[Mesa-dev] [PATCH 01/11] gallium: Basic compute interface.

[Marek Olšák]

On Sat, Mar 31, 2012 at 2:29 PM, Francisco Jerez <currojerez at riseup.net> wrote:
> Marek Olšák <maraeo at gmail.com> writes:
>
>> On Fri, Mar 23, 2012 at 1:40 AM, Francisco Jerez <currojerez at riseup.net> wrote:
>>> diff --git a/src/gallium/include/pipe/p_context.h b/src/gallium/include/pipe/p_context.h
>>> index aaeeb81..de990fe 100644
>>> --- a/src/gallium/include/pipe/p_context.h
>>> +++ b/src/gallium/include/pipe/p_context.h
>>> @@ -63,6 +63,7 @@ struct pipe_vertex_element;
>>>  struct pipe_video_buffer;
>>>  struct pipe_video_decoder;
>>>  struct pipe_viewport_state;
>>> +struct pipe_compute_state;
>>>  union pipe_color_union;
>>>
>>>  /**
>>> @@ -140,6 +141,9 @@ struct pipe_context {
>>>    void   (*bind_geometry_sampler_states)(struct pipe_context *,
>>>                                           unsigned num_samplers,
>>>                                           void **samplers);
>>> +   void   (*bind_compute_sampler_states)(struct pipe_context *,
>>> +                                         unsigned num_samplers,
>>> +                                         void **samplers);
>>
>> Hi,
>>
> Hi,
>
>> could you please add an "unsigned start_slot" parameter in
>> bind_compute_sampler_states? The function would then look like this:
>>
>> void   (*bind_compute_sampler_states)(struct pipe_context *,
>>                                       unsigned start_slot,
>>                                       unsigned num_samplers,
>>                                       void **samplers);
>>
>> Where start_slot+num_samplers <= max_samplers, and max_samplers =
>> get(SHADER_CAP_MAX_TEXTURE_SAMPLERS).
>>
>> The reason for this is that state changes will be faster if you only
>> need to change a subset of the bound states (less work for the pipe
>> driver).
>
> This sounds reasonable to me.  I'll make that change on the three entry
> points and resend later today.
>
>> The disadvantage is that the states will have to be unbound
>> using something like this:
>>
>> void *null_ptrs[MAX_SAMPLERS] = {NULL};
>> bind_compute_sampler_states(pipe, 0, max_samplers, null_ptrs);
>>
> Another option would be to take 'samplers == NULL' as meaning that the
> specified slot range has to be unbound, e.g.:
>
> | bind_compute_sampler_states(pipe, 0, max_samplers, NULL);
>
> To keep things as simple as possible in the driver we could add the
> restriction that no 'samplers[i]' will ever be NULL in cases where
> samplers is non-NULL.  The usefulness of being able to bind and unbind
> states at the same time with the same call is quite dubious anyway...

I have kinda already taken the other path with set_vertex_buffers that
null pointers are allowed, but your idea might better. There is
probably not much point in unbinding CSOs (void* states) regularly.
However resources won't be released if they're bound.

Marek