cancel
Showing results for 
Search instead for 
Did you mean: 

Very odd question regarding video card frame rates.

Perko
Level 8
OK bare with me a few seconds. I'll do my best to explain this.
First know that I have indeed verified this by swapping various parts between my Z97 and X99 boxes.

I have two 2560x1440 monitors. A Dell U2713HM and a Acer XB270HU. I have three 980Ti's ... 2x Classifieds and a G1.

Here my issue. IF I run any of the three cards with either of the monitors at native resolutions on the X99 (RVE) system doing Sniper Elite V2 benchmark my average frame rate is in the high 80's to high 90's depending on the core clock I set the card at.... any of them with either monitor with vsync off. When I run the same test with either monitor or any of the cards on the Z97 (Asus Deluxe 3.1) system I get from high 140's to high 150's depending on the core clock I set the card at.

Same drivers, same OS, same amount of system ram, SSD's, 1200P (Seasonic/EVGA) power supplies in both systems. No cooling issues in either as my X99 is WC with 2x360, 1x240 and the Z97 is ALL air in a ambient room of 0c right now. Even if I bring the RVE system into the cold it changes nothing.

A 60 frame difference is massive for the same test using the same major components.

Any thoughts here.

I'm using the latest bios I have all PCI at gen3 I'm not a rook with either system or board. My 5960x with run 4.9 under 1.4v and my 4770k will also. Even if I run my G1 at 1595 core in the RVE box my max frame rate is under 100fps while at 1500 core on my weakest card will knock down 145fps on the Z97 system.

As Sidious said to Skywalker - so weak .....

SS
Case Labs S8
ASUS X99 Rampage V Extreme
i7 5960X
EKWB XTX360 kit +1 XTX360 & XTX240
10x MassCool fans on a Lamptron FC5 6 Channel 30W LED Fan Controller V3
32GB Geil Potenza 3000Mhz
2x Samsung 840 EVO 1TB
WD 7,200 Hybrid 2TB
ASUS Blu-Ray RW
Gigabyte 980Ti G1 EKWB full block
Acer XB270HU monitor
EVGA 1200 P2
Razor 7.1 phones, Death Ader & Ultimate KB
Madcatz Panther XL SteveO modded
5,345 Views
12 REPLIES 12

Qwinn
Level 11
Try a different benchmark, say, Heaven 4.0 (or whatever the latest is). See if you get a similar effect.

Also, what nvidia driver are you using? I know there's an open issue with the latest drivers and high idle usage that only affects X99 apparently. Don't think it's supposed to affect under load, but who knows?

I'd try running Firestrike on both setups as well.

Also monitor power usage on the video cards during your tests using MSI Afterburner or something similar.

Is it possible one system is set up as Max Performance either in windows or the nvidia control panel and the other is balanced? Shouldnt make that much difference, just trying to eliminate variables.

if none of the above clears things up, I'd list out all bios settings for both machines that differ from the defaults, tho the only setting I can think of that might make that much difference would be if hyperthreading was disabled.

Qwinn wrote:
Try a different benchmark, say, Heaven 4.0 (or whatever the latest is). See if you get a similar effect.

Also, what nvidia driver are you using? I know there's an open issue with the latest drivers and high idle usage that only affects X99 apparently. Don't think it's supposed to affect under load, but who knows?

I'd try running Firestrike on both setups as well.

Also monitor power usage on the video cards during your tests using MSI Afterburner or something similar.

Is it possible one system is set up as Max Performance either in windows or the nvidia control panel and the other is balanced? Shouldnt make that much difference, just trying to eliminate variables.

if none of the above clears things up, I'd list out all bios settings for both machines that differ from the defaults, tho the only setting I can think of that might make that much difference would be if hyperthreading was disabled.



theres a gforce driver fix thats available in the nvidia forums, its just not published yet over gforce experienced/the download section. (its 361.60)

check GPU-z for any PerfCap. did you tried it with vsync enabled? (you dont have to disabled it anymore for gsync)
disconnect your 60hz screen, reboot and try again. there are known issues with 60hz displays and gsync displays connected at the same time.

Minsekt wrote:
theres a gforce driver fix thats available in the nvidia forums, its just not published yet over gforce experienced/the download section. (its 361.60)


Yep, I am running 361.60. I spend as much time on the NVidia forums as I do here, hehe.

The thread I'm talking about where NVidia acknowledged the X99 high idle load issue is here:

https://forums.geforce.com/default/topic/910580/geforce-drivers/increased-idle-power-usage-since-358...

The NVidia rep that acknowledged it posted it on that thread on 15 JAN 2016. Hotfix driver 361.60 was released 3 days prior to that, on 12 JAN 2016. So it's a safe bet it hasn't been addressed yet.

For the record, if you have a 144 Hz monitor, you can resolve most of the high idle load issue by switching to 120 Hz. My clocks run very high at idle on 144 Hz with the recent drivers, and mostly ok at 120 Hz.

Qwinn wrote:
Try a different benchmark, say, Heaven 4.0 (or whatever the latest is). See if you get a similar effect.

Also, what nvidia driver are you using? I know there's an open issue with the latest drivers and high idle usage that only affects X99 apparently. Don't think it's supposed to affect under load, but who knows?

I'd try running Firestrike on both setups as well.

Also monitor power usage on the video cards during your tests using MSI Afterburner or something similar.

