Author: mattip <[email protected]>
Branch: numpypy.float16
Changeset: r58668:da4184499945
Date: 2012-10-30 02:37 +0200
http://bitbucket.org/pypy/pypy/changeset/da4184499945/
Log: check overflow, underflow in halffloat bitconvert to float
diff --git a/pypy/module/micronumpy/test/test_halffloat.py
b/pypy/module/micronumpy/test/test_halffloat.py
--- a/pypy/module/micronumpy/test/test_halffloat.py
+++ b/pypy/module/micronumpy/test/test_halffloat.py
@@ -10,8 +10,8 @@
def test_bitconvert_exact(self):
#from _numpypy import array, uint32
# These test cases were created by
+ # numpy.float32(v).view(uint32)
# numpy.float16(v).view(uint16)
- # numpy.float32(v).view(uint32)
cases = [[0., 0, 0], [10, 1092616192, 18688], [-10, 3240099840,
51456],
[10e3, 1176256512, 28898], [float('inf'), 2139095040, 31744],
[-float('inf'), 4286578688, 64512]]
@@ -23,6 +23,8 @@
f = self.halffloat.halfbits_to_floatbits(hbits)
assert [f, v] == [fbits, v]
def test_bitconvert_inexact(self):
+ # finexact is
+ # numpy.float32(numpy.float16(v)).view(uint32)
cases = [[10.001, 1092617241, 1092616192, 18688],
[-10.001, 3240100889, 3240099840, 51456],
[22001.0, 1185669632, 1185669120, 30047],]
@@ -31,4 +33,14 @@
assert [f, v] == [hbits, v]
f = self.halffloat.halfbits_to_floatbits(hbits)
assert [f, v] == [finexact, v]
-
+ def test_bitconvert_overunderfloat(self):
+ cases = [[67000.0, 1199758336, 2139095040, 31744],
+ [-67000.0, 3347241984, 4286578688, 64512],
+ [1e-08, 841731191, 0, 0],
+ [-1e-08, 2989214839, 2147483648, 32768],
+ ]
+ for v, fexact, finexact, hbits in cases:
+ f = self.halffloat.floatbits_to_halfbits(fexact)
+ assert [f, v] == [hbits, v]
+ f = self.halffloat.halfbits_to_floatbits(hbits)
+ assert [f, v] == [finexact, v]
_______________________________________________
pypy-commit mailing list
[email protected]
http://mail.python.org/mailman/listinfo/pypy-commit
