Mathematica 5.2 (Solaris 10, Sparc) excessive CPU use

14 views
Skip to first unread message

Rainer Beushausen

unread,
Mar 10, 2006, 1:03:57 AM3/10/06
to
Having probs running Mathematica 5.2 (Sparc) on Solaris 10.
From within the GUI, MathKernel starts using most cpu
time on one CPU (U60 2x450) right after giving the
first result.

The command-line app 'math' which is actually the kernel
itself runs fine as long there is no other GUI
window to communicate with (like plotting output).

The same binary runs without any probs on a Solaris 9
box.

Anyone else? Any workaround?

Rainer

Dave (from the UK)

unread,
Mar 10, 2006, 3:30:57 AM3/10/06
to

Yes, I have found this.

In fact, I posted it sometime back on sci.math.symbolic, where I have
copied your post to as well.

http://groups.google.co.uk/group/sci.math.symbolic/browse_thread/thread/215514ac8121ddf1/1699406e69834c82?lnk=st&q=mathematica+cpu+solaris&rnum=3&hl=en#1699406e69834c82

I just took 3 of the processors off line so there was only one active -
I did not physically remove the other 3. The same problem occurs.

Are you running the first general release of Solaris 10 or Solaris 10
update 1? I can only recall noticing it since update 1 of Solaris was
installed, but can't be sure of this.

I've also noticed something very odd with a Mathematica benchmark

http://smc.vnet.net/timings50.html

Although my results are not shown on that page, I have had some
discussions with the author about them. The results from a quad
processor Ultra 80 (4 x 450) give it a considerably poorer score than a
single processor Ultra 60 (360 MHz). That happens using the command line
'math' without the GUI. The 4 x 450 MHz Sun finishes the benchmark in
less overall time (as measured by a clock) than the single 360 MHz
processors Sun, but eats up loads of CPU time in doing it, so ends up
with a poorer score.

Any workaround?

Not found one.

> Rainer
>

--
Dave K

Minefield Consultant and Solitaire Expert (MCSE).

Please note my email address changes periodically to avoid spam.
It is always of the form: month-year@domain. Hitting reply will work
for a couple of months only. Later set it manually.

ba...@smaalders.net

unread,
Mar 10, 2006, 4:05:02 AM3/10/06
to
What does truss and/or pstack report?

- Bart

Rainer Beushausen

unread,
Mar 10, 2006, 4:16:55 AM3/10/06
to
Dave (from the UK) wrote:

> Are you running the first general release of Solaris 10 or Solaris 10
> update 1? I can only recall noticing it since update 1 of Solaris was
> installed, but can't be sure of this.

It is the first release patched to recent. I am going to install another
box right now so I will check the first release unpatched.

Also will do some dtracing.

Rainer


Rainer Beushausen

unread,
Mar 10, 2006, 4:24:18 AM3/10/06
to
ba...@smaalders.net wrote:
> What does truss and/or pstack report?
>
> - Bart
>
first glance, trussing MathKernel:


/3: pollsys(0xFFFFFFFF79875450, 1, 0xFFFFFFFF798F5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79875450, 1, 0xFFFFFFFF798F5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79875450, 1, 0xFFFFFFFF798F5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79875450, 1, 0xFFFFFFFF798F5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79875450, 1, 0xFFFFFFFF798F5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79875450, 1, 0xFFFFFFFF798F5510, 0x00000000) = 0

and so on...

Rainer

Dave (from the UK)

unread,
Mar 10, 2006, 4:52:06 AM3/10/06
to

Almost the same here, but not quite, although perhaps if you showed
a bit more output, it might be the same

/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/1: lwp_park(0xFFFFFFFF7FFFE020, 0) Err#62 ETIME
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0
/3: pollsys(0xFFFFFFFF79A75450, 1, 0xFFFFFFFF79AF5510, 0x00000000) = 0

KJ

unread,
Mar 10, 2006, 8:06:04 AM3/10/06
to
Love that sig Dave!

Anyway, just thought I'd let you gents know I've got access to a
Blade100 w/ Solaris 10 update 1 (500MHz Sparc IIe & fully patched) with
Mathematica 5.2 on it in case you'd like another data point with a
similar configuration for troubleshooting.

Honestly I haven't noticed this much at all since I rarely use the
Mathematica GUI but it certainly would be nice to have resolved.

Dave (from the UK)

unread,
Mar 10, 2006, 10:02:35 AM3/10/06
to
KJ wrote:
> Love that sig Dave!

