- Dec 26, 2015
-
-
Raymond Toy authored
This makes it quite a bit easier to run the test suite instead of trying to remember exactly how to invoke it from the command line.
-
Raymond Toy authored
-
Raymond Toy authored
On sparc and ppc (setf vm:floating-point-modes) takes an (unsigned-byte 32) arg, so adjust the ldb byte appopriately. Clean up code by putting the docstring into the macro.
-
- Dec 25, 2015
-
-
Raymond Toy authored
Also replae WITH-INXACT-EXCEPTION-ENABLED with WITH-FLOAT-TRAPS-ENABLED. All tests still pass, as expected.
-
Raymond Toy authored
When enabling traps, need to take just the low 24 bits of the arg because (setf floating-point-modes) wants an (unsigned-byte 24) argument. The logorc2 makes the result negative when enabling traps.
-
Raymond Toy authored
WITH-FLOAT-TRAPS-ENABLED.
-
- Dec 24, 2015
-
-
Raymond Toy authored
Merge the body of both macros into one since they only differ in how the bits are merged with the actual mode bits.
-
Raymond Toy authored
This works like WITH-FLOAT-TRAPS-MASKED, except that the specified traps are enabled. Use this in fdlibm to enable the inexact trap.
-
Raymond Toy authored
Note that original code didn't actually signal inexact probably because the compiler constant-folded one - tiny to one.
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
- Dec 23, 2015
-
-
Raymond Toy authored
o Add tests for this o Use setexception for inexact in e_exp.c.
-
Raymond Toy authored
-
Raymond Toy authored
o Add tests for this o Use setexception for inexact in e_asin.c.
-
Raymond Toy authored
-
Raymond Toy authored
o Update fdlibm.h and setexception.c to support the inexact execption. o Use this in asinh. Tests pass.
-
Raymond Toy authored
-
- Dec 20, 2015
-
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
- Dec 19, 2015
-
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
- Dec 12, 2015
-
-
Raymond Toy authored
Define new vops for the digit shifters that take a constant (unsigned-byte 5) value. The previous version, while correct, still causes the ecx register to spill because it was a temporary. This doens't cause the compiler to spill ecx unnecessarily anymore.
- Dec 06, 2015
-
-
Raymond Toy authored
-
Raymond Toy authored
-
- Dec 03, 2015
-
-
Raymond Toy authored
Cut and paste error from unix-glibc2.lisp copied unix-getitimer instead of unix-setitimer which is needed by SAVE-LISP.
-
- Dec 02, 2015
-
-
Raymond Toy authored
Make the vops handle the case when the known constant shift amount is so large that the result is a known value. Plus, the instructions have a fixed immediate argument size and the amount is taken mod 32 which would produce the wrong result if the actual shift amount were used.
-
- Nov 29, 2015
-
-
Raymond Toy authored
-
Raymond Toy authored
This gets rid of the load of the shift amount to ecx, saving one instruction and reducing pressure on the ecx register.
-
- Nov 11, 2015
-
-
Raymond Toy authored
Can't ever remember the order of arguments for RENAME-PACKAGE so just rename the variables to make it clearer. And update the docstring to be a little more verbose.
-
- Nov 01, 2015
-