Update load_and_broadcast built-in
Change function suffix to "_i32", etc, from "_32" Improve load_and_broadcast macro in util.m4 to grab vector width from WIDTH variable rather than taking it as a parameter.
This commit is contained in:
@@ -353,10 +353,10 @@ define i64 @__reduce_max_uint64(<16 x i64>) nounwind readnone alwaysinline {
|
||||
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
||||
;; unaligned loads/loads+broadcasts
|
||||
|
||||
load_and_broadcast(16, i8, 8)
|
||||
load_and_broadcast(16, i16, 16)
|
||||
load_and_broadcast(16, i32, 32)
|
||||
load_and_broadcast(16, i64, 64)
|
||||
load_and_broadcast(i8)
|
||||
load_and_broadcast(i16)
|
||||
load_and_broadcast(i32)
|
||||
load_and_broadcast(i64)
|
||||
|
||||
; no masked load instruction for i8 and i16 types??
|
||||
masked_load(16, i8, 8, 1)
|
||||
|
||||
Reference in New Issue
Block a user