blender
/
blender
134
398
Fork 581
Code Issues 6.3k Pull Requests 757 Projects 19 Wiki Activity

Fix ubo vec3 alignment issue 

This fixes the Principled shader in Eevee, among other nodes.

Basically before we were treating all the vec3 as vec4 as far as memory
goes. We now only do it when required (aka, when the vec3 is not
followed by a float).

We can be even smarter about that and move the floats around to provide
padding for the vec3s. However this is for a separate patch.

That said, there seems to be some strong consensus in corners of the
internet against using vec3 at all [1]. Basically even if we get all the
padding correct, we may still suffer from poor driver implementations in
some consumer graphic cards.

It's not hard to move to vec4, but I think we can avoid doing it as much
as possible. By the time 2.8 is out hopefully most drivers will be
implementing things correctly.

[1] - https://stackoverflow.com/questions/38172696
This commit is contained in:
Dalai Felinto 2017-08-18 12:00:30 +02:00
parent 6fe38bf1cc
commit f4bd416b34
1 changed files with 25 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
source/blender/gpu/intern/gpu_uniformbuffer.c
View File

@ -76,6 +76,7 @@ struct GPUUniformBufferDynamicItem {
/* Prototypes */
static GPUType get_padded_gpu_type(struct LinkData *link);
static void gpu_uniformbuffer_inputs_sort(struct ListBase *inputs);
static GPUUniformBufferDynamicItem *gpu_uniformbuffer_populate(
@ -153,7 +154,8 @@ GPUUniformBuffer *GPU_uniformbuffer_dynamic_create(ListBase *inputs, char err_ou
for (LinkData *link = inputs->first; link; link = link->next) {
GPUInput *input = link->data;
gpu_uniformbuffer_populate(ubo, input->type, input->dynamicvec);
GPUType gputype = get_padded_gpu_type(link);
gpu_uniformbuffer_populate(ubo, gputype, input->dynamicvec);
}
ubo->data = MEM_mallocN(ubo->buffer.size, __func__);
@ -224,6 +226,26 @@ void GPU_uniformbuffer_dynamic_update(GPUUniformBuffer *ubo_)
ubo->flag &= ~GPU_UBO_FLAG_DIRTY;
}
/**
* We need to pad some data types (vec3) on the C side
* To match the GPU expected memory block alignment.
*/
static GPUType get_padded_gpu_type(LinkData *link)
{
GPUInput *input = link->data;
GPUType gputype = input->type;
/* Unless the vec3 is followed by a float we need to treat it as a vec4. */
if (gputype == GPU_VEC3 &&
(link->next != NULL) &&
(((GPUInput *)link->next->data)->type != GPU_FLOAT))
{
gputype = GPU_VEC4;
}
return gputype;
}
/**
* Returns 1 if the first item shold be after second item.
* We make sure the vec4 uniforms come first.
@ -254,12 +276,9 @@ static GPUUniformBufferDynamicItem *gpu_uniformbuffer_populate(
BLI_assert(gputype <= GPU_VEC4);
GPUUniformBufferDynamicItem *item = MEM_callocN(sizeof(GPUUniformBufferDynamicItem), __func__);
/* Treat VEC3 as VEC4 because of UBO struct alignment requirements. */
GPUType type = gputype == GPU_VEC3 ? GPU_VEC4 : gputype;
item->gputype = type;
item->gputype = gputype;
item->data = num;
item->size = type * sizeof(float);
item->size = gputype * sizeof(float);
ubo->buffer.size += item->size;
ubo->flag |= GPU_UBO_FLAG_DIRTY;