Any news about optimized client?
6 Jul 2006 9:19:16 UTC
Topic 191520
(moderation:
Hi all
AFAIK last one is S5S0007.DAT but there are some suspicions it's not valid one.
Are any statements from project leaders/programmers about this version exist? (I havent seen but maybe I miss it in one of threads? )
Meantime if someone still uses this patch please post links on ur hosts here in this thread. It's interesting to observe validation rate against stock clients.
If they validated against "official" client - they are good ones by definition.
(BTW if they are not validated it _doesnt'_ mean they are "bad", official client may be wrong too ;) But anyway good statistic is needed! )
Any news about optimized client?
)
hm, at the moment we only know we shall not use it. I continued to use the S5T0712 some days to see if its really bad, but i stopped with that and now only use the official linux-client. Here is the PC:
http://einsteinathome.org/host/38017/tasks
The two "Client Error" Results i aborted manually, so that has nothing to do with the client.
RE: hm, at the moment we
)
Thanx a lot. If you claim these 2 errors should not to be counted it seems this version validates against 4.02 and 4.01 too ('einstein_S5R1_4.01_i686-pc-linux-gnu').
Lets see what situation is at another hosts.
Yea, I had read that thread completelly. Just as with small children: dont touch this cause it's bad to touch this :)
Think some explanation is needed (or at least "dont' touch it" from real project "parents" ;) )
If it will be considered just as any another beta version (cause now there too small chances to get 2 suspicious results in one WU for validation) it could be approved or rejected based just on validator statistics (as it should be) not on some "opinions".
We are still busy with
)
We are still busy with digging trough the DB (including the already purged results) for trouble that was caused by these unofficial optimized Apps. Presently this involves re-calculation and careful examination of potentially troublesome results. As long as we are doing this, it would be nice not to add such results to the DB.
BM
BM
RE: Hi all If they
)
No, not being invalid does not mean something is valid. Something is valid if it IS valid...
Somnio ergo sum
RE: RE: Hi all If they
)
Fully agreed with second part :)
But what state "not invalid but not valid" is mean? AFAIK there are 2 possibly final situations:
1) There is quorum.
2) There is no quorum.
If 1) and there is stock app is used in one of results then all validated are valid by definition (if official has bug in it - bad times :) )
If 1) and only optimised applications used in result computations then validity is questionable( but we should consider they are valid if we have no or very little failures of used app in comparision with official ones. (here statistic is needed)).
If 2) and different apps are used for results then debugging is needed to discover who is wrong and statistic again to rule out just too OCed or buggy hosts.
But "official" app has no additional "weight" besides all other apps should generate results that validate with "official" results.
If it's done - all ok. If it's done and more faster - it's wonderful :)
RE: We are still busy with
)
Thank you for answer!
It would be great if later, when you will have some free time, you will post some description of situation you fight with now.
Will await patiently :)
RE: We are still busy with
)
Bernd,
I'm running two servers at a remote location (approx. 60 km / 40 miles away).
On 24th of june I installed optimized app S5S0007 on these servers - few days later we were asked to no longer use optimized apps. As I'm normally only once per month at that location - and during summer vacation I will have no possibility to go to that location before september - can you release a new windows app (different version number but same code) which will be automatically fetched by the clients?
This will 'reset' a lot of other computers still using optimized apps.
Udo
RE: I'm running two servers
)
Don't you have GUI RPC connectivity to those two machines?
If yes, then just reset the project (after the queue drained of course) using BOINC Manager or something alike.
If not, then you did a mighty irresponsible thing installing unofficial beta application there.
[edit] spilling [/edit]
Metod ...
RE: RE: We are still busy
)
Good idea.
I wonder those who argued
)
I wonder those who argued that the optimized applications were still OK to run will change their mind now that Bernard has stated it is causing problems, and has more or less asked people to stop?