std::decimal

This is an implementation of the proposed C++ ISO/IEC JTC1 SC22 WG21 N2849 std::decimal including the ISO/IEC JTC1 SC22 WG21 N3871 revisions. It is built on the Intel Floating Point Library that does all the hard work.

Intent

I've used this implementation successfully in the past however I wanted build a modern version that relied on templates instead of macros to generate the numerous operators required.

Building

• std::decimal currently requires a C++ 23 compiler. It is built and tested on macOS with Apple Clang from the latest Xcode and on Ubuntu with Clang 21.
• The general_decimal_arithmetic/test_runner requires boost. If you don't want to install this just comment it out in the top level CMakeLists.txt
• Check the supplied Dockerfile for detailed build requirements. This is the environment used to build in Github Actions here.

Floating Point Environment

The specification adds the following functions which have been implemented.

namespace std::decimal {

    int fe_dec_getround();
    int fe_dec_setround(int round);  

}

The specification does not add decimal specific exception state functions as described here. Because the state functions are opaque and we can't really use them I've introduced a decimal specific set by adding DEC_ and _dec_ to the names as follows.

namespace std::decimal {

    constexpr fexcept_t FE_DEC_DIVBYZERO;
    constexpr fexcept_t FE_DEC_INEXACT;
    constexpr fexcept_t FE_DEC_INVALID;
    constexpr fexcept_t FE_DEC_OVERFLOW;
    constexpr fexcept_t FE_DEC_UNDERFLOW;
    constexpr fexcept_t FE_DEC_ALL_EXCEPT;

    struct fenv_t;

    int fe_dec_clearexcept(int except);
    int fe_dec_raiseexcept(int except);
    int fe_dec_setexceptflag(const fexcept_t *pflag, int except);
    int fe_dec_getexceptflag(fexcept_t *pflag, int except);
    int fe_dec_testexcept(int except);
    int fe_dec_getenv(fenv_t *penv);
    int fe_dec_setenv(const fenv_t *penv);
    int fe_dec_holdexcept(fenv_t *penv);
    int fe_dec_updateenv(const fenv_t *penv);

}

Extensions

Support has been added for selectively throwing C++ exceptions.

// Throw std::decimal::exception if any of the exceptions in the except mask are raised.
// NB: Some operations will result in multiple floating point exceptions being raised and
// this will result in a single std::decimal::exception being thrown. The flags property of
/// this exception will include all raised exceptions.
void set_exceptions(int except) noexcept;
// Remove an exceptions in the except mask from the throw list.
void clear_exceptions(int except) noexcept;
// What exceptions have been configured to throw with set_exceptions?
int get_exceptions() noexcept;
// Check the decimal floating point environment exception flags and throw a std:decimal::exception if
// any exceptions have been raised.
// This function is called internally if any exceptions have been enabled with set_exceptions but it
// can also be called manually by the user.
void check_exceptions();

namespace std::decimal {

class exception : public std::exception
{
public:

    exception(fexcept_t flags);
    constexpr fexcept_t flags() const { return m_flags; }
    const char* what() const noexcept override { return m_what.c_str(); }

private:

    fexcept_t m_flags;
    std::string m_what;

};

}

Testing

The library is validated using the General Decimal Arithmetic Test Cases.

• The large majority of failures are not calculation bugs - they're a consequence of how the test harness runs the cases, not the underlying arithmetic. Each .decTest file declares a precision: and an exponent range (maxExponent:/minexponent:) that its expected results assume, but the harness doesn't apply these - every test just runs at the native precision and exponent range of whichever type (decimal32/64/128) is under test. Where a file's declared context is narrower or wider than the type's real limits, results legitimately diverge from what the file expects even though the type itself is behaving correctly for its own precision (e.g. a file assuming 9 significant digits will expect rounding that a native 7-digit decimal32 needs to apply sooner, or won't expect the overflow/underflow that a decimal32's much smaller exponent range hits on values the file treats as perfectly ordinary). Enforcing each file's declared context is a much larger undertaking than a bug fix, and hasn't been done.
• A smaller share of failures come from exceptions genuinely being raised, just not carrying every condition flag the test expects alongside the primary one (most often a missing companion Inexact), or from the underlying Intel library occasionally mislabelling which flag it raises (e.g. signalling Overflow on a power result that's actually underflowing to zero).
• Test cases exercising operations or special values that aren't applicable in a given file's mode (e.g. the subset arithmetic tests don't support NaN/Infinity as operands) or that aren't implemented yet are explicitly skipped, so this column won't necessarily reach zero for every file - though as of writing it's at zero across the entire subset arithmetic suite.

