GPU monitoring with multiple software.


Well-Known Member
Is there going to be conflicts/potential issues when multiple programs are monitoring the GPU particularly during gaming or stress testing? Does this concern HWiNFO? With my EVGA graphics card I don't have any choice but to run their Precision X overclocking software to control the fans, the EVGA GTX 1080Ti FTW3 does not have a standard fan controller. This means I cannot control fans exclusively with MSI Afterburner but I prefer to use Afterburner for it's OSD. This means three programs are monitoring not just the GPU but almost all other hardware also...
Yes, there is a potential possibility of a conflict, especially when multiple applications would access the ICX devices.
I'm not familiar how the other tools work internally and don't have much experience with proprietary ICX. But I assume that all the other tools are using standard NVIDIA API, hence I think that the changes of a conflict are rather small. So sorry that I can't give you an exact answer, you will have to try and see how it works...
Sorry, the potential consequences are unpredictable. It could be invalid readouts, or fan control issues.
No worries. I will make some inquiry's at EVGA's forum to see whether someone might have something more to add. :)
I got a nice little reply from Unwinder of MSI Afterburner, I think he may have summed it up quite well:

"In general case - yes, multiple applications accessing hardware directly may conflict due to unsynchronized low-level access to hardware. But in modern software world developers of hardware monitoring solutions try to add common hardware access synchronization mechanisms in order to avoid such cases. And in case of NVIDIA GPUs, the days of low-level access to NVIDIA GPUs are in the past, today everyone rely on official NVAPI, which has its own internal synchronization mechanisms. So I'd say that the chances of conflict are close to zero if we're talking about multiple modern tools monitoring NVIDIA GPU."
That's right and what I said earlier, I just didn't want to go into much details :)
But we never know how reliable NVAPI is for concurrent access (even though it seems to be very good) and what tools like Precision X do, if that also relies completely on (synchronized) NVAPI or perhaps is using something low-level too.