https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116032
--- Comment #7 from Christophe Lyon <clyon at gcc dot gnu.org> ---
The slp2 dump says:
[...]
pr116032.c:6:8: note: ***** Analysis succeeded with vector mode V4SI
pr116032.c:6:8: note: SLPing BB part
pr116032.c:6:8: note: Costing subgraph:
pr116032.c:6:8: note: node 0xb47b7c0 (max_nunits=2, refcnt=1) vector(2) int
pr116032.c:6:8: note: op template: *p_2(D) = 1;
pr116032.c:6:8: note: stmt 0 *p_2(D) = 1;
pr116032.c:6:8: note: stmt 1 MEM[(int *)p_2(D) + 4B] = 0;
pr116032.c:6:8: note: children 0xb47b850
pr116032.c:6:8: note: node (constant) 0xb47b850 (max_nunits=1, refcnt=1)
vector(2) int
pr116032.c:6:8: note: { 1, 0 }
pr116032.c:6:8: note: Cost model analysis:
1 1 times scalar_store costs 1 in body
0 1 times scalar_store costs 1 in body
1 1 times unaligned_store (misalign -1) costs 1 in body
<unknown> 1 times vector_load costs 1 in prologue
pr116032.c:6:8: note: Cost model analysis for part in loop 0:
Vector cost: 2
Scalar cost: 2
pr116032.c:6:8: note: Basic block will be vectorized using SLP
pr116032.c:6:8: optimized: basic block part vectorized using 8 byte vectors
pr116032.c:6:8: note: Vectorizing SLP tree:
pr116032.c:6:8: note: node 0xb47b7c0 (max_nunits=2, refcnt=1) vector(2) int
pr116032.c:6:8: note: op template: *p_2(D) = 1;
pr116032.c:6:8: note: stmt 0 *p_2(D) = 1;
pr116032.c:6:8: note: stmt 1 MEM[(int *)p_2(D) + 4B] = 0;
pr116032.c:6:8: note: children 0xb47b850
pr116032.c:6:8: note: node (constant) 0xb47b850 (max_nunits=1, refcnt=1)
vector(2) int
pr116032.c:6:8: note: { 1, 0 }
pr116032.c:6:8: note: ------>vectorizing SLP node starting from: *p_2(D) = 1;
[...]