Subset Arithmetic

decimal 32

Type	Operation	Passed	Failed	Skipped
decimal32	abs	59	0	0
decimal32	add	348	44	0
decimal32	base	488	286	0
decimal32	compare	413	0	0
decimal32	comparetotal	428	27	0
decimal32	divide	180	10	0
decimal32	divideint	162	11	0
decimal32	exp	319	0	0
decimal32	fma	9	0	0
decimal32	inexact	112	8	0
decimal32	ln	282	31	0
decimal32	log10	290	23	0
decimal32	max	67	0	0
decimal32	min	67	0	0
decimal32	minus	45	0	0
decimal32	multiply	192	0	0
decimal32	plus	64	0	0
decimal32	power	242	17	0
decimal32	quantize	394	34	0
decimal32	randombound32	2142	258	0
decimal32	randoms	3503	497	0
decimal32	reduce	106	0	0
decimal32	remainder	240	9	0
decimal32	remaindernear	254	9	0
decimal32	rescale	181	22	0
decimal32	rounding	511	217	0
decimal32	samequantum	64	0	0
decimal32	squareroot	2798	19	0
decimal32	subtract	430	89	0
decimal32	testall	0	0	0
decimal32	tointegral	70	0	0
decimal32	trim	93	0	0
decimal32		14553	1611	0

decimal 64

Type	Operation	Passed	Failed	Skipped
decimal64	abs	56	3	0
decimal64	add	253	139	0
decimal64	base	514	260	0
decimal64	compare	413	0	0
decimal64	comparetotal	426	29	0
decimal64	divide	171	19	0
decimal64	divideint	161	12	0
decimal64	exp	319	0	0
decimal64	fma	9	0	0
decimal64	inexact	74	46	0
decimal64	ln	301	12	0
decimal64	log10	296	17	0
decimal64	max	61	6	0
decimal64	min	61	6	0
decimal64	minus	42	3	0
decimal64	multiply	171	21	0
decimal64	plus	61	3	0
decimal64	power	236	23	0
decimal64	quantize	420	8	0
decimal64	randombound32	2177	223	0
decimal64	randoms	3235	765	0
decimal64	reduce	106	0	0
decimal64	remainder	238	11	0
decimal64	remaindernear	252	11	0
decimal64	rescale	199	4	0
decimal64	rounding	224	504	0
decimal64	samequantum	64	0	0
decimal64	squareroot	2817	0	0
decimal64	subtract	303	216	0
decimal64	testall	0	0	0
decimal64	tointegral	70	0	0
decimal64	trim	93	0	0
decimal64		13823	2341	0

decimal 128

Type	Operation	Passed	Failed	Skipped
decimal128	abs	56	3	0
decimal128	add	236	156	0
decimal128	base	536	238	0
decimal128	compare	413	0	0
decimal128	comparetotal	439	16	0
decimal128	divide	171	19	0
decimal128	divideint	161	12	0
decimal128	exp	318	1	0
decimal128	fma	9	0	0
decimal128	inexact	57	63	0
decimal128	ln	303	10	0
decimal128	log10	305	8	0
decimal128	max	61	6	0
decimal128	min	61	6	0
decimal128	minus	42	3	0
decimal128	multiply	169	23	0
decimal128	plus	61	3	0
decimal128	power	236	23	0
decimal128	quantize	422	6	0
decimal128	randombound32	2326	74	0
decimal128	randoms	3042	958	0
decimal128	reduce	106	0	0
decimal128	remainder	238	11	0
decimal128	remaindernear	252	11	0
decimal128	rescale	199	4	0
decimal128	rounding	210	518	0
decimal128	samequantum	64	0	0
decimal128	squareroot	2817	0	0
decimal128	subtract	269	250	0
decimal128	testall	0	0	0
decimal128	tointegral	70	0	0
decimal128	trim	93	0	0
decimal128		13742	2422	0

