[Bug target/93588] Vectorized load followed by FMA pessimizes on Haswell from version 8.1

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93588

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> ---
It is because load/store costs of vector registers for register allocator
are too low.  This patch fixes it:

diff --git a/gcc/config/i386/x86-tune-costs.h
b/gcc/config/i386/x86-tune-costs.h
index c73917e5a62..54b9dad932c 100644
--- a/gcc/config/i386/x86-tune-costs.h
+++ b/gcc/config/i386/x86-tune-costs.h
@@ -2574,9 +2574,9 @@ struct processor_costs core_cost = {
   {6, 6},                              /* cost of storing MMX registers
                                           in SImode and DImode */
   2, 2, 4,                             /* cost of moving XMM,YMM,ZMM register
*/
-  {6, 6, 6, 6, 12},                    /* cost of loading SSE registers
+  {6, 6, 6, 10, 20},                   /* cost of loading SSE registers
                                           in 32,64,128,256 and 512-bit */
-  {6, 6, 6, 6, 12},                    /* cost of storing SSE registers
+  {6, 6, 6, 12, 24},                   /* cost of storing SSE registers
                                           in 32,64,128,256 and 512-bit */
   6, 6,                                        /* SSE->integer and
integer->SSE moves */
   /* End of register allocator costs.  */