Is it possible one system is set up as Max Performance either in windows or the nvidia control panel and the other is balanced? Shouldnt make that much difference, just trying to eliminate variables.

if none of the above clears things up, I'd list out all bios settings for both machines that differ from the defaults, tho the only setting I can think of that might make that much difference would be if hyperthreading was disabled.
Latest drivers on both. Same card bios on both. Adaptive or Max performance in the NVCP makes no change. Heaven, FireStrike, MetroLL all seem normal but the frame rates don't run as high either. On the Z97 rig I can see over 200fps (through PX16 OSD) where on the X99 box I see over 150fps at points during a run. I've double checked my NVCP settings on both systems and they are the same for each even when I swap monitors/cards/drivers. I've tried drivers back into the 34x.xx series with no changes.

Minsekt wrote:
theres a gforce driver fix thats available in the nvidia forums, its just not published yet over gforce experienced/the download section. (its 361.60)

check GPU-z for any PerfCap. did you tried it with vsync enabled? (you dont have to disabled it anymore for gsync)
disconnect your 60hz screen, reboot and try again. there are known issues with 60hz displays and gsync displays connected at the same time.
Depending on the card bios I run I can see performance caps in all tests on both machines (stock bios) or no performance caps (modded bios) on either machine. No card throttling on any as 2 cards are under blocks and the one on air never runs over 60c. Yes I've run the blocked cards with the air coolers, they don't throttle on air either. For the record I don't run duel monitors on either system.

Qwinn wrote:
Yep, I am running 361.60. I spend as much time on the NVidia forums as I do here, hehe.

The thread I'm talking about where NVidia acknowledged the X99 high idle load issue is here:

https://forums.geforce.com/default/topic/910580/geforce-drivers/increased-idle-power-usage-since-358...

The NVidia rep that acknowledged it posted it on that thread on 15 JAN 2016. Hotfix driver 361.60 was released 3 days prior to that, on 12 JAN 2016. So it's a safe bet it hasn't been addressed yet.

For the record, if you have a 144 Hz monitor, you can resolve most of the high idle load issue by switching to 120 Hz. My clocks run very high at idle on 144 Hz with the recent drivers, and mostly ok at 120 Hz.

The Dell monitor in not a G-sync and the default refresh is 60Hz but I have it OC'd to 85Hz which neither setting makes a difference. Again, even if I swap only monitors nothing changes in the results. In adaptive mode on the Acer monitor at idle the clocks run very low like the Dell. I can try the hot fix drivers tonight I guess.

Hum .... the only thing I haven't swapped between them is the DP cables. I guess I could try that since I have a few spares.

SS
Case Labs S8
ASUS X99 Rampage V Extreme
i7 5960X
EKWB XTX360 kit +1 XTX360 & XTX240
10x MassCool fans on a Lamptron FC5 6 Channel 30W LED Fan Controller V3
32GB Geil Potenza 3000Mhz
2x Samsung 840 EVO 1TB
WD 7,200 Hybrid 2TB
ASUS Blu-Ray RW
Gigabyte 980Ti G1 EKWB full block
Acer XB270HU monitor
EVGA 1200 P2
Razor 7.1 phones, Death Ader & Ultimate KB
Madcatz Panther XL SteveO modded

Qwinn wrote:
Yep, I am running 361.60. I spend as much time on the NVidia forums as I do here, hehe.

The thread I'm talking about where NVidia acknowledged the X99 high idle load issue is here:

https://forums.geforce.com/default/topic/910580/geforce-drivers/increased-idle-power-usage-since-358...

The NVidia rep that acknowledged it posted it on that thread on 15 JAN 2016. Hotfix driver 361.60 was released 3 days prior to that, on 12 JAN 2016. So it's a safe bet it hasn't been addressed yet.

For the record, if you have a 144 Hz monitor, you can resolve most of the high idle load issue by switching to 120 Hz. My clocks run very high at idle on 144 Hz with the recent drivers, and mostly ok at 120 Hz.


thats interesting, my card is sometimes idling at 135mhz, but sometimes it just wont downclock. (i was aware of the issue but i thought the last 361.60 fix solved it) apparently not 😛

Vlada011
Level 10
My card boost and downclock in idle normal.
X99 - Kepler G110.
361.43 WHQL Driver.

I don't want to hear someone to blame X99 for something.
X99 is PERFECT PLATFORM... I mean nothing is perfect but best in this moment.

Chino
Level 15
I would take in game benchmarks with a pinch of salt. As suggested by Qwinn, Unigine Heaven is a good benchmark for GPUs. Try running it on a single GPU on your X99 and Z97 systems. Your result should be very close to each other.

Chino wrote:
I would take in game benchmarks with a pinch of salt. As suggested by Qwinn, Unigine Heaven is a good benchmark for GPUs. Try running it on a single GPU on your X99 and Z97 systems. Your result should be very close to each other.


I don't place much on benches as HARD evidence but that kind of difference if too much to discount.

I'm only running one card in each system. I'll try Heaven later.

SS
Case Labs S8
ASUS X99 Rampage V Extreme
i7 5960X
EKWB XTX360 kit +1 XTX360 & XTX240
10x MassCool fans on a Lamptron FC5 6 Channel 30W LED Fan Controller V3
32GB Geil Potenza 3000Mhz
2x Samsung 840 EVO 1TB
WD 7,200 Hybrid 2TB
ASUS Blu-Ray RW
Gigabyte 980Ti G1 EKWB full block
Acer XB270HU monitor
EVGA 1200 P2
Razor 7.1 phones, Death Ader & Ultimate KB
Madcatz Panther XL SteveO modded

Qwinn
Level 11
"Adaptive or Max performance in the NVCP makes no change."

What about in Windows? Balanced power plan versus High Performance. I stated earlier I didn't think it could make that much difference, but I just discovered it makes a 30,000 point difference (from 86k to 116k) when running the Image Editing portion of Realbench benchmark. That's a whole lot.