Extended Arithmetic

decimal 32

Type	Operation	Passed	Failed	Skipped
decimal32	abs	89	0	0
decimal32	add	1862	234	4
decimal32	and	217	62	0
decimal32	base	556	598	16
decimal32	clamp	0	0	132
decimal32	class	52	32	0
decimal32	compare	612	27	0
decimal32	comparesig	598	27	0
decimal32	comparetotal	533	137	0
decimal32	comparetotmag	524	140	0
decimal32	copy	43	0	0
decimal32	copyabs	43	0	0
decimal32	copynegate	43	0	0
decimal32	copysign	111	0	0
decimal32	ddAbs	75	0	0
decimal32	ddAdd	994	95	2
decimal32	ddAnd	215	72	0
decimal32	ddBase	455	492	0
decimal32	ddCanonical	125	21	84
decimal32	ddClass	26	16	0
decimal32	ddCompare	645	4	0
decimal32	ddCompareSig	555	4	0
decimal32	ddCompareTotal	513	100	0
decimal32	ddCompareTotalMag	513	100	0
decimal32	ddCopy	43	0	0
decimal32	ddCopyAbs	43	0	0
decimal32	ddCopyNegate	43	0	0
decimal32	ddCopySign	107	0	0
decimal32	ddDivide	702	15	0
decimal32	ddDivideInt	326	47	0
decimal32	ddEncode	0	0	376
decimal32	ddFMA	1261	115	2
decimal32	ddInvert	83	68	0
decimal32	ddLogB	92	16	0
decimal32	ddMax	254	2	0
decimal32	ddMaxMag	240	2	0
decimal32	ddMin	245	2	0
decimal32	ddMinMag	231	2	0
decimal32	ddMinus	43	0	0
decimal32	ddMultiply	444	2	0
decimal32	ddNextMinus	12	72	0
decimal32	ddNextPlus	12	72	0
decimal32	ddNextToward	187	117	0
decimal32	ddOr	195	42	0
decimal32	ddPlus	43	0	0
decimal32	ddQuantize	635	48	0
decimal32	ddReduce	130	4	0
decimal32	ddRemainder	437	68	0
decimal32	ddRemainderNear	461	68	0
decimal32	ddRotate	69	143	0
decimal32	ddSameQuantum	325	8	0
decimal32	ddScaleB	180	4	0
decimal32	ddShift	70	142	0
decimal32	ddSubtract	434	82	0
decimal32	ddToIntegral	176	2	0
decimal32	ddXor	222	56	0
decimal32	decDouble	0	0	0
decimal32	decQuad	0	0	0
decimal32	decSingle	0	0	0
decimal32	divide	616	15	0
decimal32	divideint	324	65	0
decimal32	dqAbs	75	0	0
decimal32	dqAdd	920	90	2
decimal32	dqAnd	215	142	0
decimal32	dqBase	440	488	0
decimal32	dqCanonical	112	18	114
decimal32	dqClass	26	16	0
decimal32	dqCompare	655	4	0
decimal32	dqCompareSig	555	4	0
decimal32	dqCompareTotal	513	100	0
decimal32	dqCompareTotalMag	513	100	0
