Re: DRM/DRI accelerated (3D) graphics?
Hi Emeric,
Did some more testing with my FireGL X1 (R300) and can confirm:
[drm] initializing kernel modesetting (R300 0x1002:0x4E47).
[drm] register mmio base: 0xD8030000
[drm] register mmio size: 65536
[drm:radeon_agp_init] *ERROR* Unable to acquire AGP: -19
[drm] Forcing AGP to PCI mode
[drm] Generation 1 PCI interface in multifunction mode
This error is always there no matter the patch or not.
More annoyingly I get a corrupted display even in GLXGears with the R300
(although 700fps) where I was getting perfect image but 400fps on the RV610.
I cannot understand why the R300 is still getting the corrupted display,
it got it without any recompiling (which I have done now many times to
test the above) even when just swapping the cards over (same driver so
should work).
Ill continue playing around, but any input, let me know!
Alan
On 08/24/11 09:57, Émeric Maschino wrote:
> Hi Alan,
>
>> Thanks again for the troubleshooting and patches. I applied your patch to my
>> Gentoo (just trying it out I swear!) 2.6.39 system, compiled with KMS on by
>> default on my zx6000 / RV610 (Radeon HD2400) system.
>> I did notice this error in dmesg, but its still working fine as above:
>>
>> [drm:radeon_agp_init] *ERROR* Unable to acquire AGP: -19
>> [drm] Forcing AGP to PCIE mode
> Is this error also reported without the patch?
>
> A noticeable difference between your Radeon HD 2400 and my FireGL X1
> cards is that the RV610 GPU in HD 2400 doesn't natively support AGP.
> Indeed, RV610 is natively a PCIe chipset, AGP support being brought by
> the Rialto PCIe -> AGP bridge on the back of the card PCB.
> By contrast, the R300 VPU in my FireGL card is a native AGP(Pro) chipset.
>
> Could this explain the AGP and PCIe-related errors you're reporting?
>
> Emeric
>
>
Reply to: