This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [Patch-86512]: Subnormal float support in armv7(with -msoft-float) for intrinsics
- From: Wilco Dijkstra <Wilco dot Dijkstra at arm dot com>
- To: "umesh dot kalappa0 at gmail dot com" <umesh dot kalappa0 at gmail dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>
- Cc: nd <nd at arm dot com>, Szabolcs Nagy <Szabolcs dot Nagy at arm dot com>, Ramana Radhakrishnan <Ramana dot Radhakrishnan at arm dot com>
- Date: Fri, 20 Jul 2018 13:37:54 +0000
- Subject: Re: [Patch-86512]: Subnormal float support in armv7(with -msoft-float) for intrinsics
Hi Umesh,
Looking at your patch, this would break all results which need to be normalized.
Index: libgcc/config/arm/ieee754-df.S
===================================================================
--- libgcc/config/arm/ieee754-df.S (revision 262850)
+++ libgcc/config/arm/ieee754-df.S (working copy)
@@ -203,8 +203,11 @@
#endif
@ Determine how to normalize the result.
+ @ if result is denormal i.e (exp)=0,then don't normalise the result,
LSYM(Lad_p):
cmp xh, #0x00100000
+ blt LSYM(Lad_e)
+ cmp xh, #0x00100000
bcc LSYM(Lad_a)
cmp xh, #0x00200000
bcc LSYM(Lad_e)
It seems Lad_a doesn't correctly handle the case where the result is a denormal. For this case
the result is correct so nothing else needs to be done. This requires an explicit test that the
exponent is zero - other cases still need to be renormalized as usual. This code looks overly
complex so any change will require extensive testing of all the corner cases.
Wilco