dexorder/contract
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

WIP: IEEE754

Browse Source
This commit is contained in:
7400
2023-11-17 08:30:23 -08:00
parent e9fd472ec1
commit 8b871940eb
2 changed files with 41 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
test/TestIEEE754.py
View File

@@ -20,9 +20,30 @@ def fixedPoint(n, shift) :
else :
return Fraction(n) / (1<<-shift)
number = fixedPoint(0xffffff, 127-23) # Largest number
number = fixedPoint(0x1, -128) # Smallest number
def print_float_fix(number) :
float_int32, float_bytes = to_float_bits(number)
print (f"float hex: 0x{float_int32:08x} 0x{int(number*(1<<128)):x}")
# print ("float hex:", float_bytes.hex(), hex(int(number*(1<<128))))
float_int32, float_bytes = to_float_bits(number)
print ("float hex:", float_bytes.hex(), hex(int(number*(1<<128))))
print ("float hex:", hex(float_int32), hex(int(number*(1<<128))))
print("Assuming 128.128 fixed point")
print("largest number:")
number = fixedPoint(0xffffff, 127-24) # Largest number that converts to FP and fits in 128.128
assert number < 1<<256
# print (hex(1<<255))
# print (hex(int(number)<<128))
print_float_fix(number)
print("negative largest number:")
number = fixedPoint(-0xffffff, 127-24) # Largest number
print_float_fix(number)
print("smallest number:")
number = fixedPoint(0x1, -128) # Smallest number
print_float_fix(number)
print("negative smallest number:")
number = fixedPoint(-0x1, -128)
print_float_fix(number)
print()
# number = fixedPoint(0x7fffff, 127-23)
# print_float_fix(number)
# number = fixedPoint(0xffffff, 127-23)
# print_float_fix(number)