Skip to content
This repository was archived by the owner on Dec 22, 2021. It is now read-only.

Commit 81081fe

Browse files
authored
Remove text about flushing of subnormals (#59)
Default to full IEEE 754 support for now. The issue of handling subnormals needs to be addressed (possibly by a global FTZ flag or mode switching), but allowing platforms to behave differently on FTZ semantics should not be baked into the specification.
1 parent 8fb2dba commit 81081fe

File tree

1 file changed

+0
-5
lines changed

1 file changed

+0
-5
lines changed

proposals/simd/SIMD.md

-5
Original file line numberDiff line numberDiff line change
@@ -99,11 +99,6 @@ WebAssembly's scalar floating-point operations. In particular, the rules about
9999
NaN propagation and default NaN values are the same, and all operations use the
100100
default *roundTiesToEven* rounding mode.
101101

102-
An implementation is allowed to flush subnormals in arithmetic floating-point
103-
operations. This means that any subnormal operand is treated as 0, and any
104-
subnormal result is rounded to 0. Note that this differs from WebAssembly
105-
scalar floating-point semantics which require correct subnormal handling.
106-
107102
# JavaScript API and SIMD Values
108103

109104
Accessing WebAssembly module imports or exports containing SIMD Type from JavaScript will throw.

0 commit comments

Comments
 (0)