From 06b20075dd3c1a5d0dd007a93643282832221612 Mon Sep 17 00:00:00 2001 From: John Hauser Date: Fri, 26 Jan 2018 12:52:36 -0800 Subject: Release 3e. See "doc/TestFloat-history.html". --- doc/TestFloat-general.html | 60 +++++++++++++++++++++++----------------------- doc/TestFloat-history.html | 34 +++++++++++++++++++++++--- doc/TestFloat-source.html | 23 +++++++++--------- doc/testfloat.html | 21 ++++++++++++++-- doc/testfloat_gen.html | 4 ++-- doc/testfloat_ver.html | 21 ++++++++++++++-- doc/testsoftfloat.html | 4 ++-- doc/timesoftfloat.html | 4 ++-- 8 files changed, 117 insertions(+), 54 deletions(-) (limited to 'doc') diff --git a/doc/TestFloat-general.html b/doc/TestFloat-general.html index c82f06e..55f67d7 100644 --- a/doc/TestFloat-general.html +++ b/doc/TestFloat-general.html @@ -7,11 +7,11 @@ -

Berkeley TestFloat Release 3d: General Documentation

+

Berkeley TestFloat Release 3e: General Documentation

John R. Hauser
-2017 August 18
+2018 January 20

@@ -78,16 +78,24 @@ Details about the standard are available elsewhere.

-The current version of TestFloat is Release 3d. -This version differs from earlier releases 3b and 3c in only minor -ways. -Compared to the original Release 3, Release 3b added -the ability to test the 16-bit half-precision format, while +The current version of TestFloat is Release 3e. +This version differs from earlier releases 3b through 3d in only minor ways. +Compared to the original Release 3: +

+This release adds a few more small improvements, including modifying the +expected behavior of rounding mode odd and fixing a minor bug in +the all-in-one testfloat program.

@@ -124,7 +132,7 @@ bugs can be found through links posted on the TestFloat Web page, The TestFloat package was written by me, John R. Hauser. Release 3 of TestFloat was a completely new implementation supplanting earlier releases. -The project to create Release 3 (now through 3d) was +The project to create Release 3 (now through 3e) was done in the employ of the University of California, Berkeley, within the Department of Electrical Engineering and Computer Sciences, first for the Parallel Computing Laboratory (Par Lab) and then for the ASPIRE Lab. @@ -158,12 +166,12 @@ Oracle, and Samsung.

-The following applies to the whole of TestFloat Release 3d as well +The following applies to the whole of TestFloat Release 3e as well as to each source file individually.

-Copyright 2011, 2012, 2013, 2014, 2015, 2016, 2017 The Regents of the +Copyright 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018 The Regents of the University of California. All rights reserved.

@@ -280,8 +288,9 @@ Any NaN is considered as good a result as another. This laxness can be overridden so that TestFloat checks for particular bit patterns within NaN results. See section 8 below, Variations Allowed by the IEEE -Floating-Point Standard, plus the -checkNaNs option documented -for programs testfloat_ver and testfloat. +Floating-Point Standard, plus the -checkNaNs and +-checkInvInts options documented for programs +testfloat_ver and testfloat.

@@ -585,13 +594,6 @@ Conversions to integer types are expected never to raise the overflow exception.

-

-When converting to an integer, if the rounding mode is odd -(possible only when the rounding mode is not in the function name), TestFloat -expects the result to be rounded not to an odd integer but rather to -minimum magnitude, the same as when the rounding mode is minMag. -

-

6.2. Basic Arithmetic Operations

@@ -691,13 +693,6 @@ This irregular system follows the IEEE Standard’s particular specification for the round-to-integer operations.

-

-If the rounding mode is odd (possible only when the rounding mode -is not in the function name), TestFloat expects the result to be rounded -not to an odd integer but rather to minimum magnitude, the same as -when the rounding mode is minMag. -

-

6.6. Comparison Operations

@@ -1131,7 +1126,12 @@ exception if the source value cannot be rounded to a representable integer. In such cases, TestFloat expects the result value to be the largest-magnitude positive or negative integer or zero, as detailed earlier in section 6.1, Conversion Operations. -The current version of TestFloat provides no means to alter these expectations. +If option -checkInvInts is selected with programs +testfloat_ver and testfloat, integer results of +invalid operations are checked for an exact match. +In order for this option to be sensible, TestFloat must have been compiled so +that its internal floating-point implementation (SoftFloat) generates the +proper integer results for the system being tested.

diff --git a/doc/TestFloat-history.html b/doc/TestFloat-history.html index ed34043..1c247de 100644 --- a/doc/TestFloat-history.html +++ b/doc/TestFloat-history.html @@ -7,11 +7,11 @@ -

History of Berkeley TestFloat, to Release 3d

+

History of Berkeley TestFloat, to Release 3e

John R. Hauser
-2017 August 18
+2018 January 20

@@ -23,13 +23,41 @@ corresponding release of SoftFloat.

+

Release 3e (2018 January)

+ + + +

Release 3d (2017 August)