On 27.05.2015 04:29, Grigori Goronzy wrote:
> On 26.05.2015 09:28, Michel Dänzer wrote:
>> From: Michel Dänzer
>>
>> Based on the corresponding SI support. Same as that, this is currently
>> only enabled for one-dimensional buffer copies due to issues with
>> multi-dimensional SDMA copies.
>>
>
>
On Tue, May 26, 2015 at 3:29 PM, Grigori Goronzy wrote:
> On 26.05.2015 09:28, Michel Dänzer wrote:
>> From: Michel Dänzer
>>
>> Based on the corresponding SI support. Same as that, this is currently
>> only enabled for one-dimensional buffer copies due to issues with
>> multi-dimensional SDMA co
I suggest we implement a thorough SDMA test first before trying to fix
it. We definitely want to test all implemented tiling modes,
dimensions, subrectangles of those (for the copies), bit depths,
number of samples, etc. and some of those parameters can differ
between src and dst. The list of teste
On 26.05.2015 09:28, Michel Dänzer wrote:
> From: Michel Dänzer
>
> Based on the corresponding SI support. Same as that, this is currently
> only enabled for one-dimensional buffer copies due to issues with
> multi-dimensional SDMA copies.
>
What a pity, so CIK has exactly the same issues as SI?
On Tue, May 26, 2015 at 3:28 AM, Michel Dänzer wrote:
> From: Michel Dänzer
>
> Based on the corresponding SI support. Same as that, this is currently
> only enabled for one-dimensional buffer copies due to issues with
> multi-dimensional SDMA copies.
>
> Signed-off-by: Michel Dänzer
For the se
From: Michel Dänzer
Based on the corresponding SI support. Same as that, this is currently
only enabled for one-dimensional buffer copies due to issues with
multi-dimensional SDMA copies.
Signed-off-by: Michel Dänzer
---
src/gallium/drivers/radeonsi/Makefile.sources | 1 +
src/gallium/driver
