Via KT400 agpgart issues

Via KT400 agpgart issues

[Mark Watts]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Ok, I _am_ using the nVidia modules (on 2.4.21) but the fact that the agpgart 
driver can't sense my agp apature size is concerning... (its set to 256MB in 
the bios)

0: nvidia: loading NVIDIA Linux x86 nvidia.o Kernel Module  1.0-4496  Wed Jul 
16 19:03:09 PDT 2003
Linux agpgart interface v0.99 (c) Jeff Hartmann
agpgart: Maximum main memory to use for agp memory: 439M
agpgart: Detected Via Apollo Pro KT400 chipset
agpgart: unable to determine aperture size.
0: NVRM: AGPGART: unable to retrieve symbol table

I'm running a GeForce FX 5600 128Mb with AGP 3.0 support enabled.
It seems to run ok, but its slower than the Ti 4200-4x I took out by a larger 
margin than it really should be (according to the benchmarks I've seen 
scattered through the internet.)
I'm also running with the Local APIC turned on (it doesnt seem to break 
anything).

Is anything actually broken or is this nothing to worry about?

Cheers,

Mark.

- -- 
Mark Watts
Senior Systems Engineer
QinetiQ TIM
St Andrews Road, Malvern
GPG Public Key ID: 455420ED

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD4DBQE/O1V4Bn4EFUVUIO0RAu0VAJip/2qWkG6e6uc/YVYKD5u2f6RwAJ9msH9y
y1VIT2ZaUL5dCQPGoRvjTg==
=a21r
-----END PGP SIGNATURE-----

Re: Via KT400 agpgart issues

[Rahul Karnik]

Mark Watts wrote:
<snip>
> I'm running a GeForce FX 5600 128Mb with AGP 3.0 support enabled.

AFAIK 2.4.21 does not have AGP 3.0 support. Can you turn that off? If 
not, use a more recent kernel (2.6.x has what you need, not sure about 
2.4.22pre).

Thanks,
Rahul
-- 
Rahul Karnik
rahul@genebrew.com

Re: Via KT400 agpgart issues

[Dave Jones]

On Thu, Aug 14, 2003 at 10:25:12AM +0100, Mark Watts wrote:

 > Ok, I _am_ using the nVidia modules (on 2.4.21) but the fact that the agpgart 
 > driver can't sense my agp apature size is concerning... (its set to 256MB in 
 > the bios)
 > 
 > 0: nvidia: loading NVIDIA Linux x86 nvidia.o Kernel Module  1.0-4496  Wed Jul 
 > 16 19:03:09 PDT 2003
 > Linux agpgart interface v0.99 (c) Jeff Hartmann
 > agpgart: Maximum main memory to use for agp memory: 439M
 > agpgart: Detected Via Apollo Pro KT400 chipset
 > agpgart: unable to determine aperture size.
 > 0: NVRM: AGPGART: unable to retrieve symbol table
 > 
 > I'm running a GeForce FX 5600 128Mb with AGP 3.0 support enabled.
 > It seems to run ok, but its slower than the Ti 4200-4x I took out by a larger 
 > margin than it really should be (according to the benchmarks I've seen 
 > scattered through the internet.)
 > I'm also running with the Local APIC turned on (it doesnt seem to break 
 > anything).
 > 
 > Is anything actually broken or is this nothing to worry about?

Caused due to lack of AGP 3.x support in 2.4
See 2.6.0-test

		Dave

-- 
 Dave Jones     http://www.codemonkey.org.uk

Re: Via KT400 agpgart issues

[Dave Jones]

On Thu, Aug 14, 2003 at 11:53:02AM -0700, dleonard@dleonard.net wrote:
 > For my geforce3 and geforce4 I didn't notice anything odd about using the
 > KT400 chipset in spite of agpgart not fully supporting it.

KT400 has two modes of operation. If you have an AGP 2.x graphics card,
the registers are in the same place they are on earlier VIA chipsets.
If you plug in an AGP 3.x card, you have to read from alternative
registers, and some of the legacy ones change meaning.

		Dave


-- 
 Dave Jones     http://www.codemonkey.org.uk

Re: Via KT400 agpgart issues

[dleonard]

For my geforce3 and geforce4 I didn't notice anything odd about using the
KT400 chipset in spite of agpgart not fully supporting it.

For my fx5800 I actually *had* to use the nvidia agp driver to get
anything other than console mode to function.  Now it could be that I just
needed to drop the AGP rate to 4x in bios but frankly I didn't think of
that option until I'd already switched to using the nvidia code.  The
nvidia agp code has been behaving quite well under 2.4.20 and 2.4.22-pre*
for me.

-- 

<Douglas Leonard>
<dleonard@dleonard.net>

On Thu, 14 Aug 2003, Mark Watts wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
>
> Ok, I _am_ using the nVidia modules (on 2.4.21) but the fact that the agpgart
> driver can't sense my agp apature size is concerning... (its set to 256MB in
> the bios)
>
> 0: nvidia: loading NVIDIA Linux x86 nvidia.o Kernel Module  1.0-4496  Wed Jul
> 16 19:03:09 PDT 2003
> Linux agpgart interface v0.99 (c) Jeff Hartmann
> agpgart: Maximum main memory to use for agp memory: 439M
> agpgart: Detected Via Apollo Pro KT400 chipset
> agpgart: unable to determine aperture size.
> 0: NVRM: AGPGART: unable to retrieve symbol table
>
> I'm running a GeForce FX 5600 128Mb with AGP 3.0 support enabled.
> It seems to run ok, but its slower than the Ti 4200-4x I took out by a larger
> margin than it really should be (according to the benchmarks I've seen
> scattered through the internet.)
> I'm also running with the Local APIC turned on (it doesnt seem to break
> anything).
>
> Is anything actually broken or is this nothing to worry about?
>
> Cheers,
>
> Mark.
>
> - --
> Mark Watts
> Senior Systems Engineer
> QinetiQ TIM
> St Andrews Road, Malvern
> GPG Public Key ID: 455420ED
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.2 (GNU/Linux)
>
> iD4DBQE/O1V4Bn4EFUVUIO0RAu0VAJip/2qWkG6e6uc/YVYKD5u2f6RwAJ9msH9y
> y1VIT2ZaUL5dCQPGoRvjTg==
> =a21r
> -----END PGP SIGNATURE-----
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

Re: Via KT400 agpgart issues

[Mark Watts]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


> On Thu, Aug 14, 2003 at 10:25:12AM +0100, Mark Watts wrote:
>  > Ok, I _am_ using the nVidia modules (on 2.4.21) but the fact that the
>  > agpgart driver can't sense my agp apature size is concerning... (its set
>  > to 256MB in the bios)
>  >
>  > 0: nvidia: loading NVIDIA Linux x86 nvidia.o Kernel Module  1.0-4496 
>  > Wed Jul 16 19:03:09 PDT 2003
>  > Linux agpgart interface v0.99 (c) Jeff Hartmann
>  > agpgart: Maximum main memory to use for agp memory: 439M
>  > agpgart: Detected Via Apollo Pro KT400 chipset
>  > agpgart: unable to determine aperture size.
>  > 0: NVRM: AGPGART: unable to retrieve symbol table
>  >
>  > I'm running a GeForce FX 5600 128Mb with AGP 3.0 support enabled.
>  > It seems to run ok, but its slower than the Ti 4200-4x I took out by a
>  > larger margin than it really should be (according to the benchmarks I've
>  > seen scattered through the internet.)
>  > I'm also running with the Local APIC turned on (it doesnt seem to break
>  > anything).
>  >
>  > Is anything actually broken or is this nothing to worry about?
>
> Caused due to lack of AGP 3.x support in 2.4
> See 2.6.0-test

I thought the KT400 chipset (at least) had been backported to 2.4.21 ?

I'm not experiancing any technical problems. The nvidia /proc interfaces are 
reporting that I'm running at agp 8x, and I have full hardware acceleration 
enough to play games like Unreal Tournament 2003 quite fast (~50fps), its 
just that agpgart cant find my aperature size...

I just tried my vendors (mandrake) 2.4.22 kernel and I get exactly the same 
output.

- -- 
Mark Watts
Senior Systems Engineer
QinetiQ TIM
St Andrews Road, Malvern
GPG Public Key ID: 455420ED

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE/PJLHBn4EFUVUIO0RAmpxAJ498GmynIcnEJuFHsqpEOwwo+41JQCg1Dzt
R1AKFPcR3kZzAz/s1/NyKbk=
=4oQ+
-----END PGP SIGNATURE-----

Re: Via KT400 agpgart issues

[Dave Jones]

On Fri, Aug 15, 2003 at 08:59:03AM +0100, Mark Watts wrote:

 > I thought the KT400 chipset (at least) had been backported to 2.4.21 ?

Only for AGP2.x mode.

 > I'm not experiancing any technical problems. The nvidia /proc interfaces are 
 > reporting that I'm running at agp 8x

if its using its own agpgart routines, maybe. If it isn't, it's broken,
or lying.

 > , and I have full hardware acceleration 
 > enough to play games like Unreal Tournament 2003 quite fast (~50fps), its 
 > just that agpgart cant find my aperature size...

which happens when agpgart tries to read the AGP2.x size, when the
chipset is in 3.x mode.

 > I just tried my vendors (mandrake) 2.4.22 kernel and I get exactly the same 
 > output.

No vendor that I know of has backported the AGP3 support it its
entirity. The closest to a backport so far is the trainwreck that
is the ATI FireGL drivers.

		Dave

-- 
 Dave Jones     http://www.codemonkey.org.uk