On Saturday 07 August 2004 06:30 pm, Lee Revell wrote: > On Sat, 2004-08-07 at 06:24, Florian Schmidt wrote: > > do you know how, in the case of volountary preemption and threaded irq > > handlers, the irq's used by X11 are handled? i removed the nvidia module > > from my kernelvand rebooted to have it untainted, but of course the X11 > > nv module does neither report to /proc/interrupts nor /proc/irq. SInce > > it is on another bus [agp] i suppose this is no problem but i always > > wondered about the interplay of the kernel an X. It looks to me like an > > area which might still have some surprises.. > > I believe that video cards use interrupts for 3d engine completion - you > tell the card to render such and such, rotate 30 degrees, zoom, and you > get an interrupt when it's done and ready for the next command. > > There are certainly some surprises in this area. If 2D acceleration is > enabled on my onboard Via CLE266 video, the 2D engine will cause > interrupts from the PCI slot to be disabled for long periods. I am not > sure yet if this a bug or a feature. > Tweaking the PCI latency timer values makes a big difference WRT the video card hogging the buss. I know thats not what you're talking about per se, but it's apropos WRT xruns/performance. > I recommend disabling all hardware video acceleration when tuning a > system for low latency (set Option "NoAccel" and commend out the "dri" > line). This will prevent X from interacting directly with your hardware > - it is complicated enough when only the kernel can access hardware. > Then, once you get that working, re-enable 2D acceleration, then DRI/3D > acceleration, if these do not cause problems. This will make it much > easier to get the problematic video drivers fixed. > > Lee Definitely a good strategy, but nobody is going to take a system with no video acceleration seriously.
