Date: Tue, 29 Apr 1997 18:15:55 +0200 (MET DST) From: Christoph Kukulies <kuku@gilberto.physik.rwth-aachen.de> To: freebsd-hackers@freefall.FreeBSD.org Subject: FP problem Message-ID: <199704291615.SAA02704@gilberto.physik.rwth-aachen.de>
next in thread | raw e-mail | index | archive | help
We are running some FPU intense jobs on a PPRO that run a couple of days.
My colleague says that the program ran before but after recompilation
on the PPRO machine he got the errors.
I'm wondering what might have changed. Could the fact that another
job is running in parallel (freebsd-bovrc-client) affect the job?
I don't think so but who knows?
Here is his FPU excerpt:
(gdb) info float
u: status 0x83e1: exceptions: INVALID LOS FPSTACK; flags: 0011; top 0
e: status 0x300: flags: 0011; top 0
control 0x1272: compute to 53 bits; round NEAREST; mask: DENORM UNDERF LOS;
last instruction: opcode 0x1d0; pc 0x8:0xf01c9e7e; operand 0x10:0x0
regno tag msb lsb value
%st(7) valid 3ff7f8d857e26d20c800 0.00759415
%st(6) valid c001ada01ce384b7cd7a -5.4258
%st(5) valid bff7908c816f2bddc000 -0.00441128
%st(4) valid bff5908c816f2bddc000 -0.00110282
%st(3) valid bfe8d0f1965952e82000 -1.94594e-07
%st(2) valid bfe9b1ac57cb15b7a800 -3.30942e-07
%st(1) valid bfe9cb23c3f349790800 -3.78377e-07
%st(0) => valid bfe9d7d9d68f0db23800 -4.02054e-07
He says the expression in question is a relatively complex
fortran expression:
rsp=-z*(1.D0+a1*z*(1.D0+a2*z*(1.D0+a3*z2*(1.D0+a4*z2*
1 (1.D0+a5*z2*(1.D0+a6*z2*(1.D0+a7*z2*(1.D0+a8*z2*(1.D0+a9*z2*
2 (1.D0+a10*z2))))))))))
3 -zeta2-0.5D0*log(-x)**2
And the incoming values all were OK.
--
Chris Christoph P. U. Kukulies kuku@gil.physik.rwth-aachen.de
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199704291615.SAA02704>