is there any progress


Advanced search

Message boards : Macintosh : is there any progress

Sort
Author Message
jc.
Volunteer tester
Avatar

Joined: Nov 1 06
Posts: 5
ID: 195
Credit: 499,917
RAC: 0
Message 1664 - Posted 30 Nov 2006 12:50:41 UTC

is there anything happening on the mac front
Andre told me i shouldn't expect much credits
but all the zeros are getting a bit boring
it would be a nice sinterklaas present to get some cred.

Jan willem

Profile Andre Kerstens
Forum moderator
Project tester
Volunteer tester
Avatar

Joined: Sep 11 06
Posts: 749
ID: 1
Credit: 15,199
RAC: 0
Message 1675 - Posted 1 Dec 2006 17:10:36 UTC - in response to Message ID 1664 .

Hi Jan Willem,

Thanks for your post. Unfortunately (and to our despair) we seem to have a BIG problem with the Charmm code on Mac OS X. For example in this workunit, every single result we received is different and the validator couldn't get a quorum together. To be honest with you, we are not focusing on Mac right now, because of the fact that we don't have enough people available to look into all the problems we are having right now (the project is mostly a group of part-time students and me officially working on it for 2 days a week). So to conclude, if you want, please suspend the D@H project on your machine until we have things on the Macs sorted out. I will also write a message on the front page news about this.

Thanks for your help and keep an eye on our RSS feed...

Andre

PS En sorry dat we je geen sinterklaas cadeau aan kunnen bieden :-(

is there anything happening on the mac front
Andre told me i shouldn't expect much credits
but all the zeros are getting a bit boring
it would be a nice sinterklaas present to get some cred.

Jan willem


____________
D@H the greatest project in the world... a while from now!
jc.
Volunteer tester
Avatar

Joined: Nov 1 06
Posts: 5
ID: 195
Credit: 499,917
RAC: 0
Message 1678 - Posted 2 Dec 2006 1:49:33 UTC - in response to Message ID 1675 .
Last modified: 2 Dec 2006 1:50:06 UTC

So to conclude, if you want, please suspend the D@H project on your machine until we have things on the Macs sorted out. I will also write a message on the front page news about this.

Thanks for your help and keep an eye on our RSS feed...

Andre

PS En sorry dat we je geen sinterklaas cadeau aan kunnen bieden :-(



Thanks for the answer Andre,
I won't suspend Docking completely,
but i've reduced the effort down to 2.5% of cpu time
just to remind me the project is still there.
and because i'm to lazy to read the news :o)
I'll just wait for positive results to turn up



Jan willem

Ps. I know christmas is to soon to hope for.
i'm placing my bets on easter :o)

Dotsch
Volunteer tester
Avatar

Joined: Sep 13 06
Posts: 49
ID: 75
Credit: 57,728
RAC: 0
Message 1684 - Posted 2 Dec 2006 18:52:13 UTC - in response to Message ID 1675 .

Andre, I think it is a little bit strange, that a same type of CPUs get out of the range, that the results are not valid.

But, how did your validator work ?
SETI has a validator which comparing the results and accepting a difference by imho 1 %. If you doing a different optimisation and using different compiler flags, there could be a difference...
SIMAP validates with diff, this caused a little problems, because the RISCs (HP PARISC, SPARC, IA64, Alpha) computed otherwise than the CISCs (x86). But all CISC (and RISC) results was OK, if there compared again the same class of CPU.

I am not shure, if it is OK for your results to have some/a difference...

Have you tried to reduce or switch off the compiler optimisations ? - It could also be, that there are some additional Flags are needed, to get them compute right. - For example, I had problems to get the old SETI 4.x computing right at NetBSD. The results has differed heavly between the reference result and also was different at the different runs. It worked, if i compiled with the CFLAGS "-ffloat-store -funsafe-math-optimizations".

Profile Andre Kerstens
Forum moderator
Project tester
Volunteer tester
Avatar

Joined: Sep 11 06
Posts: 749
ID: 1
Credit: 15,199
RAC: 0
Message 1691 - Posted 3 Dec 2006 1:44:25 UTC - in response to Message ID 1684 .

