE 600T Silver MX518 Desk 
CPUMotherboardGraphicsRAM
i5-4330 Z87 Gryphon GTX690 G.Skill RipJawsX 
Hard DriveCoolingOSMonitor
WD Blacks Hyper 212 W8.1 ACER 
KeyboardPowerCaseMouse
Logitech Ultra X3 1000w Corsair 230T Orange Razer 
  hide details  
Reply
Mighty-iTX
(12 items)
 
One foot in..
(17 items)
 
mATX
(12 items)
 
CPUMotherboardGraphicsRAM
i7-4770K M7-Impact Strix GTX980 Samsung 
Hard DriveCoolingOSMonitor
Crucial M4 256 Glacer 240L W8.1 K272HUL 
KeyboardPowerCaseMouse
Saitek Seasonic X-650 Prodigy MX518 
CPUMotherboardGraphicsGraphics
4770K/45/43 Maximus VI Extreme R290X R290X 
RAMHard DriveHard DriveOptical Drive
Samsung Crucial M4 64GB RAIDR LSSWM 
CoolingOSMonitorKeyboard
Cooler Master Glacer 7 Ultimate 64 SyncMaster Eclipse 
PowerCaseMouseMouse Pad
XFX 1050 BE 600T Silver MX518 Desk 
CPUMotherboardGraphicsRAM
i5-4330 Z87 Gryphon GTX690 G.Skill RipJawsX 
Hard DriveCoolingOSMonitor
WD Blacks Hyper 212 W8.1 ACER 
KeyboardPowerCaseMouse
Logitech Ultra X3 1000w Corsair 230T Orange Razer 
  hide details  
Reply
post #16 of 20
Thread Starter 
According to both CPU-Z and GPU-Z they're both at 8x, which I think is the max this board can do for SLI, and hypothetically shouldn't be bottle necking me by any more than 1-2%
also, I tried setting Force Alternate Frame Rendering 1 and 2 under nvidia settings, didn't do anything

btw, big thanks to everyone, especially brettjv and grunion for helping me out with this. I'd +rep you guys but there's no button

edit: running at high settings instead of very high nets me an average fps in the high 50s, so I think I'm just gonna cut my losses and enjoy the game, even if I'm a little peeved. I blame 4A games and thq. thanks again guys!
Edited by tommy1993 - 5/10/11 at 8:32pm
Gaming Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k @ 5.2ghz 1.43v Asus P8P67 Deluxe Asus + MSI GTX470 SLI - 885/1770/1800@1.1v 3x2gb Corsair Dominator DDR3 1600 
Hard DriveOptical DriveOSMonitor
1TB + 500GB Seagates, 80gb Intel X25-M SSD LG 22x CD/DVD Reader/Writer Win7 Home Premium x64 24" 1920x1200 Samsung 
KeyboardPowerCaseMouse
Logitech G15 XFX 850w Black Edition Coolermaster HAF932 Logitech G500 
Mouse Pad
Razer Goliathus 
  hide details  
Reply
Gaming Rig
(13 items)
 
  
CPUMotherboardGraphicsRAM
i5 2500k @ 5.2ghz 1.43v Asus P8P67 Deluxe Asus + MSI GTX470 SLI - 885/1770/1800@1.1v 3x2gb Corsair Dominator DDR3 1600 
Hard DriveOptical DriveOSMonitor
1TB + 500GB Seagates, 80gb Intel X25-M SSD LG 22x CD/DVD Reader/Writer Win7 Home Premium x64 24" 1920x1200 Samsung 
KeyboardPowerCaseMouse
Logitech G15 XFX 850w Black Edition Coolermaster HAF932 Logitech G500 
Mouse Pad
Razer Goliathus 
  hide details  
Reply
post #17 of 20
brettjv was saying something that made me think that the HDD may also be playing a part. If the game engine is not very smart at pulling data before it's needed, it might get choppy if, for example, you have a fragmented HDD. Have you looked into that ? If you have the game on Steam look into the "verify integrity of game cache" and "defragment cache files".

And since you've got an SSD, I'd put the game in there just for the sake of trying every possible thing.


P.S.: brettjv, congrats on being nominated a mod!
Edited by tpi2007 - 5/10/11 at 9:04pm
 
Metro 2033 review
Metro 2033
CPUMotherboardGraphicsRAM
Core i7-3820 Asus Sabertooth X79 MSI GTX 750 Ti TF Gaming 16 GB Corsair DDR3 1866 Mhz Dominator 
Hard DriveOptical DriveCoolingOS
Samsung SSD 830 128GB + WD Caviar Black 1TB Sony Optiarc DVD-RW Corsair A70 + Noiseblocker M12-P Windows 7 Home Premium 64-bit 
MonitorKeyboardPowerCase
BenQ RL2455HM Cooler Master Octane Corsair AX750 Professional Modular 80 Plus Gold Cooler Master HAF 912 Plus 
Mouse
Cooler Master Octane 
  hide details  
Reply
 