decimal32	dqCopy	43	0	0
decimal32	dqCopyAbs	43	0	0
decimal32	dqCopyNegate	43	0	0
decimal32	dqCopySign	107	0	0
decimal32	dqDivide	665	23	0
decimal32	dqDivideInt	326	48	0
decimal32	dqEncode	1	0	367
decimal32	dqFMA	1335	114	2
decimal32	dqInvert	83	110	0
decimal32	dqLogB	93	16	0
decimal32	dqMax	254	2	0
decimal32	dqMaxMag	240	2	0
decimal32	dqMin	245	2	0
decimal32	dqMinMag	231	2	0
decimal32	dqMinus	43	0	0
decimal32	dqMultiply	471	2	0
decimal32	dqNextMinus	12	72	0
decimal32	dqNextPlus	12	72	0
decimal32	dqNextToward	187	117	0
decimal32	dqOr	195	146	0
decimal32	dqPlus	43	0	0
decimal32	dqQuantize	635	51	0
decimal32	dqReduce	130	4	0
decimal32	dqRemainder	436	64	0
decimal32	dqRemainderNear	461	69	0
decimal32	dqRotate	69	179	0
decimal32	dqSameQuantum	325	8	0
decimal32	dqScaleB	198	4	0
decimal32	dqShift	70	178	0
decimal32	dqSubtract	438	82	0
decimal32	dqToIntegral	176	2	0
decimal32	dqXor	222	126	0
decimal32	dsBase	652	257	0
decimal32	dsEncode	0	0	268
decimal32	exp	430	10	0
decimal32	fma	2373	237	2
decimal32	inexact	130	22	0
decimal32	invert	83	45	0
decimal32	ln	374	40	0
decimal32	log10	328	61	0
decimal32	logb	98	30	0
decimal32	max	319	8	0
decimal32	maxmag	304	8	0
decimal32	min	309	8	0
decimal32	minmag	295	8	0
decimal32	minus	113	0	0
decimal32	multiply	508	13	0
decimal32	nextminus	12	92	0
decimal32	nextplus	12	94	0
decimal32	nexttoward	191	150	0
decimal32	or	222	54	0
decimal32	plus	122	0	0
decimal32	power	1026	181	0
decimal32	powersqrt	2826	30	0
decimal32	quantize	672	103	0
decimal32	randombound32	2138	262	0
decimal32	randoms	3503	497	0
decimal32	reduce	163	5	0
decimal32	remainder	486	31	0
decimal32	remaindernear	406	40	0
decimal32	rescale	504	87	26
decimal32	rotate	70	125	0
decimal32	rounding	781	249	0
decimal32	samequantum	325	8	0
decimal32	scaleb	145	10	0
decimal32	shift	71	129	0
decimal32	squareroot	3225	361	0
decimal32	subtract	559	122	0
decimal32	testall	0	0	0
decimal32	tointegral	153	15	0
decimal32	tointegralx	180	0	0
decimal32	trim	110	0	0
decimal32	xor	223	54	0
decimal32		54381	9343	1397

