|
|
Message-ID: <20261029215739.166761-2-bernie@codewiz.org> Date: Fri, 30 Oct 2026 06:57:39 +0900 From: Bernie Innocenti <bernie@...ewiz.org> To: musl@...ts.openwall.com Cc: Bernie Innocenti <bernie@...ewiz.org> Subject: [PATCH libc-test] math: add fma(x,y,0) double rounding test cases x*y must round once even when the arithmetic has excess precision (FLT_EVAL_METHOD==2): on i386 these fail with x*y rounded twice. --- src/math/special/fma.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/math/special/fma.h b/src/math/special/fma.h index d63f816..918575f 100644 --- a/src/math/special/fma.h +++ b/src/math/special/fma.h @@ -849,6 +849,8 @@ T(RD, -0x1p-1000, 0x1p-200, 0x1p-1022, T(RZ, -0x1p-1000, 0x1p-200, 0x1p-1022, 0x1.ffffffffffffep-1023, -0x1p+0, INEXACT|UNDERFLOW) T(RN, -0x1p-1000, 0x1p-100, 0x0p+0, -0x0p+0, 0x1p-26, INEXACT|UNDERFLOW) T(RN, -0x1p-1000, 0x1p-100, -0x0p+0, -0x0p+0, 0x1p-26, INEXACT|UNDERFLOW) +T(RN, 0x1.27365a6456ba5p+0, 0x1.63fdaf7759a4fp+0, 0x0p+0, 0x1.9a84ea6842dcdp+0, 0x1.ffe0e2p-2, INEXACT) +T(RN, 0x1.27365a6456ba5p+0, 0x1.63fdaf7759a4fp+0, -0x0p+0, 0x1.9a84ea6842dcdp+0, 0x1.ffe0e2p-2, INEXACT) T(RZ, -0x1p-1000, 0x1p-100, 0x0p+0, -0x0p+0, 0x1p-26, INEXACT|UNDERFLOW) T(RZ, -0x1p-1000, 0x1p-100, -0x0p+0, -0x0p+0, 0x1p-26, INEXACT|UNDERFLOW) T(RD, -0x1p-1000, 0x1p-100, 0x0p+0, -0x1p-1074, -0x1p+0, INEXACT|UNDERFLOW) -- 2.56.0
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.