Metro 2033 review
Metro 2033
CPUMotherboardGraphicsRAM
Core i7-3820 Asus Sabertooth X79 MSI GTX 750 Ti TF Gaming 16 GB Corsair DDR3 1866 Mhz Dominator 
Hard DriveOptical DriveCoolingOS
Samsung SSD 830 128GB + WD Caviar Black 1TB Sony Optiarc DVD-RW Corsair A70 + Noiseblocker M12-P Windows 7 Home Premium 64-bit 
MonitorKeyboardPowerCase
BenQ RL2455HM Cooler Master Octane Corsair AX750 Professional Modular 80 Plus Gold Cooler Master HAF 912 Plus 
Mouse
Cooler Master Octane 
  hide details  
Reply
post #18 of 20
Quote:
Originally Posted by tpi2007 View Post
brettjv was saying something that made me think that the HDD may also be playing a part. If the game engine is not very smart at pulling data before it's needed, it might get choppy if, for example, you have a fragmented HDD. Have you looked into that ? If you have the game on Steam look into the "verify integrity of game cache" and "defragment cache files".

And since you've got an SSD, I'd put the game in there just for the sake of trying every possible thing.


P.S.: brettjv, congrats on being nominated a mod!
Thanks tpi

Upthread he said he tried those very suggestions, actually.

So Tommy ... how did you you do on the benchmark at the settings specified in the Metro benching thread? I would think it was comparable to my setup, w/perhaps a bit better result due to your CPU ...

BTW, forgot to mention ... 90% is like the exact max GPU usage I get in the bench as well, and I get drops into the 65-70% range as well.

Bottom-line, running down GPU usage issues can be REALLY frustrating, because there's actually soooo many things that can cause it, so many different types of bottleneck that can come into play.

In my experience, it's really best to just try to compare benchies w/others w/similar systems, and as long as you're doing the same as they are ... let it go. You'll save yourself a lot of headache by just accepting that you won't always get 99%.

Aside from OC'ing your CPU (which can certainly be viable solution in some cases) or trying other driver versions, there's typically not a whole lot else you can do about it.
    
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 #19 of 20
Quote:
Originally Posted by tommy1993 View Post
Edit: review sites like this one show stock 470 sli on a 4ghz i7 920 getting a minimum of 35 frames. I'm overclocked a good 25% and I constantly dip below 20 fps
That site is crap...
Here they show a different story:

http://www.overclock.net/graphics-ca...am-really.html

So their FPS is all over the place and not even close to reality.
post #20 of 20
Just remembered something:

I did quite a few benchmarks with Metro 2033 back when I had the E8400, to compare what it would be like after switching to the Q9550, and there was one curious thing that affected minimum fps: the resolution. No, not the obvious thing you guys are thinking, but something to do with certain games not liking certain resolutions.

Since with the E8400 I wasn't going very far, I benched at 1280x768 (my former desktop resolution - now it's 1280x800 - and one of the resolutions my picky monitor works well with, including the desktop), and afterwards, after getting the GTX460 (the GTX480 behaves the same), I was able to get 1280x800 working. And while the max framerates went down a bit (but way more than enough still), the min fps were more consistent from the 1st to the 3rd run. I've also heard some games perform worse at 1680x1050 than custom resolutions like 1680x1048 (or something along the lines).

Don't know if this helps, but I figured it wouldn't hurt sharing.

Oh, by the way, if you want to see the benchmarks, they're on this and this post.
Edited by tpi2007 - 5/11/11 at 12:34am
 
Metro 2033 review
Metro 2033
CPUMotherboardGraphicsRAM
Core i7-3820 Asus Sabertooth X79 MSI GTX 750 Ti TF Gaming 16 GB Corsair DDR3 1866 Mhz Dominator 
Hard DriveOptical DriveCoolingOS
Samsung SSD 830 128GB + WD Caviar Black 1TB Sony Optiarc DVD-RW Corsair A70 + Noiseblocker M12-P Windows 7 Home Premium 64-bit 
MonitorKeyboardPowerCase
BenQ RL2455HM Cooler Master Octane Corsair AX750 Professional Modular 80 Plus Gold Cooler Master HAF 912 Plus 
Mouse
Cooler Master Octane 
  hide details  
Reply
 
Metro 2033 review
Metro 2033
CPUMotherboardGraphicsRAM
Core i7-3820 Asus Sabertooth X79 MSI GTX 750 Ti TF Gaming 16 GB Corsair DDR3 1866 Mhz Dominator 
Hard DriveOptical DriveCoolingOS
Samsung SSD 830 128GB + WD Caviar Black 1TB Sony Optiarc DVD-RW Corsair A70 + Noiseblocker M12-P Windows 7 Home Premium 64-bit 
MonitorKeyboardPowerCase
BenQ RL2455HM Cooler Master Octane Corsair AX750 Professional Modular 80 Plus Gold Cooler Master HAF 912 Plus 
Mouse
Cooler Master Octane 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › Erratic GPU Usage in Metro 2033