decimal 64

Type	Operation	Passed	Failed	Skipped
decimal64	abs	86	3	0
decimal64	add	1795	301	4
decimal64	and	275	4	0
decimal64	base	602	552	16
decimal64	clamp	0	0	132
decimal64	class	70	14	0
decimal64	compare	627	12	0
decimal64	comparesig	613	12	0
decimal64	comparetotal	548	122	0
decimal64	comparetotmag	539	125	0
decimal64	copy	43	0	0
decimal64	copyabs	43	0	0
decimal64	copynegate	43	0	0
decimal64	copysign	111	0	0
decimal64	ddAbs	75	0	0
decimal64	ddAdd	1089	0	2
decimal64	ddAnd	287	0	0
decimal64	ddBase	690	257	0
decimal64	ddCanonical	125	21	84
decimal64	ddClass	42	0	0
decimal64	ddCompare	649	0	0
decimal64	ddCompareSig	559	0	0
decimal64	ddCompareTotal	515	98	0
decimal64	ddCompareTotalMag	515	98	0
decimal64	ddCopy	43	0	0
decimal64	ddCopyAbs	43	0	0
decimal64	ddCopyNegate	43	0	0
decimal64	ddCopySign	107	0	0
decimal64	ddDivide	717	0	0
decimal64	ddDivideInt	322	51	0
decimal64	ddEncode	0	0	376
decimal64	ddFMA	1376	0	2
decimal64	ddInvert	151	0	0
decimal64	ddLogB	108	0	0
decimal64	ddMax	254	2	0
decimal64	ddMaxMag	240	2	0
decimal64	ddMin	245	2	0
decimal64	ddMinMag	231	2	0
decimal64	ddMinus	43	0	0
decimal64	ddMultiply	446	0	0
decimal64	ddNextMinus	84	0	0
decimal64	ddNextPlus	84	0	0
decimal64	ddNextToward	304	0	0
decimal64	ddOr	237	0	0
decimal64	ddPlus	43	0	0
decimal64	ddQuantize	683	0	0
decimal64	ddReduce	124	10	0
decimal64	ddRemainder	498	7	0
decimal64	ddRemainderNear	522	7	0
decimal64	ddRotate	111	101	0
decimal64	ddSameQuantum	333	0	0
decimal64	ddScaleB	180	4	0
decimal64	ddShift	111	101	0
decimal64	ddSubtract	516	0	0
decimal64	ddToIntegral	176	2	0
decimal64	ddXor	278	0	0
decimal64	decDouble	0	0	0
decimal64	decQuad	0	0	0
decimal64	decSingle	0	0	0
decimal64	divide	603	28	0
decimal64	divideint	324	65	0
decimal64	dqAbs	75	0	0
decimal64	dqAdd	940	70	2
decimal64	dqAnd	280	77	0
decimal64	dqBase	481	447	0
decimal64	dqCanonical	112	18	114
decimal64	dqClass	28	14	0
decimal64	dqCompare	657	2	0
decimal64	dqCompareSig	557	2	0
decimal64	dqCompareTotal	515	98	0
decimal64	dqCompareTotalMag	515	98	0
decimal64	dqCopy	43	0	0
decimal64	dqCopyAbs	43	0	0
decimal64	dqCopyNegate	43	0	0
decimal64	dqCopySign	107	0	0
decimal64	dqDivide	669	19	0
decimal64	dqDivideInt	322	52	0
decimal64	dqEncode	1	0	367
decimal64	dqFMA	1357	92	2
decimal64	dqInvert	83	110	0
decimal64	dqLogB	94	15	0
decimal64	dqMax	254	2	0
decimal64	dqMaxMag	240	2	0
decimal64	dqMin	245	2	0
decimal64	dqMinMag	231	2	0
decimal64	dqMinus	43	0	0
decimal64	dqMultiply	472	1	0
decimal64	dqNextMinus	12	72	0
decimal64	dqNextPlus	12	72	0
decimal64	dqNextToward	187	117	0
decimal64	dqOr	230	111	0
decimal64	dqPlus	43	0	0
decimal64	dqQuantize	676	10	0
decimal64	dqReduce	130	4	0
decimal64	dqRemainder	459	41	0
decimal64	dqRemainderNear	487	43	0
decimal64	dqRotate	69	179	0
decimal64	dqSameQuantum	325	8	0
decimal64	dqScaleB	198	4	0
decimal64	dqShift	70	178	0
decimal64	dqSubtract	512	8	0
decimal64	dqToIntegral	176	2	0
decimal64	dqXor	271	77	0
decimal64	dsBase	501	408	0
decimal64	dsEncode	0	0	268
decimal64	exp	433	7	0
decimal64	fma	2294	316	2
decimal64	inexact	92	60	0
decimal64	invert	83	45	0
decimal64	ln	393	21	0
decimal64	log10	348	41	0
decimal64	logb	101	27	0
decimal64	max	313	14	0
decimal64	maxmag	292	20	0
decimal64	min	303	14	0
decimal64	minmag	295	8	0
decimal64	minus	110	3	0
decimal64	multiply	491	30	0
decimal64	nextminus	12	92	0
decimal64	nextplus	12	94	0
decimal64	nexttoward	189	152	0
decimal64	or	264	12	0
decimal64	plus	119	3	0
decimal64	power	1076	131	0
decimal64	powersqrt	2847	9	0
decimal64	quantize	706	69	0
decimal64	randombound32	2141	259	0
decimal64	randoms	3237	763	0
decimal64	reduce	163	5	0
decimal64	remainder	504	13	0
decimal64	remaindernear	427	19	0
decimal64	rescale	526	67	24
decimal64	rotate	70	125	0
decimal64	rounding	413	617	0
decimal64	samequantum	325	8	0
decimal64	scaleb	149	6	0
decimal64	shift	86	114	0
decimal64	squareroot	3275	311	0
decimal64	subtract	525	156	0
decimal64	testall	0	0	0
decimal64	tointegral	153	15	0
decimal64	tointegralx	180	0	0
decimal64	trim	110	0	0
decimal64	xor	266	11	0
decimal64		55609	8117	1395

