From 57c49a74f9f4baf3ba864bbce35d975efd18a8e3 Mon Sep 17 00:00:00 2001
From: John Hauser
The source code for TestFloat is intended to be relatively machine-independent.
Most programs in the TestFloat package should be compilable with any
-ISO-standard C compiler that also supports
@@ -311,7 +315,7 @@ are changed.
testfloat
program will be used to test a new
floating-point implementation, additional effort will likely be required to
retarget that program to invoke the new floating-point operations.
@@ -205,8 +205,10 @@ The supplied directory structure is as follows:
subj-C
build
template
- Win32-MinGW
Linux-386-GCC
+ Linux-x86_64-GCC
+ Win32-MinGW
+ Win64-MinGW-w64
The majority of the TestFloat sources are provided in the source
directory.
@@ -240,9 +242,11 @@ Ignoring the template
directory, the supplied target directories
are intended to follow a naming system of
<execution-environment>-<compiler>
<execution-environment>
Win32
-and Linux-386
, and <compiler>
MinGW
and GCC
, respectively.
+<execution-environment>
Linux-386
Linux-x86_64
Win32
, or Win64
, and
+<compiler>
GCC
,
+MinGW
, or MinGW-w64
SOFTFLOAT_LIB
softfloat.a
).
+softfloat.a
or libsoftfloat.a
).
rand
is better on a given target platform, or if another,
better random number generator is available (such as rand48
on
-most UNIX-derived systems), TestFloat can be improved by overriding the given
+UNIX-derived systems), TestFloat can be improved by overriding the given
random.c
with a target-specific one.
--
cgit v1.1