This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Large parallel glibc builds.
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: Tulio Magno Quites Machado Filho <tuliom at linux dot vnet dot ibm dot com>
- Cc: Carlos Eduardo Seo <cseo at linux dot vnet dot ibm dot com>, GNU C Library <libc-alpha at sourceware dot org>
- Date: Wed, 23 Dec 2015 09:12:33 -0500
- Subject: Re: Large parallel glibc builds.
- Authentication-results: sourceware.org; auth=none
- References: <5679A3BE dot 2060303 at redhat dot com> <5679AE4E dot 3020907 at linux dot vnet dot ibm dot com> <5679AEFC dot 9000102 at redhat dot com> <871tae2cqa dot fsf at totoro dot br dot ibm dot com>
On 12/22/2015 08:08 PM, Tulio Magno Quites Machado Filho wrote:
> "Carlos O'Donell" <carlos@redhat.com> writes:
>
>> On 12/22/2015 03:10 PM, Carlos Eduardo Seo wrote:
>>>
>>>
>>> On 12/22/15 5:25 PM, Carlos O'Donell wrote:
>>>> Community,
>>>>
>>>> I am consistently doing parallel (-j56) builds for glibc on Intel
>>>> Xeon CPU E5-2697 v3 @ 2.60GHz (2 socket x 14 cores x 2HT).
>>>>
>>>> I have not seen any failures or artifacts that look like hazards or
>>>> races in our makefiles.
>>>>
>>>> Is anyone else using larger boxes for builds and still seeing
>>>> problems?
>>>>
>>>> Cheers, Carlos.
>>>>
>>>
>>> No, I've been doing -j80 builds on POWER8 just fine. I could run a
>>> test on larger systems, if you wish.
>>
>> If you have a chance it would be good to collect a little more data.
>>
>> We will also get some larger boxes to double check.
>
> We also have a POWER7 build slave running with with -j128 since Jan/2013
> without problems.
> From time to time, a testcase fails. In some cases, due to a timeout.
> For the record, in this particular server, we're still using PARALLELMFLAGS.
> But we don't use it on other servers.
>
> What kind of data are you looking for?
Exactly that. We've had reports from developers in the past that they
have seen parallel build failures that looked like dependency problems
in the makefiles. I'm glad to hear that this anecdotal evidence doesn't
apply anymore to our master builds, or perhaps is so rare, or related
to older Make versions.
Cheers,
Carlos.