[PPL-devel] Testing ppl-0.10pre24 on i686-apple-darwin9

[Roberto Bagnara]

Dominique Dhumieres wrote:
> I have tried to test ppl-0.10pre24 built on i686-apple-darwin9 with gcc
> 4.4.0 from some yesterday trunk.  So far I got 399 "PASS" for 24 "FAIL":
> 
> 15 of them are missing symbols: 
> ...21Grid_Generator_SystemENS_14Grid_GeneratorEEEvRT... (14 times)
> ...17Congruence_SystemENS_10CongruenceEEEvRT...         (once)
> 
> 2 bus errors and 7 "failed tests".

Dear Dominique,

[Sorry for the previous message: I don't know how the "Send" button got pressed.]

Can you please compare with what you obtain with a stable version of GCC?
I guess many of the failures you observe will disappear.

> Also the test 'memory1' has been running for more than four hours without 
> completing. Is this normal? How long is it supposed to run on a 2.1Ghz 
> Core2?

See http://www.cs.unipr.it/pipermail/ppl-devel/2008-September/012324.html
Can you please check what you obtain by defining the environment variable
PPL_NOISY_TESTS before running memory1?  Here is what I obtain:

$ export PPL_NOISY_TESTS=yes
$ ./memory1
Trying dimension 1
Trying dimension 2
Trying dimension 3
Trying dimension 4
Trying dimension 5
Trying dimension 6
Trying dimension 7
Trying dimension 8
out of virtual memory
Trying upper bound 6291456
out of virtual memory
Trying upper bound 12582912
out of virtual memory
Trying upper bound 25165824
out of virtual memory
Trying upper bound 50331648
Probing 37748736
Probing 31457280
Probing 28311552
Probing 26738688
Probing 25952256
Probing 25559040
Probing 25362432
Probing 25264128
Probing 25214976
Probing 25190400
Probing 25178112
Probing 25171968
Probing 25168896
Probing 25167360
Probing 25166592
Estimated memory for dimension 8: 25166208 bytes
$

> Note that the present gcc 4.4.0 trunk is rather buggy due to several large 
> merges which can explain some of the failures. If this is the case I can 
> try a build with the 4.3.2 release.

Yes, this is what I would recommend.
All the best,

    Roberto

-- 
Prof. Roberto Bagnara
Computer Science Group
Department of Mathematics, University of Parma, Italy
http://www.cs.unipr.it/~bagnara/
mailto:bagnara at cs.unipr.it