decimal 128

Type	Operation	Passed	Failed	Skipped
decimal128	abs	74	15	0
decimal128	add	1296	800	4
decimal128	and	275	4	0
decimal128	base	610	544	16
decimal128	clamp	0	0	132
decimal128	class	72	12	0
decimal128	compare	627	12	0
decimal128	comparesig	613	12	0
decimal128	comparetotal	561	109	0
decimal128	comparetotmag	552	112	0
decimal128	copy	43	0	0
decimal128	copyabs	43	0	0
decimal128	copynegate	43	0	0
decimal128	copysign	111	0	0
decimal128	ddAbs	75	0	0
decimal128	ddAdd	605	484	2
decimal128	ddAnd	287	0	0
decimal128	ddBase	511	436	0
decimal128	ddCanonical	125	21	84
decimal128	ddClass	36	6	0
decimal128	ddCompare	649	0	0
decimal128	ddCompareSig	559	0	0
decimal128	ddCompareTotal	515	98	0
decimal128	ddCompareTotalMag	515	98	0
decimal128	ddCopy	43	0	0
decimal128	ddCopyAbs	43	0	0
decimal128	ddCopyNegate	43	0	0
decimal128	ddCopySign	107	0	0
decimal128	ddDivide	673	44	0
decimal128	ddDivideInt	322	51	0
decimal128	ddEncode	0	0	376
decimal128	ddFMA	931	445	2
decimal128	ddInvert	83	68	0
decimal128	ddLogB	108	0	0
decimal128	ddMax	254	2	0
decimal128	ddMaxMag	240	2	0
decimal128	ddMin	245	2	0
decimal128	ddMinMag	231	2	0
decimal128	ddMinus	43	0	0
decimal128	ddMultiply	370	76	0
decimal128	ddNextMinus	10	74	0
decimal128	ddNextPlus	10	74	0
decimal128	ddNextToward	142	162	0
decimal128	ddOr	237	0	0
decimal128	ddPlus	43	0	0
decimal128	ddQuantize	671	12	0
decimal128	ddReduce	124	10	0
decimal128	ddRemainder	498	7	0
decimal128	ddRemainderNear	522	7	0
decimal128	ddRotate	70	142	0
decimal128	ddSameQuantum	333	0	0
decimal128	ddScaleB	165	19	0
decimal128	ddShift	92	120	0
decimal128	ddSubtract	385	131	0
decimal128	ddToIntegral	176	2	0
decimal128	ddXor	278	0	0
decimal128	decDouble	0	0	0
decimal128	decQuad	0	0	0
decimal128	decSingle	0	0	0
decimal128	divide	603	28	0
decimal128	divideint	324	65	0
decimal128	dqAbs	75	0	0
decimal128	dqAdd	1010	0	2
decimal128	dqAnd	357	0	0
decimal128	dqBase	669	259	0
decimal128	dqCanonical	112	18	114
decimal128	dqClass	42	0	0
decimal128	dqCompare	659	0	0
decimal128	dqCompareSig	559	0	0
decimal128	dqCompareTotal	515	98	0
decimal128	dqCompareTotalMag	515	98	0
decimal128	dqCopy	43	0	0
decimal128	dqCopyAbs	43	0	0
decimal128	dqCopyNegate	43	0	0
decimal128	dqCopySign	107	0	0
decimal128	dqDivide	688	0	0
decimal128	dqDivideInt	323	51	0
decimal128	dqEncode	1	0	367
decimal128	dqFMA	1447	2	2
decimal128	dqInvert	193	0	0
decimal128	dqLogB	109	0	0
decimal128	dqMax	254	2	0
decimal128	dqMaxMag	240	2	0
decimal128	dqMin	245	2	0
decimal128	dqMinMag	231	2	0
decimal128	dqMinus	43	0	0
decimal128	dqMultiply	473	0	0
decimal128	dqNextMinus	84	0	0
decimal128	dqNextPlus	84	0	0
decimal128	dqNextToward	304	0	0
decimal128	dqOr	341	0	0
decimal128	dqPlus	43	0	0
decimal128	dqQuantize	686	0	0
decimal128	dqReduce	124	10	0
decimal128	dqRemainder	493	7	0
decimal128	dqRemainderNear	523	7	0
decimal128	dqRotate	147	101	0
decimal128	dqSameQuantum	333	0	0
decimal128	dqScaleB	198	4	0
decimal128	dqShift	147	101	0
decimal128	dqSubtract	520	0	0
decimal128	dqToIntegral	176	2	0
decimal128	dqXor	348	0	0
decimal128	dsBase	496	413	0
decimal128	dsEncode	0	0	268
decimal128	exp	408	32	0
decimal128	fma	1752	858	2
decimal128	inexact	75	77	0
decimal128	invert	83	45	0
decimal128	ln	399	15	0
decimal128	log10	355	34	0
decimal128	logb	114	14	0
decimal128	max	307	20	0
decimal128	maxmag	280	32	0
decimal128	min	297	20	0
decimal128	minmag	295	8	0
decimal128	minus	98	15	0
decimal128	multiply	451	70	0
decimal128	nextminus	10	94	0
decimal128	nextplus	10	96	0
decimal128	nexttoward	151	190	0
decimal128	or	264	12	0
decimal128	plus	103	19	0
decimal128	power	1032	175	0
decimal128	powersqrt	2847	9	0
decimal128	quantize	708	67	0
decimal128	randombound32	2170	230	0
decimal128	randoms	3044	956	0
decimal128	reduce	151	17	0
decimal128	remainder	504	13	0
decimal128	remaindernear	429	17	0
decimal128	rescale	528	65	24
decimal128	rotate	70	125	0
decimal128	rounding	395	635	0
decimal128	samequantum	333	0	0
decimal128	scaleb	137	18	0
decimal128	shift	86	114	0
decimal128	squareroot	3245	341	0
decimal128	subtract	473	208	0
decimal128	testall	0	0	0
decimal128	tointegral	153	15	0
decimal128	tointegralx	180	0	0
decimal128	trim	110	0	0
decimal128	xor	266	11	0
decimal128		53572	10154	1395