ahmed/OpenCL-CTS
1
0
Fork 0
mirror of https://github.com/KhronosGroup/OpenCL-CTS.git synced 2026-03-19 06:09:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

math_brute_force: only test correctly rounded divide/sqrt for fp32 (#1997)

Browse Source 
Skip the correctly rounded divide (divide_cr) and sqrt (sqrt_cr) tests
for fp16 and fp64.

The corresponding build option to enable correctly rounded divide and
sqrt is named `-cl-fp32-correctly-rounded-divide-sqrt` and the
description refers only to "single precision floating-point", so this
option should not apply to fp16 or fp64.

The specification states that fp16 and fp64 divide and sqrt must be
correctly rounded for the full profile, without needing any additional
build options. This is already tested by the regular divide and sqrt
tests. For the embedded profile the ULP requirement is non-zero, but
there is no build option to request a correctly rounded implementation
anyway.

Fixes https://github.com/KhronosGroup/OpenCL-CTS/issues/1901 .
Relates to https://github.com/KhronosGroup/OpenCL-CTS/issues/1996 .

Signed-off-by: Sven van Haastregt <sven.vanhaastregt@arm.com>
This commit is contained in:
Sven van Haastregt
2024-08-06 18:29:08 +02:00
committed by GitHub
parent 13848621f1
commit 1a9d41867e
1 changed files with 21 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
test_conformance/math_brute_force/function_list.cpp
View File

@@ -65,6 +65,7 @@
#define binaryOF NULL
#define binaryF_nextafter NULL
#define binaryOperatorF NULL
#define binaryOperatorOF NULL
#define binaryF_i NULL
#define macro_binaryF NULL
#define ternaryF NULL
@@ -169,6 +170,13 @@ static constexpr vtbl _binary_operator = {
TestFunc_Half_Half_Half_Operator,
};
static constexpr vtbl _binary_operator_of = {
"binaryOperator_of",
TestFunc_Float_Float_Float_Operator,
nullptr,
nullptr,
};
static constexpr vtbl _binary_i = {
"binary_i",
TestFunc_Float_Float_Int,
@@ -227,6 +235,7 @@ static constexpr vtbl _mad_tbl = {
#define binaryF_nextafter &_binary_nextafter
#define binaryOF &_binaryof
#define binaryOperatorF &_binary_operator
#define binaryOperatorOF &_binary_operator_of
#define binaryF_i &_binary_i
#define macro_binaryF &_macro_binary
#define ternaryF &_ternary
@@ -370,17 +379,17 @@ const Func functionList[] = {
{ "sqrt_cr",
"sqrt",
{ (void*)reference_sqrt },
{ (void*)reference_sqrtl },
{ nullptr },
{ NULL },
0.0f,
0.0f,
0.0f,
0.0f,
INFINITY,
INFINITY,
INFINITY,
INFINITY,
INFINITY,
FTZ_OFF,
RELAXED_OFF,
unaryF },
unaryOF /* only for single precision */ },
// In derived mode it the ulp error is calculated as sin/cos.
// In non-derived mode it is the same as half_tan.
@@ -426,17 +435,17 @@ const Func functionList[] = {
{ "divide_cr",
"/",
{ (void*)reference_divide },
{ (void*)reference_dividel },
{ (void*)reference_relaxed_divide },
{ nullptr },
{ nullptr },
0.0f,
0.0f,
0.0f,
0.0f,
0.f,
INFINITY,
INFINITY,
INFINITY,
INFINITY,
INFINITY,
FTZ_OFF,
RELAXED_OFF,
binaryOperatorF },
binaryOperatorOF /* only for single precision */ },
OPERATOR_ENTRY(multiply, "*", 0.0f, 0.0f, 0.0f, FTZ_OFF, binaryOperatorF),
OPERATOR_ENTRY(assignment, "", 0.0f, 0.0f, 0.0f, FTZ_OFF,
unaryF), // A simple copy operation