Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Pull 2019-09-03T14-15 Recent NVIDIA Changes #794

Merged
merged 6 commits into from Sep 3, 2019

Conversation

@gklimowicz
Copy link
Contributor

commented Sep 3, 2019

No description provided.

gklimowicz added 6 commits Aug 29, 2019
Fix issue with bogus error message
Fix a bogus error message that is generated with
an argument check of an interface associated
with a procedure dummy argument in an internal
procedure.

The problem is that the interface symbol becomes
stale when the internal procedure is processed
between semantic pass 0 and 1. The compiler
will propagate symbol information from the
host procedure to the internal procedure but
not always from the internal procedure to the
host procedure. To fix this problem, defer the
argument check until processing the internal
procedure during semantic pass 1.
Fix issue with atan on x86
Fixed issue with the new atan function where
passing the argument(-0.0) was failing to
return -0.0. Fixed both the scalar and vector
versions.

Also changed which scalar function we call for
double precision atan when using the relaxed
version. It turns out that the previous version
of scalar double precision atan function was
slightly faster than our new one, therefore
decided to make a way that it could still
be called.
Fix incorrect DWARF for function parameters
Remove the redundant dereference DW_OP on
formal argument's expression mdnode when the
LL_BaseDataType is non-pointer.
Fix issue #763 MINLOC intrinsic incorrect return result
MINLOC returns last minimum rather than first
as if BACK was specified as .TRUE.
For SCALE intrinsic, correct order of initialization
BZERO 'rslt' after get_item, but before assigning
values to 'rslt'.
Add support for non-decomposed complex operations
Added xbit 201 0x4000000 in order to control alignment
assumption for complex arrays. When enabled, it is
assumed that single complex arrays are 8-byte aligned
and that double complex arrays are 16-byte aligned.

@sscalpone sscalpone merged commit a0a67c5 into flang-compiler:master Sep 3, 2019

@sscalpone sscalpone deleted the ThePortlandGroup:nv_stage branch Sep 3, 2019

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
2 participants
You can’t perform that action at this time.