Performance Decrease with PG278Q

[gridironcpj]

Hello folks,

I've noticed a performance decrease with G-sync enabled or disabled with my PG278Q. I did some extensive testing with this and I've come to the conclusion that having G-sync available as an option in the NVCP causes a performance hit (bad drivers I suppose?). This is consistent with the last 3 Nvidia drivers and is consistent over Windows 7 and Windows 8.1 (I upgraded to 8.1 recently). In order to pinpoint this issue on the PG278Q, I turned my PC off, unplugged my PG278Q from my PC, and only had my PB278Q plugged in. This automatically removes G-sync as an option in the NVCP. My performance is normal in this setting. My graphics score in Fire Strike drops 1,000-2,000 if the PG278Q is plugged in (even if it isn't even selected as an active monitor!). Once again, I attribute this to G-sync being an option in the NVCP (enabled or disabled).

Does anyone have a fix for this or is it truly a driver issue? Can anyone else replicate my experiment and see if they come to the same conclusion? To be specific, try this:

(1) Have the PG278Q plugged into your PC and run Fire Strike (G-sync disabled).
(2) Unplug the PG278Q and use a different monitor. Run Fire Strike with the exact same overclock and fan profiles.
(3) Compare the scores.

For reference, here is my relevant hardware:
GPU(s): GTX Titan 2-way SLI (both typically overclocked ~1.1GHz in gaming and benchmark applications)
CPU: i7 3930K @4.3GHz (cooled via Corsair H110)
RAM: Corsair Vengeance 4x4GB 1866MHz
Mobo: Rampage IV Formula
PSU: Corsair AX1200i

[fenderjaguar]

What scores you get when disabling sli?

[RealNC]

If you ONLY connect the PG278Q, does the issue persist? Just trying to make sure this isn't a dual monitor issue.

[Bishi]

I don't have 3Dmark installed to test at the moment I'm afraid, but I did some testing in Unigine Valley:
As you can see I didn't get the same problem you have but I did notice a large Gsync performance hit I wasn't expecting.

Valley in DX11, Ultra, 2560x1440, no AA. 340.52 NVIDIA drivers

Gsync enabled using PG278Q
FPS: 91.3 (min 39.2, max 139)
SCORE: 3822

Vsync enabled using PG278Q
FPS: 78.1 (min 38.4, max 104.3)
SCORE: 3267

Gsync / vsync disabled using PG278Q
FPS: 112.9 (min 29.7, max 190.8) (what the hell? I thought Gsync didn' t affect performance...)
SCORE: 4724

PG278Q unplugged and using other 2560x1440 IPS screen for benchmark
FPS: 113.2 (min 42.5, max 191.3)
SCORE: 4735

My rig is:
780 GTX SLI @ 1201/6908
PG278Q (and two other 60hz IPS monitors on DVI)

5960X @ 4.4
16GB Crucial DDR4 @ 2400


This may be a silly question but your monitors are all plugged into the same (primary) video card, right?

[gridironcpj]

What scores you get when disabling sli?

I have not tested this without SLI, so I can do that next.

[gridironcpj]

If you ONLY connect the PG278Q, does the issue persist? Just trying to make sure this isn't a dual monitor issue.

Yes, only the PG278Q connected. My tests were with one monitor connected to my PC at a time.

[gridironcpj]

I don't have 3Dmark installed to test at the moment I'm afraid, but I did some testing in Unigine Valley:
As you can see I didn't get the same problem you have but I did notice a large Gsync performance hit I wasn't expecting.

Valley in DX11, Ultra, 2560x1440, no AA. 340.52 NVIDIA drivers

Gsync enabled using PG278Q
FPS: 91.3 (min 39.2, max 139)
SCORE: 3822

Vsync enabled using PG278Q
FPS: 78.1 (min 38.4, max 104.3)
SCORE: 3267

Gsync / vsync disabled using PG278Q
FPS: 112.9 (min 29.7, max 190.8) (what the hell? I thought Gsync didn' t affect performance...)
SCORE: 4724

PG278Q unplugged and using other 2560x1440 IPS screen for benchmark
FPS: 113.2 (min 42.5, max 191.3)
SCORE: 4735

My rig is:
780 GTX SLI @ 1201/6908
PG278Q (and two other 60hz IPS monitors on DVI)

5960X @ 4.4
16GB Crucial DDR4 @ 2400


This may be a silly question but your monitors are all plugged into the same (primary) video card, right?

Thank you for taking the time to do those tests. All of my monitors are plugged into the same video card, but my tests were conducted with only one connected at a time (so I manually unplugged the other while testing).

Here are some numbers for you guys using the Valley benchmark. The settings are as follows:

API: DirectX 11
Quality: Ultra
Stereo 3D: Disabled
Monitors: Single
Anti-aliasing: x8
Full Screen: Check
Resolution: 2560x1440

Results:

2x GTX Titans on PB278Q 60Hz - 81.0FPS Avg.
2x GTX Titans on PG278Q 144Hz (G-sync disabled) - 78.1FPS Avg.
1x GTX Titan on PB278Q 60Hz - 42.3FPS Avg.
1x GTX Titan on PG278Q 144Hz (G-sync disabled) - 42.3FPS Avg.

From this little test, it appears that SLI is the culprit of the performance hit. It's not a deal-breaker, but it is annoying. The effect is more prominent in 3D Mark 2013 Fire Strike.

[gridironcpj]

So is anyone else having similar issues or does anyone have any idea what the problem could be? I really hope it's not a hardware problem with the monitor.

[masterotaku]

I think it's pretty easy to see when this "performance decrease" happens:

Gsync enabled using PG278Q
FPS: 91.3 (min 39.2, max 139)
SCORE: 3822

Gsync / vsync disabled using PG278Q
FPS: 112.9 (min 29.7, max 190.8) (what the hell? I thought Gsync didn' t affect performance...)
SCORE: 4724

G-Sync is capping your fps, so that's going to impact your average fps. Those minimum fps could be in the margin of error, or while loading something from the HDD, I don't know. Unless that's maintained during some time.

PG278Q unplugged and using other 2560x1440 IPS screen for benchmark
FPS: 113.2 (min 42.5, max 191.3)
SCORE: 4735

See? You get more or less the same when G-Sync isn't capping your fps. The monitor doesn't matter if vsync is off and resolution is the same.

[gridironcpj]

I think it's pretty easy to see when this "performance decrease" happens:

Gsync enabled using PG278Q
FPS: 91.3 (min 39.2, max 139)
SCORE: 3822

Gsync / vsync disabled using PG278Q
FPS: 112.9 (min 29.7, max 190.8) (what the hell? I thought Gsync didn' t affect performance...)
SCORE: 4724

G-Sync is capping your fps, so that's going to impact your average fps. Those minimum fps could be in the margin of error, or while loading something from the HDD, I don't know. Unless that's maintained during some time.

PG278Q unplugged and using other 2560x1440 IPS screen for benchmark
FPS: 113.2 (min 42.5, max 191.3)
SCORE: 4735

See? You get more or less the same when G-Sync isn't capping your fps. The monitor doesn't matter if vsync is off and resolution is the same.

That's not the case for me though. I get a performance decrease just by having the PG278Q plugged in (if I'm using SLI).