I'm glad you do.

> Anyway, just thought I'd let you gents know I've got access to a
> Blade100 w/ Solaris 10 update 1 (500MHz Sparc IIe & fully patched) with
> Mathematica 5.2 on it in case you'd like another data point with a
> similar configuration for troubleshooting.

If you can test it, it would be useful. I suspect the Solaris kernel for
the Blade will be different from the Ultra 80. The OP had a U60, which
is quite similar to this machine (I know a boot disk from a U60
will boot a U80 and visa-versa.)

I've just checked Mathematica 4.0, 4.2, 5.1 and 5.2 on this machine
(Sun Ultra 80, 4 x 450 MHz, Solaris 10 update 1, Solaris
kernel 118822-26)

Mathematica 5.2 - exhibits the problem of excessive CPU usage
Mathematica 5.1 - exhibits the problem of excessive CPU usage
Mathematica 4.2 - No problem found.
Mathematica 4.0 - No problem found.

But I'm pretty sure this was not apparent on older releases of
Solaris.

KJ

unread,
Mar 10, 2006, 11:13:02 AM3/10/06
to
Dave (from the UK) wrote:
> If you can test it, it would be useful. I suspect the Solaris kernel for
> the Blade will be different from the Ultra 80. The OP had a U60, which
> is quite similar to this machine (I know a boot disk from a U60
> will boot a U80 and visa-versa.)
>
> I've just checked Mathematica 4.0, 4.2, 5.1 and 5.2 on this machine
> (Sun Ultra 80, 4 x 450 MHz, Solaris 10 update 1, Solaris
> kernel 118822-26)
>
> Mathematica 5.2 - exhibits the problem of excessive CPU usage
> Mathematica 5.1 - exhibits the problem of excessive CPU usage
> Mathematica 4.2 - No problem found.
> Mathematica 4.0 - No problem found.
>
> But I'm pretty sure this was not apparent on older releases of
> Solaris.
>
I'll have local login to that machine this evening and report back then
with performance timings from Karl's MMA 5.0 as well.

Rainer Beushausen

unread,
Mar 10, 2006, 6:01:56 PM3/10/06
to

I am running it now on a Blade 1000 (2x900), Solaris 10 first release
fully patched. Runs without any probs. No excessive CPU
usage.

Still no solution for U60 and friends.

Rainer

Wes Williams

unread,
Mar 10, 2006, 6:32:05 PM3/10/06
to
Dave (from the UK) wrote:
> KJ wrote:
>
>> Love that sig Dave!
>
>
> I'm glad you do.
>
>> Anyway, just thought I'd let you gents know I've got access to a
>> Blade100 w/ Solaris 10 update 1 (500MHz Sparc IIe & fully patched)
>> with Mathematica 5.2 on it in case you'd like another data point with
>> a similar configuration for troubleshooting.
>
>
> If you can test it, it would be useful. I suspect the Solaris kernel for
> the Blade will be different from the Ultra 80. The OP had a U60, which
> is quite similar to this machine (I know a boot disk from a U60
> will boot a U80 and visa-versa.)
>
> I've just checked Mathematica 4.0, 4.2, 5.1 and 5.2 on this machine
> (Sun Ultra 80, 4 x 450 MHz, Solaris 10 update 1, Solaris
> kernel 118822-26)
>
> Mathematica 5.2 - exhibits the problem of excessive CPU usage
> Mathematica 5.1 - exhibits the problem of excessive CPU usage
> Mathematica 4.2 - No problem found.
> Mathematica 4.0 - No problem found.
>
> But I'm pretty sure this was not apparent on older releases of
> Solaris.
>

Blade100 (500MHz UltraSPARC IIe & 512Mb RAM) Solaris 10 update 1
Fully patched - just installed 121563-02 before testing.

Problem with runaway CPU does exist with this setup.