Hi Dotsch,

We are using the Intel Fortran compiler

Our scientists (the people that actually analyze the results) want to see three exact the same results so that they can actually trust what they are working with. Unless we can convince them that the system is so good that we can do with 2 similar results, this will most likely stay for a while. This means we can not use a fuzzy comparison like seti does.

Regarding the Mac results being so different, we don't know yet exactly why this is happening, but there will be a meeting on Monday to discuss this very issue. Since the Mac binary of Charmm is the only one that is not statically compiled (Apple doesn't support that and makes your life hard if you try), we think it might have to do with different libraries on the different machines (not very likely), or the fact that the Intel processors are not of the same type and therefore give different results. We hope we'll find out soon. Your compiler options seem to come from the gnu compiler; we use the Intel Fortran compiler for our app and have tried a gazillion compiler options and option combinations already, but none of these helped solve the problem up to now.

Thanks
Andre

Andre, I think it is a little bit strange, that a same type of CPUs get out of the range, that the results are not valid.

But, how did your validator work ?
SETI has a validator which comparing the results and accepting a difference by imho 1 %. If you doing a different optimisation and using different compiler flags, there could be a difference...
SIMAP validates with diff, this caused a little problems, because the RISCs (HP PARISC, SPARC, IA64, Alpha) computed otherwise than the CISCs (x86). But all CISC (and RISC) results was OK, if there compared again the same class of CPU.

I am not shure, if it is OK for your results to have some/a difference...

Have you tried to reduce or switch off the compiler optimisations ? - It could also be, that there are some additional Flags are needed, to get them compute right. - For example, I had problems to get the old SETI 4.x computing right at NetBSD. The results has differed heavly between the reference result and also was different at the different runs. It worked, if i compiled with the CFLAGS "-ffloat-store -funsafe-math-optimizations".


____________
D@H the greatest project in the world... a while from now!
Dotsch
Volunteer tester
Avatar

Joined: Sep 13 06
Posts: 49
ID: 75
Credit: 57,728
RAC: 0
Message 1703 - Posted 4 Dec 2006 8:54:21 UTC - in response to Message ID 1691 .

Regarding the Mac results being so different, we don't know yet exactly why this is happening, but there will be a meeting on Monday to discuss this very issue. Since the Mac binary of Charmm is the only one that is not statically compiled (Apple doesn't support that and makes your life hard if you try), we think it might have to do with different libraries on the different machines (not very likely), or the fact that the Intel processors are not of the same type and therefore give different results. We hope we'll find out soon.

Have you compared the results come back from different OS releases against each other ? - Is it posible to configure the HR testwise so that only same CPUs and same OS release gets the same WUs ?
I hope you will support Macs in the future !


Your compiler options seem to come from the gnu compiler; we use the Intel Fortran compiler for our app and have tried a gazillion compiler options and option combinations already, but none of these helped solve the problem up to now.

Sorry, did not know that you use the Intel compiler.
Good, that you have tried the different compiler switches. But bad, that it did not work. So I think the problem is not in the optimisation.

At SIMAP we had some problems with the rounding functions, which produced differences in the results. Could also be, that different OS releases and patchlevels have other math libs with other roundings.
Dotsch
Volunteer tester
Avatar

Joined: Sep 13 06
Posts: 49
ID: 75
Credit: 57,728
RAC: 0
Message 1788 - Posted 16 Dec 2006 14:53:21 UTC

Andre, you have announced at the front page news, that there will be a scripted mailed to the MacOS users. - I have not recived any mail yet.

Memo
Forum moderator
Project developer
Project tester

Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
Message 1795 - Posted 17 Dec 2006 7:16:41 UTC

Expect that email during the next week.

