First things first my gaming rig:

Asus Rampage V Extreme ROG 2011-3 socket
Gigabyte GTX 980 Ti (XTreme edition, 6GB of VRam, Coverclock Core clock: 1447 Mhz, Mem clock OC: 3304 MHZ)
Intel i7 5930k, 6 cores, 12 threads, overclocked to 4.3 GHz from 3.5
16 Gigs of DDR4 Ram (Quad Channel)
Samsung 850 pro evo SSD
Windows 10 x64

My problem is that recently my GPU started to reach a maximum of only 37% load (short runs) or 77% load (long runs) and not my beloved 99% which costs me some computing time ("some", the 980Ti is still one of the best GPUs out there). Moreover, GPU-Z tells me that my GPU only supports PCI-e 2.0 and uses PCI-e 2.0 with 8 lanes even though I "said" my mainboard (UEFI) to use PCI-e 3.0 (of course I use a PCIe 3.0 x16 slot and I don't use power save modes). I think or rather hope that it's a bug but if it's not I don't know how to fix that. Does the maximum load of my GPU depends on the PCIe interface?

I tried to get more load on my GPU by editing the app_config.xml file.
Now it looks like this:

<app_config>
<app>
<name>acemdlong</name>
<max_concurrent>9999</max_concurrent>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>

<app>
<name>acemdshort</name>
<max_concurrent>9999</max_concurrent>
<gpu_versions>
<gpu_usage>0.5</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>
</app_config>

and I get a load of 90 % constantly which is "acceptable". However, I'd like to use 98%-99% of my GPU and I do prefer to run only one task at a time (cuz of personal reasons) which I couldn't achieve with the following configurations:


1 cpu + 1 nvidia card = 77% use
12 cpus + 1 nvidia card = 83% use (<- so low even with 12 threads feeding the GPU)
(12 cpus + 2*0.5 nvidia card = 83% use)
(1 cpu + 2*0.5 nvidia card = 90% use)
(6 cpus + 2*0.5 nvidia card = 91% use)

I hope somebody can help me out of this :) I WANT MY 99% GPU-LOAD BACK #FirstWorldCruncherProblems

(And don't mind my bad English)

Note: I will not tolerate more than one core per GPU task because I want to run World Community Grid.

UPDATE: Now I managed that my GPU uses PCI-e 3.0 but only with 8 Lanes (which is ok becuase 8 or 16 lanes don't make a difference). And my GPU utilization is now 1 - 2% higher. So I reached 92% utilization.

From what I have read these new tasks require more communication between the CPU and GPU. This causes the GPU to run at less than 90%. Running 2 tasks per GPU does raise that percentage but does that really help? I tried that on my 970s and found the increased GPU load led to more error tasks and lower overall efficiency so I went back to 1 task per card. I don't think the GPU load rates will be getting any higher with the current tasks.