[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ppc/fpu_helper.c

From: Matheus K. Ferst
Subject: Re: ppc/fpu_helper.c
Date: Tue, 16 Nov 2021 09:36:27 -0300
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.13.0

On 14/11/2021 07:15, 罗勇刚(Yonggang Luo) wrote:
On Sun, Nov 14, 2021 at 5:07 PM Richard Henderson <richard.henderson@linaro.org <mailto:richard.henderson@linaro.org>> wrote:
 > On 11/13/21 11:07 AM, 罗勇刚(Yonggang Luo) wrote:
 > > I've seen nothing in fpu_helper.c to update
 > > the fpscr to FP_FR,
 > > that is there is no code with `fpscr |= FP_FR`,
 > > is that valid for PowerPC? or it's just because this is hard
 > > to implement and then the FP_FR  are always not set  for  fpscr
> It is unimplemented, yes.  I think that no one has spent the time; I don't think that it
 > should be hard, necessarily.

Thanks, I also have a question, where is the complete PowerPC floating point instrunctions tests case? I wanna improve the performance of powerpc floating point calculation and don't lost the accuracy of it
So I need a complete testsuite for it.

AFAIK we only have the multiarch tests, float_convs and float_madds. You'll need the reference files in tests/tcg/ppc{,64{,le}}/ to run them. I'm attaching the ppc64le ones (obtained from a POWER9), but you can also generate them by running the test in the real hardware and saving the output (e.g.: ./float_convs > float_convs.ref).

However, both tests currently fail. I guess it's related to https://bugs.launchpad.net/qemu/+bug/1841592, but I'm not sure if the comments on this bug are still valid/up-to-date.

Matheus K. Ferst
Instituto de Pesquisas ELDORADO <http://www.eldorado.org.br/>
Analista de Software
Aviso Legal - Disclaimer <https://www.eldorado.org.br/disclaimer.html>

Attachment: float_convs.ref
Description: Text document

Attachment: float_madds.ref
Description: Text document

reply via email to

[Prev in Thread] Current Thread [Next in Thread]