Commit 0e9ec869 authored by Zack Rusin's avatar Zack Rusin

draw: cleanup and fix instance id computation

The instance id system value always starts at 0, even if the
specified start instance is larger than 0. Instead of implicitly
setting instance id to instance id plus start instance and then
having to subtract instance id when computing the buffer offsets
lets just set instance id to the proper instance id. This fixes
instance id computation and cleansup buffer offset computation.
Signed-off-by: default avatarZack Rusin <zackr@vmware.com>
Reviewed-by: default avatarRoland Scheidegger <sroland@vmware.com>
Reviewed-by: default avatarBrian Paul <brianp@vmware.com>
Reviewed-by: default avatarJose Fonseca <jfonseca@vmware.com>
parent 0ac31647
......@@ -707,15 +707,14 @@ generate_fetch(struct gallivm_state *gallivm,
if (velem->instance_divisor) {
/* Index is equal to the start instance plus the number of current
* instance divided by the divisor. In this case we compute it as:
* index = start_instance + ((instance_id - start_instance) / divisor)
* index = start_instance + (instance_id / divisor)
*/
LLVMValueRef current_instance;
index = lp_build_const_int32(gallivm, draw->start_instance);
current_instance = LLVMBuildSub(builder, instance_id, index, "");
current_instance = LLVMBuildUDiv(builder, current_instance,
current_instance = LLVMBuildUDiv(builder, instance_id,
lp_build_const_int32(gallivm, velem->instance_divisor),
"instance_divisor");
index = LLVMBuildAdd(builder, index, current_instance, "instance");
index = lp_build_uadd_overflow(gallivm, index, current_instance, &ofbit);
}
stride = lp_build_umul_overflow(gallivm, vb_stride, index, &ofbit);
......
......@@ -542,11 +542,12 @@ draw_vbo(struct draw_context *draw,
*/
for (instance = 0; instance < info->instance_count; instance++) {
draw->instance_id = instance + info->start_instance;
unsigned instance_idx = instance + info->start_instance;
draw->start_instance = info->start_instance;
draw->instance_id = instance;
/* check for overflow */
if (draw->instance_id < instance ||
draw->instance_id < info->start_instance) {
if (instance_idx < instance ||
instance_idx < draw->start_instance) {
/* if we overflown just set the instance id to the max */
draw->instance_id = 0xffffffff;
}
......
......@@ -625,8 +625,7 @@ static ALWAYS_INLINE void PIPE_CDECL generic_run_one( struct translate_generic *
if (tg->attrib[attr].instance_divisor) {
index = start_instance;
index += (instance_id - start_instance) /
tg->attrib[attr].instance_divisor;
index += (instance_id / tg->attrib[attr].instance_divisor);
/* XXX we need to clamp the index here too, but to a
* per-array max value, not the draw->pt.max_index value
* that's being given to us via translate->set_buffer().
......
......@@ -1094,10 +1094,6 @@ static boolean init_inputs( struct translate_sse *p,
struct x86_reg tmp_EDX = p->tmp2_EDX;
struct x86_reg tmp_ECX = p->src_ECX;
/* instance_num = instance_id - start_instance */
x86_mov(p->func, tmp_EDX, start_instance);
x86_sub(p->func, tmp_EAX, tmp_EDX);
/* TODO: Add x86_shr() to rtasm and use it whenever
* instance divisor is power of two.
*/
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment