Re: [Maxima-discuss] Maxima works with gcl 2.7.1

[David S. <d.s...@go...>]

Yes, I use 1 GiB hugepages on this machine. Commonly supported hugepage
sizes are 2 MiB and 1 GiB. It's not unusual to use the 1 GiB option,
especially with virtualization.

MemTotal:       131817344 kB
MemFree:        29411484 kB
MemAvailable:   29138728 kB
Buffers:           18628 kB
Cached:          1021516 kB
SwapCached:         6620 kB
Active:           224524 kB
Inactive:          63028 kB
Active(anon):      17732 kB
Inactive(anon):     3124 kB
Active(file):     206792 kB
Inactive(file):    59904 kB
Unevictable:      791724 kB
Mlocked:           27744 kB
SwapTotal:      16759804 kB
SwapFree:       16563004 kB
Zswap:                 0 kB
Zswapped:              0 kB
Dirty:                 0 kB
Writeback:             0 kB
AnonPages:         37684 kB
Mapped:            64448 kB
Shmem:              3108 kB
KReclaimable:      56948 kB
Slab:             206984 kB
SReclaimable:      56948 kB
SUnreclaim:       150036 kB
KernelStack:        9464 kB
PageTables:         4216 kB
SecPageTables:         0 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    32336828 kB
Committed_AS:     549828 kB
VmallocTotal:   34359738367 kB
VmallocUsed:      250676 kB
VmallocChunk:          0 kB
Percpu:            31744 kB
HardwareCorrupted:     0 kB
AnonHugePages:         0 kB
ShmemHugePages:        0 kB
ShmemPmdMapped:        0 kB
FileHugePages:         0 kB
FilePmdMapped:         0 kB
Unaccepted:            0 kB
HugePages_Total:      96
HugePages_Free:       96
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:    1048576 kB
Hugetlb:        100663296 kB
DirectMap4k:      278288 kB
DirectMap2M:     4882432 kB
DirectMap1G:    128974848 kB

Am Do., 24. Apr. 2025 um 21:31 Uhr schrieb Camm Maguire <
ca...@ma...>:

> Following up to myself, I now think this is the problem, maybe in both
> cases:
>
> page size:            1073741824 bytes
>
> GCL will support Linux huge pages with considerable performance gain,
> but this value when available should be around 2Mb.
>
> I suspect a configure script failure:
>
> j=""
> if test -e /proc/meminfo ; then
>    j=`awk '/^Hugepagesize:/ {if ($3!="kB") next;j=1;for (i=0;i<20 &&
> j!=$2;i++) j=j*2;if (j==$2) printf("%d\n",i+10)}'  /proc/meminfo`
> fi
>
> Could you please send me `cat /proc/meminfo`.
>
> Take care,
>
>
> David Scherfgen via Maxima-discuss
> <max...@li...> writes:
>
> > Thanks, any help is welcome!
> > Camm, I think I already sent you the config and make logs from trying to
> build the 2.7.1 tag, it was a few weeks ago before you went on a vacation
> or something.
> > But here it is again, plus new logs from trying to build the Git master:
> >
> > Version_2_7_1 tag: First error "ERROR: File /tmp/gazonk_625863_0.o has
> been compiled for a restricted address space"
> > config.log: https://pastebin.com/V4jCCAsH
> > make.log: https://pastebin.com/1sY8UPMF
> >
> > master branch: First error "The assertion sigaltstack(&estack, 0)>=0 on
> line 1285 of o/alloc.c in function gcl_init_alloc failed: Operation not
> permitted"
> > config.log: https://pastebin.com/8tgqpDfH
> > make.log: https://pastebin.com/xJheBrEa
> >
> > I hope this helps.
> >
> > Best regards
> > David
> >
> > Am Do., 24. Apr. 2025 um 18:19 Uhr schrieb Camm Maguire <
> ca...@ma...>:
> >
> >  Greetings!
> >
> >  David Scherfgen via Maxima-discuss
> >  <max...@li...> writes:
> >
> >  > That's great news!
> >  > Increasing numerical thresholds should not be a problem, the
> differences are probably due to a different order of floating point
> operations. By the way,
> >  some
> >  > other Lisps also fail a few numerical tests, IIRC.
> >  >
> >  > I still haven't been able to build GCL on my Ubuntu 22.04.5 system.
> The 2.7.1 source distribution (after patching as described on the website)
> leads to one
> >  error,
> >  > trying to build from the current Git master leads to another error.
> Raymond, could you share your GCL build procedure?
> >  >
> >
> >  I am more than happy to help with both of these if you could send me the
> >  failed logs.
> >
> >  Take care,
> >  --
> >  Camm Maguire
> ca...@ma...
> >
> ==========================================================================
> >  "The earth is but one country, and mankind its citizens."  --
> Baha'u'llah
> >
> > _______________________________________________
> > Maxima-discuss mailing list
> > Max...@li...
> > https://lists.sourceforge.net/lists/listinfo/maxima-discuss
> >
>
> --
> Camm Maguire                                        ca...@ma...
> ==========================================================================
> "The earth is but one country, and mankind its citizens."  --  Baha'u'llah
>