Profile Andre Kerstens
Forum moderator
Project tester
Volunteer tester
Avatar

Joined: Sep 11 06
Posts: 749
ID: 1
Credit: 15,199
RAC: 0
Message 1806 - Posted 18 Dec 2006 0:19:00 UTC

Yep, Memo is working to get the last tests out of the way; it should work out of the box :-)

AK
____________
D@H the greatest project in the world... a while from now!

Message boards : Macintosh : is there any progress

Database Error
: The MySQL server is running with the --read-only option so it cannot execute this statement
array(3) {
  [0]=>
  array(7) {
    ["file"]=>
    string(47) "/boinc/projects/docking/html_v2/inc/db_conn.inc"
    ["line"]=>
    int(97)
    ["function"]=>
    string(8) "do_query"
    ["class"]=>
    string(6) "DbConn"
    ["object"]=>
    object(DbConn)#14 (2) {
      ["db_conn"]=>
      resource(72) of type (mysql link persistent)
      ["db_name"]=>
      string(7) "docking"
    }
    ["type"]=>
    string(2) "->"
    ["args"]=>
    array(1) {
      [0]=>
      &string(51) "update DBNAME.thread set views=views+1 where id=118"
    }
  }
  [1]=>
  array(7) {
    ["file"]=>
    string(48) "/boinc/projects/docking/html_v2/inc/forum_db.inc"
    ["line"]=>
    int(60)
    ["function"]=>
    string(6) "update"
    ["class"]=>
    string(6) "DbConn"
    ["object"]=>
    object(DbConn)#14 (2) {
      ["db_conn"]=>
      resource(72) of type (mysql link persistent)
      ["db_name"]=>
      string(7) "docking"
    }
    ["type"]=>
    string(2) "->"
    ["args"]=>
    array(3) {
      [0]=>
      object(BoincThread)#3 (16) {
        ["id"]=>
        string(3) "118"
        ["forum"]=>
        string(1) "7"
        ["owner"]=>
        string(3) "195"
        ["status"]=>
        string(1) "0"
        ["title"]=>
        string(21) "is there any progress"
        ["timestamp"]=>
        string(10) "1166639007"
        ["views"]=>
        string(4) "1393"
        ["replies"]=>
        string(1) "8"
        ["activity"]=>
        string(23) "6.3590819774287995e-127"
        ["sufferers"]=>
        string(1) "0"
        ["score"]=>
        string(1) "0"
        ["votes"]=>
        string(1) "0"
        ["create_time"]=>
        string(10) "1164891041"
        ["hidden"]=>
        string(1) "0"
        ["sticky"]=>
        string(1) "0"
        ["locked"]=>
        string(1) "0"
      }
      [1]=>
      &string(6) "thread"
      [2]=>
      &string(13) "views=views+1"
    }
  }
  [2]=>
  array(7) {
    ["file"]=>
    string(63) "/boinc/projects/docking/html_v2/user/community/forum/thread.php"
    ["line"]=>
    int(184)
    ["function"]=>
    string(6) "update"
    ["class"]=>
    string(11) "BoincThread"
    ["object"]=>
    object(BoincThread)#3 (16) {
      ["id"]=>
      string(3) "118"
      ["forum"]=>
      string(1) "7"
      ["owner"]=>
      string(3) "195"
      ["status"]=>
      string(1) "0"
      ["title"]=>
      string(21) "is there any progress"
      ["timestamp"]=>
      string(10) "1166639007"
      ["views"]=>
      string(4) "1393"
      ["replies"]=>
      string(1) "8"
      ["activity"]=>
      string(23) "6.3590819774287995e-127"
      ["sufferers"]=>
      string(1) "0"
      ["score"]=>
      string(1) "0"
      ["votes"]=>
      string(1) "0"
      ["create_time"]=>
      string(10) "1164891041"
      ["hidden"]=>
      string(1) "0"
      ["sticky"]=>
      string(1) "0"
      ["locked"]=>
      string(1) "0"
    }
    ["type"]=>
    string(2) "->"
    ["args"]=>
    array(1) {
      [0]=>
      &string(13) "views=views+1"
    }
  }
}
query: update docking.thread set views=views+1 where id=118