[Intel] Add fast_dividef (#902)

Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
intel · Apr 17, 2024 · b10a970 · b10a970
1 parent 9a0605c
commit b10a970
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/python/triton/language/extra/intel/libdevice.py b/python/triton/language/extra/intel/libdevice.py
@@ -127,6 +127,13 @@ def saturatef(arg0, _builder=None):
     }, is_pure=True, _builder=_builder)
 
 
+@core.extern
+def fast_dividef(arg0, arg1, _builder=None):
+    return core.extern_elementwise("", "", [arg0, arg1], {
+        (core.dtype("fp32"), core.dtype("fp32")): ("__imf_fast_fdividef", core.dtype("fp32")),
+    }, is_pure=True, _builder=_builder)
+
+
 @core.extern
 def div_rn(arg0, arg1, _builder=None):
     return core.extern_elementwise(