Does anybody know if the ASUS GTX 460 cards downclock when a monitor is not connected? Performing various benchmarks (FFT/IFFT and various homegrown algorithms) shows a marked increase in performance when a monitor is connected to my GTX 460. Performance for a particular test is around 649ms when connected, 730ms when disconnected. The deviceQuery indicates the core clock is at 1.35GHz regardless. I am unable to get the memory clock speed through nvclock however as it reports a large negative number which is clearly incorrect.
If so, is there any way to force stock clock speed at all times?
We did just get another GTX 460 (this one from Palit) in order to see if there is a difference from board partner to board partner. For the Palit card we are getting around 642ms when the monitor is disconnected, and 653ms when it is connected. In this case we are seeing the exact opposite, however the difference is minimal.
The 642ms number for the Palit, and the 649ms number for the ASUS make sense because the Palit has a slight factory overclock to it. What doesn’t make sense is the dependence on a monitor.
We are not running X.
Drivers: 256.40
Well, if a card is on idle it automatically Clocks down. Atleast my GTX 465 does so. But if it is on load it should boost up with all clocks. and pluging out a monitor making the clock decreasing sounds weird to me. Anyways with MSI Afterburner you can set some clocks. Download on msi.com or use google :D
Well, if a card is on idle it automatically Clocks down. Atleast my GTX 465 does so. But if it is on load it should boost up with all clocks. and pluging out a monitor making the clock decreasing sounds weird to me. Anyways with MSI Afterburner you can set some clocks. Download on msi.com or use google :D
We purchased a PNY GTX 460 (VCGGTX4601XPB-OC) and this problem does not occur with this card.
As a result of this problem, and another IRQ related problem I’ve discussed in another post, I cannot recommend using either the Palit or the ASUS card for CUDA work with current drivers.
We purchased a PNY GTX 460 (VCGGTX4601XPB-OC) and this problem does not occur with this card.
As a result of this problem, and another IRQ related problem I’ve discussed in another post, I cannot recommend using either the Palit or the ASUS card for CUDA work with current drivers.
Non-display devices cannot be over or underclocked in Linux. Windows drivers will let you, and MSI Afterburner (which I use too) is only for Windows.
You can over/underclock display cards in Linux (using the Coolbits option in xorg.conf) but that’s usually not very useful because of the kernel time limit watchdog for display cards.
Non-display devices cannot be over or underclocked in Linux. Windows drivers will let you, and MSI Afterburner (which I use too) is only for Windows.
You can over/underclock display cards in Linux (using the Coolbits option in xorg.conf) but that’s usually not very useful because of the kernel time limit watchdog for display cards.