[SYCL][CUDA][HIP] Propagate -foffload-fp32-prec-sqrt#17044
[SYCL][CUDA][HIP] Propagate -foffload-fp32-prec-sqrt#17044dm-vodopyanov merged 4 commits intointel:syclfrom
Conversation
It follows the approach from intel#5141 and intel#5309 adding intermediate fcuda-prec-div flag. Signed-off-by: Sidorov, Dmitry <dmitry.sidorov@intel.com>
MrSidims
had a problem deploying
to
WindowsCILock
GitHub Actions
Error
| !llvm.module.flags = !{!0} | ||
| !llvm.module.flags = !{!0, !1, !2} | ||
| !0 = !{i32 4, !"nvvm-reflect-ftz", i32 1} | ||
| !1 = !{i32 4, !"nvvm-reflect-prec-sqrt", i32 1} |
There was a problem hiding this comment.
This line seem to be lost during pulldown (at least git log/blame doesn't show any patches explicitly removing it), so I'm restoring it here.
MrSidims
changed the title
MrSidims
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
MrSidims
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
Signed-off-by: Sidorov, Dmitry <dmitry.sidorov@intel.com>
MrSidims
changed the title
MrSidims
had a problem deploying
to
WindowsCILock
GitHub Actions
Error
Signed-off-by: Sidorov, Dmitry <dmitry.sidorov@intel.com>
MrSidims
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
MrSidims
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
|
@intel/dpcpp-tools-reviewers @intel/dpcpp-clang-driver-reviewers please take a look |
mdtoguchi
left a comment
mdtoguchi
left a comment
There was a problem hiding this comment.
LGTM - if you could also update the description to give a quick overview of the usage against the existing -fsycl-fp32-prec-sqrt option.
Would it make sense to work towards deprecating -fsycl-fp32-prec-sqrt?
|
@mdtoguchi deprecation (or renaming of offload options) will be followed in #17033 also added description in the PR |
|
@intel/llvm-gatekeepers please help with merge not sure, if tools team approval is necessary here, the only file that slips to the 'default' owner is llvm/docs/NVPTXUsage.rst , but approval Nicolas in this regard is (IMHO) sufficient. |
5 participants
-foffload-fp32-prec-sqrt and -fsycl-fp32-prec-sqrt options should be merged together as they have the same purpose.
In this patch ability of -fsycl-fp32-prec-sqrt to pass appropriate options to CUDA and HIP compilers was added to
-foffload-fp32-prec-sqrt to allow such merge in the future.