Hopefully, just an OOPS!! moment ....
9 Jan 2008 0:18:19 UTC
Topic 193425
(moderation:
Has anyone noticed the shenanigans going on at the top of the Top Computers list? Take a look at this supercomputer. A RAC of 382K (currently anyway) achieved on its opening day is not too shabby :). Also, 10K+ tasks in its tasks list - WOW!! :).
And it didn't happen once only!! Somebody stuffed up a few times by the look of things. The top of the list used to be #1 - Peanut and #2 - Akosf. Those two have (temporarily) been rather displaced :).
I wonder if anyone will "fess up" as to exactly what happened :).
Cheers,
Gary.
Hopefully, just an OOPS!! moment ....
)
Definitely some shenanigins going on... lol :0
...seems to be a Cluster with
)
...seems to be a Cluster with Linux 2.6.9-55.0.9.EL.cernsmp...
@peanut: Some shenanigans
)
@peanut:
Some shenanigans going on here too! :).
I had started to reply to your "dam burst" analogy only to find that the dam indeed has burst and your previous words have flowed away ... :).
I think that the problem has perhaps occurred because the combined results for a cluster of machines have inadvertently become "attached" to just a single (or perhaps a few) members of the cluster, somehow. I don't think it's the result of storing up results over time and releasing them in a single "dam burst".
Also, regarding your comments about the performance of the former #2 computer. I posted some details about the crunch time improvements here . I think it's pretty obvious that Akos is improving his optimisations (apparently fairly dramatically) and I look forward to the time that he is happy to unleash the improvements on the GUM (Great Unwashed Masses) that somebody else referred to :).
Cheers,
Gary.
@Gary, Sorry I changed my
)
@Gary,
Sorry I changed my post. I realized that I really had no idea of what is possible, so I removed my speculation. I guessed that no one saw it before I changed it. Not that it was bad or anything,
RE: I think it's pretty
)
The only way "to unleash the improvements" is in Bernd's hand.
RE: The only way "to
)
Yes, of course. I'm sorry for not wording my comment better. I certainly didn't mean to imply that you were keeping your improvements to yourself or that only you would make a unilateral decision about release. I should have said something like, "I look forward to the time that Akos' improvements can be incorporated into the official apps and then be released to the GUM". :). This is something that Bernd has indicated would happen after he is happy with the progress on bug squashing.
Cheers,
Gary.
RE: RE: I think it's
)
You could always release binaries yourself again. Or is that now frowned upon?
RE: You could always
)
It is. He was asked to stop early in S5 because the use of unoffocial apps could potentially taint the scientific validity of the results.
RE: @Gary, Sorry I changed
)
That's quite OK :). I was just having a little tease since I had been quick enough to have noticed your speculation and particularly the extra comments about RAC decaying and how fast Akos' results were being crunched.
As you surmised, the RACs of the 4 rogue machines have all started decaying and you have now overtaken the lowest ranked one. As you also correctly surmised, Akos' RAC is climbing rapidly and it looks like he will overtake you shortly. I just looked at his most recent 20 returned results which seem to be averaging not much over 7K seconds each. The last time I looked they were averaging around 9K seconds so I would guess that he has made even more improvement.
Quite amazing performance!!
Cheers,
Gary.
RE: Has anyone noticed the
)
I contacted the participant in question. It turned out that accidentally the hostid was shared on a cluster of ~50 machines.
BM
BM