On Sat, Jul 28, 2007 at 10:42:23PM -0400, john wrote:
Silly me. A friend pointed out that the monitor section in xorg.conf was set to use the VGA input of my monitor. I had my video card connected to the DVI-D input of the monitor. Once I connected the cable to the proper port, X came up perfectly. So now my LinuxBIOS is working.
Great news! :)
Still I have some observations:
- Nvidia proprietary driver does not work, however NV works well enough
for my purposes.
Do you notice any special problem with the nv driver? Or just that it doesn't offer the same functionality as the proprietary driver?
- I've lost cpu frequency control (no powernow-k8) so now both my
CPUs are running flat out at 3000 MHz. Core temperatures are hovering just below 70C (63C and 67C).
I just had a look at arch/i386/kernel/cpu/cpufreq/powernow-k8.c and it currently depends on ACPI and the BIOS keeping a table of parameters that it needs;
./powernow-k8.h:#define PSB_ID_STRING "AMDK7PNOW!" ./powernow-k8.h:#define PSB_ID_STRING_LEN 10
for (i = 0xc0000; i < 0xffff0; i += 0x10) { /* Scan BIOS looking for the signature. */ /* It can not be at ffff0 - it is too big. */
psb = phys_to_virt(i); if (memcmp(psb, PSB_ID_STRING, PSB_ID_STRING_LEN) != 0) continue;
dprintk("found PSB header at 0x%p\n", psb);
Once the table has been initialized in the driver ACPI is not used at all - the real work is done using rdmsr() and wrmsr().
I think it would be easy enough to change the initialization so that ACPI is not required.
AMD: Are these tables board dependent? Why not put the tables, arranged by cpuid, into the powernow driver and do without ACPI?
Another option is of course for someone to create the needed table, store it in the right place and reserve that memory so Linux doesn't use it - but I don't like that solution too much.
- I now can load kvm_amd (without locking up my system) and use
hardware acceleration to speed up my virtual machines.
Have you been able to make any performance measurements yet?
This is really fantastic news!
//Peter