MMA 5.0 Results:
Version = 5.2 for Sun Solaris (UltraSPARC) (June 21, 2005)
Times = {22.959, 20.1527, 26.4417, 46.6591, 38.909,
26.8787, 57.8305, 148.941, 195.881, 116.129, 206.917, 21.1517,
73.0785, 57.6422, 73.9492}
Total Kernel Time = 1133.52, Total Time = 1147.
Benchmark = 0.42459


PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP

11471 wesw 209M 154M run 59 0 0:27:37 83% MathKernel/3
11755 wesw 116M 49M sleep 49 0 0:01:01 7.8% mozilla-bin/3
451 wesw 22M 18M sleep 59 0 0:00:49 5.0% Xsun/1
1451 wesw 13M 9008K sleep 49 0 0:00:14 2.2% Mathematica/2
11760 wesw 4888K 4632K cpu0 59 0 0:00:00 0.4% prstat/1
857 wesw 9784K 6680K sleep 59 0 0:00:04 0.1% dtwm/5
929 wesw 7928K 4472K sleep 59 0 0:00:03 0.1% sdtperfmeter/1
11630 wesw 7816K 4736K sleep 59 0 0:00:00 0.1% dtterm/1
956 root 2152K 1272K sleep 59 0 0:00:02 0.1% rpc.rstatd/1
925 wesw 10M 6112K sleep 49 0 0:00:00 0.0% dtfile/1
92 root 4176K 2560K sleep 59 0 0:00:00 0.0% nscd/26
11632 wesw 2960K 2056K sleep 59 0 0:00:00 0.0% bash/1
726 root 7216K 3520K sleep 59 0 0:00:00 0.0% httpd/1
371 root 10M 7328K run 59 0 0:00:02 0.0% cctransport/8
788 root 3696K 2680K sleep 59 0 0:00:00 0.0% nscd/24
86 root 3176K 1416K sleep 59 0 0:00:00 0.0% picld/7
1133 root 7504K 2288K sleep 59 0 0:00:00 0.0% sendmail/1
1123 root 7504K 2272K sleep 59 0 0:00:00 0.0% sendmail/1
770 root 3696K 2648K sleep 59 0 0:00:00 0.0% nscd/24
546 root 9352K 6312K sleep 59 0 0:00:00 0.0% snmpd/1
300 root 7504K 1504K sleep 59 0 0:00:00 0.0% sendmail/1
Total: 134 processes, 410 lwps, load averages: 1.40, 1.41, 1.25

Dave (from the UK)

unread,
Mar 10, 2006, 6:35:32 PM3/10/06
to
Rainer Beushausen wrote:

> I am running it now on a Blade 1000 (2x900), Solaris 10 first release
> fully patched. Runs without any probs. No excessive CPU
> usage.
>
> Still no solution for U60 and friends.

That narrows it down a *lot* if the program only misbehaves on those
machines. I guess it would be anything that uses the same kernel.

I was surprised there was not some confirmation when I posted this on
some maths newsgroups a few weeks ago, but if it only affects these
machines, it probably explains it.

I can try it on a Netra T1 at some point, but I suspect it will be OK on
that too, as it is a quite different machine.

Any thoughts from those with knowledge about the Solaris kernel ?


--
Dave K MCSE.

MCSE = Minefield Consultant and Solitaire Expert.

Dave (from the UK)

unread,
Mar 10, 2006, 6:39:00 PM3/10/06
to
Dave (from the UK) wrote:

>> Still no solution for U60 and friends.

If this effects the U60 and U80, but not the Blade 100 and Blade 1000,
does anyone know what other machines it might effect? I assume anything
with the same kernel would be potential targets.

sparrow /export/home/drkirkby % uname -a
SunOS sparrow 5.10 Generic_118822-26 sun4u sparc SUNW,Ultra-80


sparrow /export/home/drkirkby % cat /etc/release
Solaris 10 1/06 s10s_u1wos_19a SPARC
Copyright 2005 Sun Microsystems, Inc. All Rights Reserved.
Use is subject to license terms.
Assembled 07 December 2005

KJ

unread,
Mar 10, 2006, 7:37:22 PM3/10/06
to
Dave (from the UK) wrote:
> Dave (from the UK) wrote:
>
>>> Still no solution for U60 and friends.
>
> If this effects the U60 and U80, but not the Blade 100 and Blade 1000,
> does anyone know what other machines it might effect? I assume anything
> with the same kernel would be potential targets.
>
> sparrow /export/home/drkirkby % uname -a
> SunOS sparrow 5.10 Generic_118822-26 sun4u sparc SUNW,Ultra-80
>
>
> sparrow /export/home/drkirkby % cat /etc/release
> Solaris 10 1/06 s10s_u1wos_19a SPARC
> Copyright 2005 Sun Microsystems, Inc. All Rights Reserved.
> Use is subject to license terms.
> Assembled 07 December 2005
>
>
Sorry to make it easy, but the problem does exist on the Blade 100. I
(Wes Williams) also forgot to post the uname earlier, but here is...

