>
> I ran the speed program, but I do not see how to "choose different
> assembly
> files". Do I have to build different speed programs with different
> assembly
> files linked in?
>
The perl script tune/many.pl is pretty useful for doing this, I can't
remember the exact options but opening
Hello Bill,
Yes, I would be happy to help. I'll keep an eye out for his email.
-David C.
On 2/21/2017 3:53 AM, 'Bill Hart' via mpir-devel wrote:
I have just been contacted off list by a very experienced assembly expert who
would like to help develop Broadwell assembly optimisations, guided by
I added model 71 to cpuid.c as a Broadwell. The other models I have left
out for now, as I don't find sufficient evidence for those identifications
just yet. Perhaps Intel has reserved them for future chips. When people
start reporting to us that they have chips in these families, we'll take a
look
I've opened [1] for the haswell bmi2 issue. I doubt it will be dealt with
in this release. As far as I can tell, such machines are vanishingly rare.
Bill.
[1] https://github.com/wbhart/mpir/issues/209
On 21 February 2017 at 11:03, Bill Hart wrote:
> Whoops, looks like I was wrong about us corr
Whoops, looks like I was wrong about us correctly handling Haswell without
BMI2. We handle Skylake correctly, but Haswell we don't check for BMI2.
It looks like the Haswell directory has been divided into haswell and
haswell/avx, which is probably intended for this distinction, but I think
we deci
I have just been contacted off list by a very experienced assembly expert
who would like to help develop Broadwell assembly optimisations, guided by
your experiments (as far as I know, he doesn't actually have a Broadwell
machine).
If you are ok with it, I could share your email with him so that h
On 21 February 2017 at 00:35, wrote:
> Hello Bill,
>
> I can report that both of my systems are now correctly detected. Thank
> you for the fix!
>
> I also noticed that there were a couple of other missing processor models
> when compared with GMP mpn/x86_64/fat/fat.c: (the * star values are mis
Hello Bill,
I can report that both of my systems are now correctly detected. Thank you for
the fix!
I also noticed that there were a couple of other missing processor models when
compared with GMP mpn/x86_64/fat/fat.c: (the * star values are missing in MPIR)
Broadwell:
0x3d=61
0x47=71 *
0x4
I have issued an alpha3 which supports your particular Broadwell chip,
which was a Family we weren't aware of. Can you verify that it is now
detected correctly.
As for optimising for Broadwell, if you are interested, we'd be happy to
have the help. I've added details to the relevant ticket [1] to
The only thing we need is the output of cat /proc/cpuinfo under Linux.
Broadwell is supported in this release, we just haven't optimised for it.
That would be 3-6 months of full time work, so there's no chance of it
happening in this release.
Bill.
On 18 February 2017 at 20:24, wrote:
> Hello
It looks like you configured your broadwell as a generic x86_64 instead of
as a haswell (as per your patch). This means that it wasn't using anything
other than generic assembly for the build, and hence the tuning values will
be way off.
Perhaps you can force it to build as a haswell and then run
I don't know if you are interested or not, but broadwell probably runs some
of the Skylake assembly code. If you drop it in the haswell directory you
set up for your machine and time the functions with speed, you can probably
figure out if any of the existing Skylake code makes Broadwell run faster
Thanks very much! I'll insert these today. I'll attach the broadwell ones
to a ticket, until someone can sort out the configuration for that system.
There are probably numerous cpuid family/model pairs that correspond to
Broadwell, so we'll have to add these as they become known.
On 14 February 20
Attached are tuning values for nehalem, ivybridge, broadwell and skylake.
(Also cpuinfo for the broadwell one)
Isuru Fernado
On Tue, Feb 14, 2017 at 1:29 PM, 'Bill Hart' via mpir-devel <
mpir-devel@googlegroups.com> wrote:
> Apparently if you have a very recent machine, yasm may fail to build th
Apparently if you have a very recent machine, yasm may fail to build the
assembly files for your architecture. To get around this, install the
latest yasm [1] and use MPIR's --with-system-yasm option.
If your system is recent and detects as core2 or k8 or simply x86_64 or
something else obviously
15 matches
Mail list logo
