My graphics card has performed very well since building this system in late 2012. Until the windows anniversary update, a few weeks ago I believe it was - either coincidentally or not, I'm not sure. Also around that time, I installed 'Bluestacks' to run a few mobile-only apps on the PC. This program seemed to bring a tremendous amount of CPU overhead with it, and was therefore uninstalled - but that overhead may have been related to this graphics issue as well.
So, the system is a custom build. Excluding case and fans, DVD burner etc, we have:
Motherboard: ASUS M5A99X EVO 990X R (Note - this motherboard has no on-board video card)
Video Card: EVGA 2G GTX 560 TI (NVIDIA)
Power Supply: 675W ToughPower XT
CPU: AMD FX8350
RAM: 32G (4 x 8G) GSkill 1600mhz RAM
OS: Windows 10 (was initially, Windows 7 Home Edition, OEM)
Here's the pattern. I've been through 6 or 7 restarts today...
The graphics card core clock is normally supposed to be something that dynamically adjusts to the amount of work required of the card. In my experience, it isn't something you, as a user, ever have to be concerned with. Running a graphics-intensive application, a game, or whatnot, will make use of the video card and adjust the core clock speed accordingly to handle the workload. This in turn, spins up the card fans, and the desired video output is produced.
Not anymore. If, shortly after startup, I run an application that normally "maxes out" the card, the clock speed still goes up, and I can hold it 'up', for hours on end, by continuing to run such an application. Usually, in my testing, the clock speed stays high until it's been in the background too long, and loses thread priority. That can go hours, or minutes, depending, but once it loses it, the card never re-increases the clock, no matter what.
It's like it is not receiving the messages to do the work -- or rather, it receives them, but does not act on them. I've seen games (which I rarely played to begin with, but needed something to test against) run at lowest quality, to the rate of 1 or 2 fps, where it seemed to be over 60, in times previous.
I wondered if something in the Windows power config got changed, as that is how it's acting, but I don't see anything in there that would be affecting this. This is so strange since the clock speed will go from 850, to 405, then to 50 - and only a reboot...quite disruptive this can be, will restore the ability to have it run higher once more.
Following the big Windows 10 update from the middle of last month, I noticed ASUS AI Charger was automatically uninstalled, as were a few other apps. I don't use the PC to charge devices so it didn't appear, initially at any rate, to affect the performance or such characteristics as I've noted above.
I have looked through the BIOS, as well, and see only one point related to this video card, in there, which is the use order of the PCI slot. PEG, PCI is what it was set to. Switching this made no difference so it was restored to the setting which I found it at. No other changes in BIOS were made.
I have searched the Internet for this problem and as close to it as I could find, suggested I go through the NVIDIA control panel and ensure the dedicated graphics (rather than on-board) graphics card is selected for use in apps. This...does not apply here because there is no on-board graphics card on this system.
What happens now, when higher graphics use is requested, is 25 to 35 percent of the CPU becomes used by the requesting application, rather than the video card. I am at a loss for how to address and resolve this matter. Any help, suggestions, etc, will be greatly appreciated.
Thank you for reading my post.
~ shaelan
Update:
- Power Supply not faulty. All voltages measure within +/- 0.05v of target (i.e., nothing is out of line).
- Selecting Custom Installation on the NVidia driver does not provide any options. It is the same as Express Installation. I saw a post which said to do a clean install of the driver, and that the option was available through the custom install.
- Running Furmark during the pattern's high-clock times yeilds 23 to 26 FPS with all Furmark options enabled. GPU temperatures do not exceed 68*C. During low-clock times, 1 FPS. Temps between 31*C and 34*C.
- Checked event log for event 41. 2 hits in the last 7 days - both of which were due to improper system shutdown - under exceptional circumstances, as I recall.
Hope this additional information assists anyone who would care to lend a helping hand with this issue.
So, the system is a custom build. Excluding case and fans, DVD burner etc, we have:
Motherboard: ASUS M5A99X EVO 990X R (Note - this motherboard has no on-board video card)
Video Card: EVGA 2G GTX 560 TI (NVIDIA)
Power Supply: 675W ToughPower XT
CPU: AMD FX8350
RAM: 32G (4 x 8G) GSkill 1600mhz RAM
OS: Windows 10 (was initially, Windows 7 Home Edition, OEM)
Here's the pattern. I've been through 6 or 7 restarts today...
The graphics card core clock is normally supposed to be something that dynamically adjusts to the amount of work required of the card. In my experience, it isn't something you, as a user, ever have to be concerned with. Running a graphics-intensive application, a game, or whatnot, will make use of the video card and adjust the core clock speed accordingly to handle the workload. This in turn, spins up the card fans, and the desired video output is produced.
Not anymore. If, shortly after startup, I run an application that normally "maxes out" the card, the clock speed still goes up, and I can hold it 'up', for hours on end, by continuing to run such an application. Usually, in my testing, the clock speed stays high until it's been in the background too long, and loses thread priority. That can go hours, or minutes, depending, but once it loses it, the card never re-increases the clock, no matter what.
It's like it is not receiving the messages to do the work -- or rather, it receives them, but does not act on them. I've seen games (which I rarely played to begin with, but needed something to test against) run at lowest quality, to the rate of 1 or 2 fps, where it seemed to be over 60, in times previous.
I wondered if something in the Windows power config got changed, as that is how it's acting, but I don't see anything in there that would be affecting this. This is so strange since the clock speed will go from 850, to 405, then to 50 - and only a reboot...quite disruptive this can be, will restore the ability to have it run higher once more.
Following the big Windows 10 update from the middle of last month, I noticed ASUS AI Charger was automatically uninstalled, as were a few other apps. I don't use the PC to charge devices so it didn't appear, initially at any rate, to affect the performance or such characteristics as I've noted above.
I have looked through the BIOS, as well, and see only one point related to this video card, in there, which is the use order of the PCI slot. PEG, PCI is what it was set to. Switching this made no difference so it was restored to the setting which I found it at. No other changes in BIOS were made.
I have searched the Internet for this problem and as close to it as I could find, suggested I go through the NVIDIA control panel and ensure the dedicated graphics (rather than on-board) graphics card is selected for use in apps. This...does not apply here because there is no on-board graphics card on this system.
What happens now, when higher graphics use is requested, is 25 to 35 percent of the CPU becomes used by the requesting application, rather than the video card. I am at a loss for how to address and resolve this matter. Any help, suggestions, etc, will be greatly appreciated.
Thank you for reading my post.
~ shaelan
Update:
- Power Supply not faulty. All voltages measure within +/- 0.05v of target (i.e., nothing is out of line).
- Selecting Custom Installation on the NVidia driver does not provide any options. It is the same as Express Installation. I saw a post which said to do a clean install of the driver, and that the option was available through the custom install.
- Running Furmark during the pattern's high-clock times yeilds 23 to 26 FPS with all Furmark options enabled. GPU temperatures do not exceed 68*C. During low-clock times, 1 FPS. Temps between 31*C and 34*C.
- Checked event log for event 41. 2 hits in the last 7 days - both of which were due to improper system shutdown - under exceptional circumstances, as I recall.
Hope this additional information assists anyone who would care to lend a helping hand with this issue.
Facebook
Twitter
Instagram