Date: Fri, 18 Dec 1998 16:45:41 +0100 From: "José Mª Alcaide" <jose@we.lc.ehu.es> To: freebsd-current@FreeBSD.ORG Subject: SMP benchmarks Message-ID: <367A78A5.3FBF4CBA@we.lc.ehu.es>
next in thread | raw e-mail | index | archive | help
After reading a recent message from Jordan (I lost its reference)
about the "makeworldbench" test results, I was a bit worried: is one
of the two CPUs of my new computer being wasted by the FreeBSD
kernel? Uh-oh! Then I did a little test running one of the BYTE
benchmarks (arithoh). I did three runs with one, two and three
simultaneous processes. These are the results:
----------------------------------------------------
TEST 1 - ONE process ALONE:
TEST|Arithmetic Test (type = arithoh)
FLAVOR|BSD
318250006 loops
10.00 real 9.99 user 0.00 sys
318090368 loops
10.00 real 9.99 user 0.00 sys
318006936 loops
10.00 real 9.98 user 0.00 sys
318271812 loops
10.00 real 10.00 user 0.00 sys
318086538 loops
10.00 real 9.99 user 0.00 sys
318002117 loops
10.00 real 9.99 user 0.00 sys
----------------------------------------------------
TEST 2 - TWO simultaneous processes:
===>PROCESS 1:
TEST|Arithmetic Test (type = arithoh)
FLAVOR|BSD
317440824 loops
10.00 real 9.98 user 0.00 sys
317409417 loops
10.00 real 9.98 user 0.00 sys
317367675 loops
10.00 real 9.99 user 0.00 sys
317256654 loops
10.00 real 9.98 user 0.00 sys
317141118 loops
10.00 real 9.97 user 0.00 sys
317716271 loops
10.00 real 10.00 user 0.00 sys
===>PROCESS 2:
TEST|Arithmetic Test (type = arithoh)
FLAVOR|BSD
317661019 loops
10.00 real 9.97 user 0.00 sys
317213738 loops
10.00 real 9.97 user 0.03 sys
316201777 loops
10.00 real 9.95 user 0.00 sys
317520076 loops
10.00 real 9.99 user 0.00 sys
317297652 loops
10.00 real 9.98 user 0.00 sys
317758290 loops
10.00 real 10.00 user 0.00 sys
----------------------------------------------------
TEST 3 - THREE simultaneous processes:
===>PROCESS 1:
TEST|Arithmetic Test (type = arithoh)
FLAVOR|BSD
218048189 loops
10.00 real 6.86 user 0.00 sys
211899735 loops
10.00 real 6.66 user 0.01 sys
212672592 loops
10.03 real 6.69 user 0.00 sys
218775511 loops
10.03 real 6.88 user 0.00 sys
212323294 loops
10.01 real 6.69 user 0.00 sys
211117933 loops
10.00 real 6.64 user 0.00 sys
===>PROCESS 2:
TEST|Arithmetic Test (type = arithoh)
FLAVOR|BSD
217233937 loops
10.00 real 6.84 user 0.00 sys
218657533 loops
10.02 real 6.88 user 0.00 sys
218244791 loops
10.00 real 6.86 user 0.00 sys
217716665 loops
10.03 real 6.85 user 0.00 sys
214327742 loops
10.03 real 6.74 user 0.00 sys
213706676 loops
10.03 real 6.72 user 0.00 sys
===>PROCESS 3:
TEST|Arithmetic Test (type = arithoh)
FLAVOR|BSD
220732030 loops
10.00 real 6.95 user 0.00 sys
226138300 loops
10.00 real 7.12 user 0.00 sys
214511860 loops
10.00 real 6.74 user 0.00 sys
209003767 loops
10.00 real 6.57 user 0.00 sys
217434544 loops
10.00 real 6.83 user 0.00 sys
218595421 loops
10.00 real 6.87 user 0.01 sys
----------------------------------------------------
Obviously this is only a number-crunching test. It does not measure
kernel (syscall) SMP performance. However, it is significant. Observe
that the results of tests 1 and 2 (one process alone and two
simultaneous processes) are essentially identical. Even the "user
time" is almost equal to the "real time" for both tests. But test 3
shows important differences: first, the number of loops is about 32%
less than the two process test; and the "user time" is about 68% of
the time dedicated to each process of test 2.
In summary, at least making pure arithmetic computing, SMP does
its work.
(BTW, I'm also getting the "Freeing (NOT implemented) redirected ISA
irq" messages at startup. The machine is a Supermicro P6DBU, two
PII-450, 256 MB RAM).
-- JMA
-----------------------------------------------------------------------
José Mª Alcaide | mailto:jose@we.lc.ehu.es
Universidad del País Vasco | http://www.we.lc.ehu.es/~jose
Dpto. de Electricidad y Electrónica |
Facultad de Ciencias - Campus de Lejona | Tel.: +34-946012479
48940 Lejona (Vizcaya) - SPAIN | Fax: +34-944858139
-----------------------------------------------------------------------
"Go ahead... make my day." - H. Callahan
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?367A78A5.3FBF4CBA>