[9fans] nvidia vs. dual processors

[9fans] nvidia vs. dual processors

[erik quanstrom]

i've been having problems with aux/vga when running
on a dual processor machine.  before the recient update
to aux/vga, the machine would get stuck in /sys/src/9/pc/vganvidia.c/^waitforidle.
the iprints would be painfully slow and the machine would became
unusable afterwards.  even "aux/vga -p" would exhibit this.
various three-fingered salutes would impress nobody.

after the recient update, "aux/vga -p" works but "monitor=vesa aux/vga -p"
puts the machine into a fairly unusable state as soon as the output hits
the bottom of the window.  the cursor makes big jumps, lags many seconds
and doesn't quite track rio's idea of the cursor location.  i coudn't
manage to select a window to to do fshalt, so i rebooted. (via the keboard.)

(this doesn't happen if i boot *nomp=1.)

i think vganvidia.c/^nvidiahwscroll is conflicting with the vesa requests in aux/vga.

in an unrelated problem. *nomp or not, the screen is shifted to the left 1".
(it works fine at the same video mode on linux.)  i'm not sure what the problem
might be.  any ideas?

- erik

Re: [9fans] nvidia vs. dual processors

[Philippe Anel]

erik quanstrom wrote:

>i've been having problems with aux/vga when running
>on a dual processor machine.  before the recient update
>to aux/vga, the machine would get stuck in /sys/src/9/pc/vganvidia.c/^waitforidle.
>the iprints would be painfully slow and the machine would became
>unusable afterwards.  even "aux/vga -p" would exhibit this.
>various three-fingered salutes would impress nobody.
>
>after the recient update, "aux/vga -p" works but "monitor=vesa aux/vga -p"
>puts the machine into a fairly unusable state as soon as the output hits
>the bottom of the window.  the cursor makes big jumps, lags many seconds
>and doesn't quite track rio's idea of the cursor location.  i coudn't
>manage to select a window to to do fshalt, so i rebooted. (via the keboard.)
>
>
realmode (and vesa) is not compatible with mp.
what happens with another monitor= value ?

>(this doesn't happen if i boot *nomp=1.)
>
>i think vganvidia.c/^nvidiahwscroll is conflicting with the vesa requests in aux/vga.
>
>in an unrelated problem. *nomp or not, the screen is shifted to the left 1".
>(it works fine at the same video mode on linux.)  i'm not sure what the problem
>might be.  any ideas?
>
>
what happens with the *good* monitor= value ?

>- erik
>
>
>
>

Re: [9fans] nvidia vs. dual processors

[andrey mirtchovski]

On 4/28/06, erik quanstrom <quanstro@quanstro.net> wrote:
> using monitor=multisync65, multisync75 i get the same results.

run xvidtune under X and get the correct values for your monitor, and
if they're different than the ones you're currently using add them to
vgadb

Re: [9fans] nvidia vs. dual processors

[erik quanstrom]

using monitor=multisync65, multisync75 i get the same results.

On Fri Apr 28 11:59:08 CDT 2006, xigh@free.fr wrote:
> >
> >i think vganvidia.c/^nvidiahwscroll is conflicting with the vesa requests in aux/vga.
> >
> >in an unrelated problem. *nomp or not, the screen is shifted to the left 1".
> >(it works fine at the same video mode on linux.)  i'm not sure what the problem
> >might be.  any ideas?
> >
> >
> what happens with the *good* monitor= value ?