Bug 16488 – [spec][optimization] broadcast scalar to simd vector
Status
RESOLVED
Resolution
FIXED
Severity
major
Priority
P1
Component
dmd
Product
D
Version
D2
Platform
All
OS
All
Creation time
2016-09-12T08:49:00Z
Last change time
2017-01-16T23:24:46Z
Keywords
SIMD
Assigned to
nobody
Creator
ilyayaroshenko
Comments
Comment #0 by ilyayaroshenko — 2016-09-12T08:49:07Z
Currently DMD is 10-20(!) times slower for GLAS matrix multiplication then LDC. This is the largest DMD FE problem for GLAS [1].
float s = 3;
...
__vector(float[4]) e = 3; // OK
__vector(float[4]) e = s; // Fails for DMD, but OK for LDC
[1] http://docs.mir.dlang.io/latest/mir_glas_l3.html
Comment #1 by ilyayaroshenko — 2016-09-12T09:00:19Z