Nicehash Multi-GPU, low hashrates on "GPU #0"?

[zcypher]

 Hi yall,

Read a bit in other forums and threads, but so far haven't found anything too conclusive.  My setup for now is:

- Asus Z170-K + G3930 + Ballistix 8GB DDR4 2400, latest BIOS installed
- 4x Asus GTX 1060 6GB 9G (dual fan)
- Win10 anniversary edition (going from non-anniversary to anniversary increased eth hashrate from 2.5mhs to 17mhs)
- Latest nvidia drivers, which were fully uninstalled and reinstalled a few times
- Latest Asus chipset / other drivers
- Not leaving MSI Afterburner or any other programs running, no OC or downclocking or undervolting at this time, stock settings
- Have run with monitor connected and not connected

It was supposed to be a Z170-A, so I have two extra 1060 soon to be put to use, but anyway, not the point of this thread.  As I've tinkered with this, I've noticed that GPU #0 often has significantly lower hashrates than the other three GPUs.  This happens in both ccminer and excavator using Nicehash.  When the hashrates are lower, I see the other values are lower as well.  I can't see anything obvious to attribute it to.  Not sure if it's just my imagination or not, but setting the "priority" in task manager for the currently running miner seems to help a bit.  That's not something I want to look towards as a solution since Nicehash is always switching back and forth, effectively resetting any priority changes.

If others have noticed the same effect in multi-GPU rigs running Nicehash, what have your experiences been, have you had any luck?

[crazydane]

I'm currently running 3 x 1080Ti with NiceHash.  I don't have a monitor directly connected, but when I open a 4K RDP session to that miner, GPU #0 production drops about 30%.

If I open a RDP session at 1280x1024, there is only a 5% or so drop in production on GPU #0.  With the RDP window minimized, GPU #0 is producing at the same speed as the other 2 GPUs.

This only started after I switched over to using 1x PCIe risers (got 3 more 1080Ti's coming this week).

I run a MSI Z270-A PRO motherboard and my GPUs are EVGA 1080Ti SC Gaming Black Edition (11G-P4-6393-KR) and I'm using V007 risers, Windows 10 latest creators and latest Nvidia drivers (384.94).

[the_blesss]

I have 2 rigs and both exhibit the GPU0 decreased performance when connected to RDP especially, but less so for Teamviewer. I thought it was MSI Afterburner at first, since I switched over to some newer GTX 1070 TI Mini's, from RX 480s, so I uninstalled the app, cleaned up, rebooted and saw the same behavior when using RDP. Using Teamviewer there is only a 10% drop in performance, as opposed to RDP that dropped 40% when running full screen RDP at 1920 x 1080. My guess is that the performance will move back up when the session is closed.

Looking at the logging data of nicehash shows there is better performance, which corresponds to the remote sessions being closed.

Both rigs running Windows 10, 1703

GPUs were MSI GAMING RX 480 4GB
GPUs now are ZOTAC GTX 1070 TI Mini 8GB

[frazier34567a]

I have a non rig that I use gotomypc to access and then use TightVNC to connect to them on the same network. I do not have monitors connected to any of the GPUs on my rigs and I do not see any decrease in hashrates using TightVNC.