Another affected machine:

-bash-3.00# uname -a
SunOS usmc 5.10 Generic_118833-02 sun4u sparc SUNW,Sun-Blade100
-bash-3.00#

-bash-3.00# cat /etc/release


Solaris 10 1/06 s10s_u1wos_19a SPARC
Copyright 2005 Sun Microsystems, Inc. All Rights Reserved.
Use is subject to license terms.
Assembled 07 December 2005

-bash-3.00#

Dave (from the UK)

unread,
Mar 10, 2006, 9:52:30 PM3/10/06
to
Rainer Beushausen wrote:

> I am running it now on a Blade 1000 (2x900), Solaris 10 first release
> fully patched. Runs without any probs. No excessive CPU
> usage.
>
> Still no solution for U60 and friends.
>
> Rainer
>

Rainer,

What is the exact release and kernel version on the Blade 1000 where
Mathematica runs without problems? Can you cat /etc/release and give us
'uname -a'?

Given that

1) All affected Sun workstations (Ultra 60, Ultra 80 and Blade 100) seem
to be running Solaris 10 update 1.

2) The only machine not affected (Blade 1000) is running the *first*
release of Solaris 10.

makes me think this may be a bug triggered in Solaris 10 update 1 (i.e.
the second release of Solaris 10), which came out at the very end of
December 2005.

I noticed the problem of excessive CPU usage first on the 19th Jan 2006

http://groups.google.co.uk/group/sci.math.symbolic/browse_frm/thread/215514ac8121ddf1/1699406e69834c82?lnk=st&q=cpu+mathematica+ultra+80&rnum=3&hl=en#1699406e69834c82

so was within 3 weeks of installing Solaris 10 update 1.

I am pretty sure I would have noticed this problem earlier, had it
existed. The fact I can see it in Mathematica 5.1, makes me think this
might be something introduced by Solaris in the last few months.

Of course, I am not saying it is Sun's fault - it could be bad
programming of Mathematica, but I think the problem has only surfaced
recently.

Even Mathematica 5.1 is affected, so I'm pretty sure this was not the
case when Mathematica 5.2 was released.


--
Dave K MCSE.

MCSE = Minefield Consultant and Solitaire Expert.

Please note my email address changes periodically to avoid spam.

Rainer Beushausen

unread,
Mar 11, 2006, 4:43:21 AM3/11/06
to
Dave (from the UK) wrote:

> Rainer,
>
> What is the exact release and kernel version on the Blade 1000 where
> Mathematica runs without problems? Can you cat /etc/release and give us
> 'uname -a'?
>

I am running Mathematica 5.2 on 3 systems:

1) U60 Solaris 10 3/05. System is *affected*.
cat /etc/release
Solaris 10 3/05 s10_74L2a SPARC


Copyright 2005 Sun Microsystems, Inc. All Rights Reserved.
Use is subject to license terms.

Assembled 22 January 2005
uname -a
SunOS faramir 5.10 Generic_118822-30 sun4u sparc SUNW,Ultra-60

prtdiag
System Configuration: Sun Microsystems sun4u Sun Ultra 60 UPA/PCI (2 X
UltraSPARC-II 450MHz)

2) Blade-1000 Solaris 10 3/05. System is *not* affected
cat /etc/release
Solaris 10 3/05 s10_74L2a SPARC


Copyright 2005 Sun Microsystems, Inc. All Rights Reserved.
Use is subject to license terms.

Assembled 22 January 2005

uname -a
SunOS elrond 5.10 Generic_118833-02 sun4u sparc SUNW,Sun-Blade-1000

prtdiag
System Configuration: Sun Microsystems sun4u SUNW,Sun-Blade-1000 (2 X
UltraSPARC-III)

3) Ultra-2 Solaris 9 12/03. System is *not* affected
cat /etc/release
Solaris 9 12/03 s9s_u5wos_08b SPARC
Copyright 2003 Sun Microsystems, Inc. All Rights Reserved.


Use is subject to license terms.

Assembled 21 November 2003
uname -a
SunOS gandalf 5.9 Generic_118558-09 sun4u sparc SUNW,Ultra-2

prtdiag
System Configuration: Sun Microsystems sun4u Sun Ultra 2 UPA/SBus (2 X
UltraSPARC-II 400MHz)

All boxes maintain a recent patch level via smpatch.

My Blade-1000 is UltraSPARC-III, the affected systems known so far
are UltraSPARC-II and UltraSPARC IIe (Blade100), respectively. All
affected systems are Sol 10, recent patch level.

Rainer


Reply all
Reply to author
Forward
0 new messages